diff --git a/README.md b/README.md index 73fcbb39b..74ab05475 100644 --- a/README.md +++ b/README.md @@ -61,8 +61,8 @@ Results are framed by **astronomy instrument** (HST, Euclid, JWST, …) rather t | `imaging/delaunay` | hst | hpc_a100_fp64 | sparse (w-tilde) | 98.0 ms | v2026.7.6.649 | | `imaging/delaunay` | hst | hpc_a100_mp | dense (mapping) | 96.8 ms | v2026.7.6.649 | | `imaging/delaunay` | hst | hpc_a100_mp | sparse (w-tilde) | 95.5 ms | v2026.7.6.649 | -| `imaging/delaunay_numba` | euclid | local_cpu_fp64 | sparse (numba) | 4.84 s | v2026.8.17.1 | -| `imaging/delaunay_numba` | hst | local_cpu_fp64 | sparse (numba) | 4.48 s | v2026.8.17.1 | +| `imaging/delaunay_numba` | euclid | local_cpu_fp64 | sparse (numba) | 1.19 s | v2026.8.17.1 | +| `imaging/delaunay_numba` | hst | local_cpu_fp64 | sparse (numba) | 3.22 s | v2026.8.17.1 | | `imaging/mge` | hst | local_cpu_fp64 | dense (mapping) | 179.5 ms | v2026.7.6.649 | | `imaging/mge` | hst | hpc_a100_fp64 | dense (mapping) | 7.8 ms | v2026.7.6.649 | | `imaging/pixelization` | hst | local_cpu_fp64 | dense (mapping) | 8.65 s | v2026.7.6.649 | diff --git a/results/README.md b/results/README.md index 3f381316c..753d7b99b 100644 --- a/results/README.md +++ b/results/README.md @@ -16,7 +16,7 @@ packages; the dashboard tables in every README are rendered from this tree by | `quick_update/` | [`quick_update/`](../scripts/misc/quick_update/README.md) | Unversioned fast re-profiling snapshots (scratch tier) | | `delaunay_nn/` | [`delaunay_nn/`](../scripts/misc/delaunay_nn/README.md) | Versioned full-mapper cap and runtime benchmarks | | `hazards/` | [`hazards/`](../scripts/misc/hazards/README.md) | Semantic finding records, reproducer plots, generated seed summary, and consumer index | -| `notes/` | humans + agents | Narrative findings and design notes (e.g. [`design_lock_in.md`](./notes/design_lock_in.md)) | +| `notes/` | humans + agents | Narrative findings and design notes (e.g. [`design_lock_in.md`](./notes/design_lock_in.md), [`nnls_solver_ledger.md`](./notes/nnls_solver_ledger.md), [`nnls_warm_start_memo.md`](./notes/nnls_warm_start_memo.md), [`nnls_warm_start_memo_matrix.md`](./notes/nnls_warm_start_memo_matrix.md)) | | `baselines/` | campaign snapshots | Named, frozen baselines (e.g. `PreOptimizationTimes/`) — see below | ## Performance artifact shapes diff --git a/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.json index 1c60f3e48..622ff8487 100644 --- a/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.json +++ b/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.json @@ -5,7 +5,7 @@ "device": "cpu:0", "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", "omp_num_threads": "1", - "cpu_count": 4 + "cpu_count": 8 }, "instrument": "euclid", "configuration": { @@ -25,29 +25,29 @@ "omp_num_threads": "1" }, "steps": { - "FitImaging construct": 0.00013836129999731384, - "Blurred image (FFT convolve)": 0.01630020360000799, - "Profile subtracted image": 5.2616499993973774e-05, - "Inversion build (trace+Delaunay+mapper)": 0.021611292699992644, - "PSF-weighted data [numba]": 0.0034058456999900956, - "MGE operated mapping matrix (60 funcs)": 0.560158383999999, - "Mapper sparse triplets": 0.014768384199993535, - "Data vector D [numba]": 0.0015318015000104878, - "Curvature matrix F [numba sparse-op]": 0.29001415160000193, - "Regularization matrix H (ConstantSplit)": 0.09172119199999998, - "F + H": 0.0031295566000096643, - "Reconstruction solve (BLAS)": 3.6148581532000037, - "Mapped reconstruction [numba+FFT]": 0.00954050639999764, - "Model data + chi^2": 0.009341346800005113, - "log det (F+H) [Cholesky]": 0.05571201360000373, - "log det H [Cholesky]": 0.045085756200006696, - "Regularization term s'Hs": 0.001553125199995975, - "Log evidence (figure of merit)": 0.10032829029999561 + "FitImaging construct": 8.208000008380623e-05, + "Blurred image (FFT convolve)": 0.008638289999908011, + "Profile subtracted image": 4.433999993125326e-05, + "Inversion build (trace+Delaunay+mapper)": 0.022416690000000017, + "PSF-weighted data [numba]": 0.003058889999965686, + "MGE operated mapping matrix (60 funcs)": 0.14983798000002935, + "Mapper sparse triplets": 0.017052480000074866, + "Data vector D [numba]": 0.0014287400000284833, + "Curvature matrix F [numba sparse-op]": 0.2916402800000014, + "Regularization matrix H (ConstantSplit)": 0.09061346000003141, + "F + H": 0.002813830000013695, + "Reconstruction solve (BLAS)": 0.47000207000000954, + "Mapped reconstruction [numba+FFT]": 0.0017286499999954684, + "Model data + chi^2": 0.0018000400000346417, + "log det (F+H) [Cholesky]": 0.03419463000004726, + "log det H [Cholesky]": 0.031532190000007176, + "Regularization term s'Hs": 0.000812050000013187, + "Log evidence (figure of merit)": 0.061450309999963794 }, - "total_step_by_step": 4.839250981400005, - "direct_log_likelihood_function_per_call": 4.994904638500003, - "warmup_incl_numba_compile_s": 7.218259176000004, - "log_likelihood": 7215.3687893658935, - "pinned_expected": null, + "total_step_by_step": 1.1891470000001392, + "direct_log_likelihood_function_per_call": 1.2646348100000069, + "warmup_incl_numba_compile_s": 3.253364899999724, + "log_likelihood": 7215.369093236955, + "pinned_expected": 7215.3687893658935, "pinned_drift": [] } \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.png index c95f2fb3a..367699f6f 100644 Binary files a/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.png and b/results/breakdown/imaging/delaunay_numba_breakdown_euclid_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.json index 7c040e342..1e6674d50 100644 --- a/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.json +++ b/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.json @@ -5,7 +5,7 @@ "device": "cpu:0", "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", "omp_num_threads": "1", - "cpu_count": 4 + "cpu_count": 8 }, "instrument": "hst", "configuration": { @@ -25,29 +25,29 @@ "omp_num_threads": "1" }, "steps": { - "FitImaging construct": 0.00015594309999187317, - "Blurred image (FFT convolve)": 0.030774156100028448, - "Profile subtracted image": 0.00011517150001054688, - "Inversion build (trace+Delaunay+mapper)": 0.03938092059999008, - "PSF-weighted data [numba]": 0.011929086600002848, - "MGE operated mapping matrix (60 funcs)": 1.1724747234000006, - "Mapper sparse triplets": 0.026465198999994756, - "Data vector D [numba]": 0.008720781399983935, - "Curvature matrix F [numba sparse-op]": 1.4052928210000004, - "Regularization matrix H (ConstantSplit)": 0.10681075329999885, - "F + H": 0.004605925099986053, - "Reconstruction solve (BLAS)": 1.4235228239000093, - "Mapped reconstruction [numba+FFT]": 0.01963783639999974, - "Model data + chi^2": 0.019632536799986156, - "log det (F+H) [Cholesky]": 0.05760797249998859, - "log det H [Cholesky]": 0.04742895750000571, - "Regularization term s'Hs": 0.0014776729999880444, - "Log evidence (figure of merit)": 0.102019350400019 + "FitImaging construct": 0.00010670000006030023, + "Blurred image (FFT convolve)": 0.019015520000039032, + "Profile subtracted image": 0.00010823000002346816, + "Inversion build (trace+Delaunay+mapper)": 0.05277517000004082, + "PSF-weighted data [numba]": 0.01558175999998639, + "MGE operated mapping matrix (60 funcs)": 0.4349745199998779, + "Mapper sparse triplets": 0.030901940000103423, + "Data vector D [numba]": 0.008124649999899702, + "Curvature matrix F [numba sparse-op]": 1.7725606600000448, + "Regularization matrix H (ConstantSplit)": 0.13839846999994734, + "F + H": 0.003469200000108685, + "Reconstruction solve (BLAS)": 0.5417054600000938, + "Mapped reconstruction [numba+FFT]": 0.006093289999989792, + "Model data + chi^2": 0.006790640000008352, + "log det (F+H) [Cholesky]": 0.050793790000079754, + "log det H [Cholesky]": 0.04692912999998953, + "Regularization term s'Hs": 0.0011736499999642548, + "Log evidence (figure of merit)": 0.0934743499999513 }, - "total_step_by_step": 4.478052631599984, - "direct_log_likelihood_function_per_call": 4.05626313570001, - "warmup_incl_numba_compile_s": 8.269797903999915, - "log_likelihood": 29090.527192303925, - "pinned_expected": null, + "total_step_by_step": 3.222977130000208, + "direct_log_likelihood_function_per_call": 3.664489109999977, + "warmup_incl_numba_compile_s": 5.48075499999959, + "log_likelihood": 29090.52721044813, + "pinned_expected": 29090.527192092646, "pinned_drift": [] } \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.png index c9392cd37..78f046425 100644 Binary files a/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.png and b/results/breakdown/imaging/delaunay_numba_breakdown_hst_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_adapt_reg_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_adapt_reg_v2026.8.17.1.json new file mode 100644 index 000000000..0b7a48b17 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_adapt_reg_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "adapt_reg", + "configuration": { + "model": "adapt_reg", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "adapt_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 3.4651003999997556, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.05419847328244275, + "mean_warm_start_error_fraction": 0.05328244274809161, + "p90_warm_start_error_fraction": 0.05656488549618321, + "max_warm_start_error_fraction": 0.05877862595419847, + "median_outer_iterations": 57.5, + "mean_outer_iterations": 57.2, + "p90_outer_iterations": 65.4, + "max_outer_iterations": 71.0, + "median_inner_iterations": 8.0, + "mean_inner_iterations": 8.8, + "p90_inner_iterations": 12.300000000000004, + "max_inner_iterations": 16.0, + "median_total_iterations": 66.0, + "mean_total_iterations": 66.0, + "p90_total_iterations": 73.30000000000001, + "max_total_iterations": 86.0, + "median_n_passive": 1227.0, + "mean_n_passive": 1226.25, + "p90_n_passive": 1230.1, + "max_n_passive": 1231.0, + "median_warm_start_errors": 71.0, + "mean_warm_start_errors": 69.8, + "p90_warm_start_errors": 74.1, + "max_warm_start_errors": 77.0, + "median_solve_s": 0.5117872499999976, + "mean_solve_s": 0.5050070050000158, + "p90_solve_s": 0.6649182499996641, + "max_solve_s": 0.8647769999997763, + "median_reconstruction_s": 1.2260003500000494, + "mean_reconstruction_s": 1.2319590699999936, + "p90_reconstruction_s": 1.6273678800003837, + "max_reconstruction_s": 1.7316133000003902, + "median_eval_s": 1.315945300000294, + "mean_eval_s": 1.3240341650000573, + "p90_eval_s": 1.75753797999987, + "max_eval_s": 1.838071299999683 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.013358778625954198, + "mean_warm_start_error_fraction": 0.015152671755725192, + "p90_warm_start_error_fraction": 0.022213740458015267, + "max_warm_start_error_fraction": 0.050381679389312976, + "median_outer_iterations": 13.5, + "mean_outer_iterations": 15.45, + "p90_outer_iterations": 24.50000000000002, + "max_outer_iterations": 65.0, + "median_inner_iterations": 2.0, + "mean_inner_iterations": 2.1, + "p90_inner_iterations": 4.100000000000001, + "max_inner_iterations": 7.0, + "median_total_iterations": 15.5, + "mean_total_iterations": 17.55, + "p90_total_iterations": 28.30000000000002, + "max_total_iterations": 72.0, + "median_n_passive": 1227.0, + "mean_n_passive": 1226.05, + "p90_n_passive": 1230.1, + "max_n_passive": 1231.0, + "median_warm_start_errors": 17.5, + "mean_warm_start_errors": 19.85, + "p90_warm_start_errors": 29.1, + "max_warm_start_errors": 66.0, + "median_solve_s": 0.11059625000007145, + "mean_solve_s": 0.14198677999997927, + "p90_solve_s": 0.17166818000005152, + "max_solve_s": 0.8357251999996151, + "median_reconstruction_s": 0.6371365499999229, + "mean_reconstruction_s": 0.7022406950000913, + "p90_reconstruction_s": 0.7975415200005046, + "max_reconstruction_s": 1.6723685999995723, + "median_eval_s": 0.7030723500001841, + "mean_eval_s": 0.784405095000011, + "p90_eval_s": 0.8793670199999725, + "max_eval_s": 1.7684423999999126 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 2.3984670360139033e-14, + "max_abs_reconstruction_deviation": 9.793721389428356e-10 + }, + "iteration_reduction_ratio": 4.258064516129032, + "solve_speedup_ratio": 4.6275280581363925, + "eval_speedup_ratio": 1.871706802294173, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 1.1219013999998424, + "eval_s": 1.2009040000002642, + "log_likelihood": 5587.183365707368, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6231876000001648, + "n": 1310, + "solve_s": 0.6231876000001648, + "outer_iterations": 65, + "inner_iterations": 7, + "total_iterations": 72, + "n_passive": 1227, + "warm_start_errors": 66, + "index": 0 + }, + { + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 1.2419534000000567, + "eval_s": 1.3385096000001795, + "log_likelihood": 5542.961266279436, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5548245999998471, + "n": 1310, + "solve_s": 0.5548245999998471, + "outer_iterations": 69, + "inner_iterations": 16, + "total_iterations": 85, + "n_passive": 1222, + "warm_start_errors": 71, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 1.3123435999996218, + "eval_s": 1.3861521999997422, + "log_likelihood": 5587.098539440693, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5756879000000481, + "n": 1310, + "solve_s": 0.5756879000000481, + "outer_iterations": 58, + "inner_iterations": 11, + "total_iterations": 69, + "n_passive": 1228, + "warm_start_errors": 72, + "index": 2 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 0.8738667000006899, + "eval_s": 0.9410330000000613, + "log_likelihood": 5621.406943488521, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28424769999946875, + "n": 1310, + "solve_s": 0.28424769999946875, + "outer_iterations": 50, + "inner_iterations": 6, + "total_iterations": 56, + "n_passive": 1227, + "warm_start_errors": 65, + "index": 3 + }, + { + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 0.9201155000000654, + "eval_s": 0.9790283000002091, + "log_likelihood": 5520.017929533509, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38139049999972485, + "n": 1310, + "solve_s": 0.38139049999972485, + "outer_iterations": 71, + "inner_iterations": 15, + "total_iterations": 86, + "n_passive": 1231, + "warm_start_errors": 71, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.7415062000000034, + "eval_s": 0.8063357000000906, + "log_likelihood": 5588.907960500332, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.267999100000452, + "n": 1310, + "solve_s": 0.267999100000452, + "outer_iterations": 57, + "inner_iterations": 9, + "total_iterations": 66, + "n_passive": 1223, + "warm_start_errors": 72, + "index": 5 + }, + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 0.902490200000102, + "eval_s": 0.9836282999995092, + "log_likelihood": 5643.356041965719, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3206606000003376, + "n": 1310, + "solve_s": 0.3206606000003376, + "outer_iterations": 52, + "inner_iterations": 7, + "total_iterations": 59, + "n_passive": 1227, + "warm_start_errors": 67, + "index": 6 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 1.6197600000004968, + "eval_s": 1.7510794999998325, + "log_likelihood": 5718.630647922662, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8647769999997763, + "n": 1310, + "solve_s": 0.8647769999997763, + "outer_iterations": 50, + "inner_iterations": 7, + "total_iterations": 57, + "n_passive": 1227, + "warm_start_errors": 65, + "index": 7 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 1.6958387999993647, + "eval_s": 1.838071299999683, + "log_likelihood": 5795.497729932087, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.664782799999557, + "n": 1310, + "solve_s": 0.664782799999557, + "outer_iterations": 59, + "inner_iterations": 7, + "total_iterations": 66, + "n_passive": 1230, + "warm_start_errors": 73, + "index": 8 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 1.3988140999999814, + "eval_s": 1.5002276000004713, + "log_likelihood": 5742.739745218305, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5353778999997303, + "n": 1310, + "solve_s": 0.5353778999997303, + "outer_iterations": 59, + "inner_iterations": 8, + "total_iterations": 67, + "n_passive": 1227, + "warm_start_errors": 74, + "index": 9 + }, + { + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 1.012268600000425, + "eval_s": 1.0926731999998083, + "log_likelihood": 5654.531999161209, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3524778000000879, + "n": 1310, + "solve_s": 0.3524778000000879, + "outer_iterations": 56, + "inner_iterations": 8, + "total_iterations": 64, + "n_passive": 1219, + "warm_start_errors": 71, + "index": 10 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 1.0886713999998392, + "eval_s": 1.1580653999999413, + "log_likelihood": 5612.135328159378, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43407200000001467, + "n": 1310, + "solve_s": 0.43407200000001467, + "outer_iterations": 61, + "inner_iterations": 10, + "total_iterations": 71, + "n_passive": 1225, + "warm_start_errors": 75, + "index": 11 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.9557448999994449, + "eval_s": 1.0490214999999807, + "log_likelihood": 5575.912296727059, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4314340999999331, + "n": 1310, + "solve_s": 0.4314340999999331, + "outer_iterations": 59, + "inner_iterations": 10, + "total_iterations": 69, + "n_passive": 1224, + "warm_start_errors": 73, + "index": 12 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 1.5849797000000763, + "eval_s": 1.69729180000013, + "log_likelihood": 5576.220072581891, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6549912000000404, + "n": 1310, + "solve_s": 0.6549912000000404, + "outer_iterations": 59, + "inner_iterations": 10, + "total_iterations": 69, + "n_passive": 1226, + "warm_start_errors": 73, + "index": 13 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 1.5019517999999152, + "eval_s": 1.6126159000004918, + "log_likelihood": 5529.0131634745785, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6661373000006279, + "n": 1310, + "solve_s": 0.6661373000006279, + "outer_iterations": 53, + "inner_iterations": 6, + "total_iterations": 59, + "n_passive": 1222, + "warm_start_errors": 69, + "index": 14 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 1.7316133000003902, + "eval_s": 1.8156643000002077, + "log_likelihood": 5522.487287365645, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5242917000005036, + "n": 1310, + "solve_s": 0.5242917000005036, + "outer_iterations": 57, + "inner_iterations": 8, + "total_iterations": 65, + "n_passive": 1226, + "warm_start_errors": 69, + "index": 15 + }, + { + "warm_start_error_fraction": 0.04732824427480916, + "reconstruction_s": 1.210047300000042, + "eval_s": 1.3014857000007396, + "log_likelihood": 5679.081804815349, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.49928279999949154, + "n": 1310, + "solve_s": 0.49928279999949154, + "outer_iterations": 47, + "inner_iterations": 5, + "total_iterations": 52, + "n_passive": 1229, + "warm_start_errors": 62, + "index": 16 + }, + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 1.4835589999993317, + "eval_s": 1.5758598000002166, + "log_likelihood": 5648.088847848746, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.643641600000592, + "n": 1310, + "solve_s": 0.643641600000592, + "outer_iterations": 61, + "inner_iterations": 8, + "total_iterations": 69, + "n_passive": 1231, + "warm_start_errors": 77, + "index": 17 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 1.2492818000000625, + "eval_s": 1.3304048999998486, + "log_likelihood": 5676.957427425598, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4223156999996718, + "n": 1310, + "solve_s": 0.4223156999996718, + "outer_iterations": 55, + "inner_iterations": 6, + "total_iterations": 61, + "n_passive": 1228, + "warm_start_errors": 73, + "index": 18 + }, + { + "warm_start_error_fraction": 0.044274809160305344, + "reconstruction_s": 0.9924737000001187, + "eval_s": 1.122631299999739, + "log_likelihood": 5680.647842354635, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3985602000002473, + "n": 1310, + "solve_s": 0.3985602000002473, + "outer_iterations": 46, + "inner_iterations": 12, + "total_iterations": 58, + "n_passive": 1226, + "warm_start_errors": 58, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 1.6723685999995723, + "eval_s": 1.7684423999999126, + "log_likelihood": 5587.183365707368, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8357251999996151, + "n": 1310, + "solve_s": 0.8357251999996151, + "outer_iterations": 65, + "inner_iterations": 7, + "total_iterations": 72, + "n_passive": 1227, + "warm_start_errors": 66, + "index": 0 + }, + { + "warm_start_error_fraction": 0.022900763358778626, + "reconstruction_s": 0.757863400000133, + "eval_s": 0.854573600000549, + "log_likelihood": 5542.961266279558, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.15980920000038168, + "n": 1310, + "solve_s": 0.15980920000038168, + "outer_iterations": 18, + "inner_iterations": 2, + "total_iterations": 20, + "n_passive": 1223, + "warm_start_errors": 30, + "index": 1 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.8107312000001912, + "eval_s": 0.8778179999999338, + "log_likelihood": 5587.098539440702, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06997449999926175, + "n": 1310, + "solve_s": 0.06997449999926175, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 1227, + "warm_start_errors": 10, + "index": 2 + }, + { + "warm_start_error_fraction": 0.01984732824427481, + "reconstruction_s": 0.7020864000005531, + "eval_s": 0.7864964000000327, + "log_likelihood": 5621.406943488643, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.15609680000034132, + "n": 1310, + "solve_s": 0.15609680000034132, + "outer_iterations": 17, + "inner_iterations": 3, + "total_iterations": 20, + "n_passive": 1227, + "warm_start_errors": 26, + "index": 3 + }, + { + "warm_start_error_fraction": 0.012213740458015267, + "reconstruction_s": 0.729008000000249, + "eval_s": 0.8301084999993691, + "log_likelihood": 5520.0179295334765, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.17239430000063294, + "n": 1310, + "solve_s": 0.17239430000063294, + "outer_iterations": 13, + "inner_iterations": 2, + "total_iterations": 15, + "n_passive": 1231, + "warm_start_errors": 16, + "index": 4 + }, + { + "warm_start_error_fraction": 0.012977099236641221, + "reconstruction_s": 0.7131031999997504, + "eval_s": 0.8371613000008438, + "log_likelihood": 5588.907960500321, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.13993639999989682, + "n": 1310, + "solve_s": 0.13993639999989682, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 1222, + "warm_start_errors": 17, + "index": 5 + }, + { + "warm_start_error_fraction": 0.011450381679389313, + "reconstruction_s": 0.7241014999999607, + "eval_s": 0.7976121000001513, + "log_likelihood": 5643.35604196573, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11349780000000464, + "n": 1310, + "solve_s": 0.11349780000000464, + "outer_iterations": 17, + "inner_iterations": 5, + "total_iterations": 22, + "n_passive": 1227, + "warm_start_errors": 15, + "index": 6 + }, + { + "warm_start_error_fraction": 0.016793893129770993, + "reconstruction_s": 0.7960760000005394, + "eval_s": 0.8933082000003196, + "log_likelihood": 5718.630647922573, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1715874999999869, + "n": 1310, + "solve_s": 0.1715874999999869, + "outer_iterations": 14, + "inner_iterations": 2, + "total_iterations": 16, + "n_passive": 1227, + "warm_start_errors": 22, + "index": 7 + }, + { + "warm_start_error_fraction": 0.017557251908396947, + "reconstruction_s": 0.7854683000005025, + "eval_s": 0.8695298000002367, + "log_likelihood": 5795.497729932107, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.13222810000024765, + "n": 1310, + "solve_s": 0.13222810000024765, + "outer_iterations": 20, + "inner_iterations": 2, + "total_iterations": 22, + "n_passive": 1230, + "warm_start_errors": 23, + "index": 8 + }, + { + "warm_start_error_fraction": 0.003816793893129771, + "reconstruction_s": 0.6124815000002855, + "eval_s": 0.67757519999941, + "log_likelihood": 5742.739745218181, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04220359999999346, + "n": 1310, + "solve_s": 0.04220359999999346, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1227, + "warm_start_errors": 5, + "index": 9 + }, + { + "warm_start_error_fraction": 0.013740458015267175, + "reconstruction_s": 0.5137052999998559, + "eval_s": 0.5939883999999438, + "log_likelihood": 5654.531999161243, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.061889000000519445, + "n": 1310, + "solve_s": 0.061889000000519445, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1219, + "warm_start_errors": 18, + "index": 10 + }, + { + "warm_start_error_fraction": 0.01984732824427481, + "reconstruction_s": 0.613814999999704, + "eval_s": 0.6772126999994725, + "log_likelihood": 5612.135328159244, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10769470000013825, + "n": 1310, + "solve_s": 0.10769470000013825, + "outer_iterations": 23, + "inner_iterations": 4, + "total_iterations": 27, + "n_passive": 1225, + "warm_start_errors": 26, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0053435114503816794, + "reconstruction_s": 0.5134821000001466, + "eval_s": 0.5813686000001326, + "log_likelihood": 5575.912296726991, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.056990899999618705, + "n": 1310, + "solve_s": 0.056990899999618705, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 1224, + "warm_start_errors": 7, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 0.5792662999992899, + "eval_s": 0.6437759999998889, + "log_likelihood": 5576.220072581915, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04775520000021061, + "n": 1310, + "solve_s": 0.04775520000021061, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1225, + "warm_start_errors": 3, + "index": 13 + }, + { + "warm_start_error_fraction": 0.008396946564885497, + "reconstruction_s": 0.5059461000000738, + "eval_s": 0.5688076999995246, + "log_likelihood": 5529.013163474511, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05356559999927413, + "n": 1310, + "solve_s": 0.05356559999927413, + "outer_iterations": 6, + "inner_iterations": 2, + "total_iterations": 8, + "n_passive": 1222, + "warm_start_errors": 11, + "index": 14 + }, + { + "warm_start_error_fraction": 0.01984732824427481, + "reconstruction_s": 0.5718130000004749, + "eval_s": 0.6857003000004624, + "log_likelihood": 5522.487287365569, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1323471999994581, + "n": 1310, + "solve_s": 0.1323471999994581, + "outer_iterations": 20, + "inner_iterations": 2, + "total_iterations": 22, + "n_passive": 1226, + "warm_start_errors": 26, + "index": 15 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.5874779000005219, + "eval_s": 0.6801481000002241, + "log_likelihood": 5679.081804815329, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08692100000007486, + "n": 1310, + "solve_s": 0.08692100000007486, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1228, + "warm_start_errors": 10, + "index": 16 + }, + { + "warm_start_error_fraction": 0.022137404580152672, + "reconstruction_s": 0.6360849000002418, + "eval_s": 0.6986697000002096, + "log_likelihood": 5648.088847848659, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1489521000003151, + "n": 1310, + "solve_s": 0.1489521000003151, + "outer_iterations": 38, + "inner_iterations": 2, + "total_iterations": 40, + "n_passive": 1231, + "warm_start_errors": 29, + "index": 17 + }, + { + "warm_start_error_fraction": 0.00916030534351145, + "reconstruction_s": 0.5817470000001776, + "eval_s": 0.658329899999444, + "log_likelihood": 5676.957427425653, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05036569999992935, + "n": 1310, + "solve_s": 0.05036569999992935, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1227, + "warm_start_errors": 12, + "index": 18 + }, + { + "warm_start_error_fraction": 0.019083969465648856, + "reconstruction_s": 0.638188199999604, + "eval_s": 0.7074750000001586, + "log_likelihood": 5680.647842354679, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09980079999968439, + "n": 1310, + "solve_s": 0.09980079999968439, + "outer_iterations": 16, + "inner_iterations": 1, + "total_iterations": 17, + "n_passive": 1226, + "warm_start_errors": 25, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.05725190839694656, + "mean_warm_start_error_fraction": 0.05958015267175574, + "p90_warm_start_error_fraction": 0.07595419847328244, + "max_warm_start_error_fraction": 0.08244274809160305, + "median_outer_iterations": 59.5, + "mean_outer_iterations": 62.9, + "p90_outer_iterations": 82.4, + "max_outer_iterations": 93.0, + "median_inner_iterations": 9.0, + "mean_inner_iterations": 8.9, + "p90_inner_iterations": 12.100000000000001, + "max_inner_iterations": 13.0, + "median_total_iterations": 70.0, + "mean_total_iterations": 71.8, + "p90_total_iterations": 92.2, + "max_total_iterations": 99.0, + "median_n_passive": 1217.5, + "mean_n_passive": 1222.0, + "p90_n_passive": 1239.6, + "max_n_passive": 1248.0, + "median_warm_start_errors": 75.0, + "mean_warm_start_errors": 78.05, + "p90_warm_start_errors": 99.5, + "max_warm_start_errors": 108.0, + "median_solve_s": 0.3524464500001159, + "mean_solve_s": 0.3542164750000666, + "p90_solve_s": 0.432117670000207, + "max_solve_s": 0.44878740000058315, + "median_reconstruction_s": 0.9252180999997108, + "mean_reconstruction_s": 0.9360591050000494, + "p90_reconstruction_s": 1.0753877899994222, + "max_reconstruction_s": 1.1118263999996998, + "median_eval_s": 0.9971095499995499, + "mean_eval_s": 1.0077609949998076, + "p90_eval_s": 1.1573722299998737, + "max_eval_s": 1.1854899999998452 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.04961832061068702, + "mean_warm_start_error_fraction": 0.05061068702290076, + "p90_warm_start_error_fraction": 0.06740458015267176, + "max_warm_start_error_fraction": 0.08015267175572519, + "median_outer_iterations": 60.0, + "mean_outer_iterations": 62.95, + "p90_outer_iterations": 88.10000000000002, + "max_outer_iterations": 151.0, + "median_inner_iterations": 7.5, + "mean_inner_iterations": 8.85, + "p90_inner_iterations": 15.300000000000004, + "max_inner_iterations": 20.0, + "median_total_iterations": 70.0, + "mean_total_iterations": 71.8, + "p90_total_iterations": 101.60000000000001, + "max_total_iterations": 165.0, + "median_n_passive": 1217.0, + "mean_n_passive": 1221.7, + "p90_n_passive": 1238.7, + "max_n_passive": 1248.0, + "median_warm_start_errors": 65.0, + "mean_warm_start_errors": 66.3, + "p90_warm_start_errors": 88.30000000000001, + "max_warm_start_errors": 105.0, + "median_solve_s": 0.2698056999997789, + "mean_solve_s": 0.2835179849999804, + "p90_solve_s": 0.3819461900000533, + "max_solve_s": 0.6365933999995832, + "median_reconstruction_s": 0.7804079000002275, + "mean_reconstruction_s": 0.8100099400001, + "p90_reconstruction_s": 0.9287371099994745, + "max_reconstruction_s": 1.1480811000001268, + "median_eval_s": 0.846219949999977, + "mean_eval_s": 0.8807003500000974, + "p90_eval_s": 0.9955983600005313, + "max_eval_s": 1.2169707000002745 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 4.308579341131159e-14, + "max_abs_reconstruction_deviation": 8.132250428616317e-10 + }, + "iteration_reduction_ratio": 1.0, + "solve_speedup_ratio": 1.3062972724460777, + "eval_speedup_ratio": 1.1783101426521285, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 1.1118263999996998, + "eval_s": 1.1850592000000688, + "log_likelihood": 3835.995333232333, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4257407000004605, + "n": 1310, + "solve_s": 0.4257407000004605, + "outer_iterations": 60, + "inner_iterations": 13, + "total_iterations": 73, + "n_passive": 1207, + "warm_start_errors": 77, + "index": 0 + }, + { + "warm_start_error_fraction": 0.06183206106870229, + "reconstruction_s": 0.9051272000006065, + "eval_s": 0.9696971999992456, + "log_likelihood": 4776.955011640415, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34407610000016575, + "n": 1310, + "solve_s": 0.34407610000016575, + "outer_iterations": 63, + "inner_iterations": 8, + "total_iterations": 71, + "n_passive": 1211, + "warm_start_errors": 81, + "index": 1 + }, + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 0.7843029000005117, + "eval_s": 0.8469677000002775, + "log_likelihood": 5852.840950742196, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26591430000007676, + "n": 1310, + "solve_s": 0.26591430000007676, + "outer_iterations": 51, + "inner_iterations": 5, + "total_iterations": 56, + "n_passive": 1245, + "warm_start_errors": 67, + "index": 2 + }, + { + "warm_start_error_fraction": 0.08244274809160305, + "reconstruction_s": 0.9135165000006964, + "eval_s": 0.9716210999995383, + "log_likelihood": 4796.742679497043, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37904340000022785, + "n": 1310, + "solve_s": 0.37904340000022785, + "outer_iterations": 93, + "inner_iterations": 6, + "total_iterations": 99, + "n_passive": 1210, + "warm_start_errors": 108, + "index": 3 + }, + { + "warm_start_error_fraction": 0.07557251908396946, + "reconstruction_s": 0.8443840999998429, + "eval_s": 0.9305352999999741, + "log_likelihood": 5609.259956570299, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.360816800000066, + "n": 1310, + "solve_s": 0.360816800000066, + "outer_iterations": 82, + "inner_iterations": 5, + "total_iterations": 87, + "n_passive": 1237, + "warm_start_errors": 99, + "index": 4 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.9557077000008576, + "eval_s": 1.0155466999995042, + "log_likelihood": 5670.369454845861, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43807809999998426, + "n": 1310, + "solve_s": 0.43807809999998426, + "outer_iterations": 56, + "inner_iterations": 9, + "total_iterations": 65, + "n_passive": 1227, + "warm_start_errors": 75, + "index": 5 + }, + { + "warm_start_error_fraction": 0.051908396946564885, + "reconstruction_s": 0.8280440000007729, + "eval_s": 0.8908833999994386, + "log_likelihood": 5308.701747308844, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2950905999996394, + "n": 1310, + "solve_s": 0.2950905999996394, + "outer_iterations": 55, + "inner_iterations": 12, + "total_iterations": 67, + "n_passive": 1197, + "warm_start_errors": 68, + "index": 6 + }, + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 0.740418000000318, + "eval_s": 0.8046570999995311, + "log_likelihood": 5005.22051955335, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25973179999982676, + "n": 1310, + "solve_s": 0.25973179999982676, + "outer_iterations": 47, + "inner_iterations": 8, + "total_iterations": 55, + "n_passive": 1239, + "warm_start_errors": 59, + "index": 7 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.8402067999995779, + "eval_s": 0.9079271000000517, + "log_likelihood": 5349.678071715674, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.306803100000252, + "n": 1310, + "solve_s": 0.306803100000252, + "outer_iterations": 55, + "inner_iterations": 9, + "total_iterations": 64, + "n_passive": 1214, + "warm_start_errors": 74, + "index": 8 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.9136042000000089, + "eval_s": 0.9752346999994188, + "log_likelihood": 3426.9188812725806, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2856705000003785, + "n": 1310, + "solve_s": 0.2856705000003785, + "outer_iterations": 59, + "inner_iterations": 10, + "total_iterations": 69, + "n_passive": 1210, + "warm_start_errors": 74, + "index": 9 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.876371399999698, + "eval_s": 0.9370176000002175, + "log_likelihood": 5461.423080103111, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31006930000057764, + "n": 1310, + "solve_s": 0.31006930000057764, + "outer_iterations": 63, + "inner_iterations": 12, + "total_iterations": 75, + "n_passive": 1230, + "warm_start_errors": 75, + "index": 10 + }, + { + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 0.959191799999644, + "eval_s": 1.0411718999994264, + "log_likelihood": 4099.747849461966, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4159924999994473, + "n": 1310, + "solve_s": 0.4159924999994473, + "outer_iterations": 79, + "inner_iterations": 13, + "total_iterations": 92, + "n_passive": 1207, + "warm_start_errors": 87, + "index": 11 + }, + { + "warm_start_error_fraction": 0.07938931297709924, + "reconstruction_s": 1.108798399999614, + "eval_s": 1.1854899999998452, + "log_likelihood": 4875.353843213199, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.44878740000058315, + "n": 1310, + "solve_s": 0.44878740000058315, + "outer_iterations": 86, + "inner_iterations": 8, + "total_iterations": 94, + "n_passive": 1213, + "warm_start_errors": 104, + "index": 12 + }, + { + "warm_start_error_fraction": 0.07022900763358779, + "reconstruction_s": 0.9368319999994128, + "eval_s": 1.0046991999997772, + "log_likelihood": 5515.656174630241, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3670260999997481, + "n": 1310, + "solve_s": 0.3670260999997481, + "outer_iterations": 76, + "inner_iterations": 9, + "total_iterations": 85, + "n_passive": 1211, + "warm_start_errors": 92, + "index": 13 + }, + { + "warm_start_error_fraction": 0.04122137404580153, + "reconstruction_s": 0.9121674999996685, + "eval_s": 0.9895198999993227, + "log_likelihood": 5900.774014331114, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2885548000003837, + "n": 1310, + "solve_s": 0.2885548000003837, + "outer_iterations": 43, + "inner_iterations": 8, + "total_iterations": 51, + "n_passive": 1248, + "warm_start_errors": 54, + "index": 14 + }, + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 1.071675499999401, + "eval_s": 1.1542958999998518, + "log_likelihood": 5631.55587735493, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42829299999993964, + "n": 1310, + "solve_s": 0.42829299999993964, + "outer_iterations": 50, + "inner_iterations": 10, + "total_iterations": 60, + "n_passive": 1233, + "warm_start_errors": 67, + "index": 15 + }, + { + "warm_start_error_fraction": 0.04580152671755725, + "reconstruction_s": 1.012732600000163, + "eval_s": 1.0938938999997845, + "log_likelihood": 5622.214876458491, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3056157999999414, + "n": 1310, + "solve_s": 0.3056157999999414, + "outer_iterations": 43, + "inner_iterations": 7, + "total_iterations": 50, + "n_passive": 1217, + "warm_start_errors": 60, + "index": 16 + }, + { + "warm_start_error_fraction": 0.06412213740458016, + "reconstruction_s": 1.0625216999997065, + "eval_s": 1.1510109000000739, + "log_likelihood": 5901.336890546579, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41443619999972725, + "n": 1310, + "solve_s": 0.41443619999972725, + "outer_iterations": 69, + "inner_iterations": 6, + "total_iterations": 75, + "n_passive": 1236, + "warm_start_errors": 84, + "index": 17 + }, + { + "warm_start_error_fraction": 0.07022900763358779, + "reconstruction_s": 1.002659499999936, + "eval_s": 1.0849753000002238, + "log_likelihood": 4990.436852851217, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43145540000023175, + "n": 1310, + "solve_s": 0.43145540000023175, + "outer_iterations": 82, + "inner_iterations": 9, + "total_iterations": 91, + "n_passive": 1230, + "warm_start_errors": 92, + "index": 18 + }, + { + "warm_start_error_fraction": 0.04885496183206107, + "reconstruction_s": 0.9410939000008511, + "eval_s": 1.0150158000005831, + "log_likelihood": 4348.4381679283515, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3131335999996736, + "n": 1310, + "solve_s": 0.3131335999996736, + "outer_iterations": 46, + "inner_iterations": 11, + "total_iterations": 57, + "n_passive": 1218, + "warm_start_errors": 64, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 1.0276201999995465, + "eval_s": 1.0965347999999722, + "log_likelihood": 3835.995333232333, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3838531999999759, + "n": 1310, + "solve_s": 0.3838531999999759, + "outer_iterations": 60, + "inner_iterations": 13, + "total_iterations": 73, + "n_passive": 1207, + "warm_start_errors": 77, + "index": 0 + }, + { + "warm_start_error_fraction": 0.03206106870229008, + "reconstruction_s": 0.7772984000002907, + "eval_s": 0.8409970000002431, + "log_likelihood": 4776.955011640361, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20241059999989375, + "n": 1310, + "solve_s": 0.20241059999989375, + "outer_iterations": 47, + "inner_iterations": 4, + "total_iterations": 51, + "n_passive": 1211, + "warm_start_errors": 42, + "index": 1 + }, + { + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 0.9177500999994663, + "eval_s": 0.9843832000005932, + "log_likelihood": 5852.840950742232, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33415289999993547, + "n": 1310, + "solve_s": 0.33415289999993547, + "outer_iterations": 87, + "inner_iterations": 12, + "total_iterations": 99, + "n_passive": 1245, + "warm_start_errors": 88, + "index": 2 + }, + { + "warm_start_error_fraction": 0.06946564885496183, + "reconstruction_s": 1.1480811000001268, + "eval_s": 1.2169707000002745, + "log_likelihood": 4796.742679497218, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6365933999995832, + "n": 1310, + "solve_s": 0.6365933999995832, + "outer_iterations": 151, + "inner_iterations": 14, + "total_iterations": 165, + "n_passive": 1210, + "warm_start_errors": 91, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 0.7883185000000594, + "eval_s": 0.8550965000003998, + "log_likelihood": 5609.2599565704595, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30176250000022264, + "n": 1310, + "solve_s": 0.30176250000022264, + "outer_iterations": 66, + "inner_iterations": 6, + "total_iterations": 72, + "n_passive": 1237, + "warm_start_errors": 59, + "index": 4 + }, + { + "warm_start_error_fraction": 0.051908396946564885, + "reconstruction_s": 0.7395175999999992, + "eval_s": 0.8072948999997607, + "log_likelihood": 5670.369454845888, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24029850000079023, + "n": 1310, + "solve_s": 0.24029850000079023, + "outer_iterations": 60, + "inner_iterations": 7, + "total_iterations": 67, + "n_passive": 1227, + "warm_start_errors": 68, + "index": 5 + }, + { + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 0.8050179000001663, + "eval_s": 0.8707410999995773, + "log_likelihood": 5308.701747308847, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2525400000004083, + "n": 1310, + "solve_s": 0.2525400000004083, + "outer_iterations": 58, + "inner_iterations": 20, + "total_iterations": 78, + "n_passive": 1197, + "warm_start_errors": 80, + "index": 6 + }, + { + "warm_start_error_fraction": 0.08015267175572519, + "reconstruction_s": 0.8274483999994118, + "eval_s": 0.8892958999995244, + "log_likelihood": 5005.220519553333, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3817343000000619, + "n": 1310, + "solve_s": 0.3817343000000619, + "outer_iterations": 98, + "inner_iterations": 9, + "total_iterations": 107, + "n_passive": 1238, + "warm_start_errors": 105, + "index": 7 + }, + { + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 0.70939510000062, + "eval_s": 0.7805852999999843, + "log_likelihood": 5349.678071715665, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22957389999919542, + "n": 1310, + "solve_s": 0.22957389999919542, + "outer_iterations": 52, + "inner_iterations": 3, + "total_iterations": 55, + "n_passive": 1214, + "warm_start_errors": 80, + "index": 8 + }, + { + "warm_start_error_fraction": 0.04732824427480916, + "reconstruction_s": 0.8809005000002799, + "eval_s": 0.9532406999996965, + "log_likelihood": 3426.9188812726425, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26786470000024565, + "n": 1310, + "solve_s": 0.26786470000024565, + "outer_iterations": 50, + "inner_iterations": 13, + "total_iterations": 63, + "n_passive": 1210, + "warm_start_errors": 62, + "index": 9 + }, + { + "warm_start_error_fraction": 0.03740458015267176, + "reconstruction_s": 0.7688385999999809, + "eval_s": 0.8363656999999876, + "log_likelihood": 5461.423080103182, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2318334000001414, + "n": 1310, + "solve_s": 0.2318334000001414, + "outer_iterations": 46, + "inner_iterations": 8, + "total_iterations": 54, + "n_passive": 1229, + "warm_start_errors": 49, + "index": 10 + }, + { + "warm_start_error_fraction": 0.042748091603053436, + "reconstruction_s": 0.761921600000278, + "eval_s": 0.8414268000005904, + "log_likelihood": 4099.747849461937, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.17987649999940913, + "n": 1310, + "solve_s": 0.17987649999940913, + "outer_iterations": 35, + "inner_iterations": 4, + "total_iterations": 39, + "n_passive": 1207, + "warm_start_errors": 56, + "index": 11 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 0.9001066999999239, + "eval_s": 0.9704072000004089, + "log_likelihood": 4875.353843213083, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.36123140000017884, + "n": 1310, + "solve_s": 0.36123140000017884, + "outer_iterations": 83, + "inner_iterations": 18, + "total_iterations": 101, + "n_passive": 1212, + "warm_start_errors": 79, + "index": 12 + }, + { + "warm_start_error_fraction": 0.03282442748091603, + "reconstruction_s": 0.6153148000003057, + "eval_s": 0.6833864000000176, + "log_likelihood": 5515.656174630203, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.13518759999988106, + "n": 1310, + "solve_s": 0.13518759999988106, + "outer_iterations": 26, + "inner_iterations": 2, + "total_iterations": 28, + "n_passive": 1211, + "warm_start_errors": 43, + "index": 13 + }, + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 0.7834411000003456, + "eval_s": 0.8489392999999836, + "log_likelihood": 5900.774014330928, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2826794000002337, + "n": 1310, + "solve_s": 0.2826794000002337, + "outer_iterations": 63, + "inner_iterations": 7, + "total_iterations": 70, + "n_passive": 1248, + "warm_start_errors": 59, + "index": 14 + }, + { + "warm_start_error_fraction": 0.030534351145038167, + "reconstruction_s": 0.7082891000000018, + "eval_s": 0.8047764000002644, + "log_likelihood": 5631.555877354901, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2043045999998867, + "n": 1310, + "solve_s": 0.2043045999998867, + "outer_iterations": 37, + "inner_iterations": 3, + "total_iterations": 40, + "n_passive": 1232, + "warm_start_errors": 40, + "index": 15 + }, + { + "warm_start_error_fraction": 0.04885496183206107, + "reconstruction_s": 0.7773747000001094, + "eval_s": 0.8435005999999703, + "log_likelihood": 5622.214876458538, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28324570000040694, + "n": 1310, + "solve_s": 0.28324570000040694, + "outer_iterations": 71, + "inner_iterations": 15, + "total_iterations": 86, + "n_passive": 1216, + "warm_start_errors": 64, + "index": 16 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 0.7538948000001255, + "eval_s": 0.8351506000008158, + "log_likelihood": 5901.336890546646, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2717466999993121, + "n": 1310, + "solve_s": 0.2717466999993121, + "outer_iterations": 62, + "inner_iterations": 8, + "total_iterations": 70, + "n_passive": 1236, + "warm_start_errors": 66, + "index": 17 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.8033079000006182, + "eval_s": 0.8750866999998834, + "log_likelihood": 4990.436852851273, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30390210000041407, + "n": 1310, + "solve_s": 0.30390210000041407, + "outer_iterations": 77, + "inner_iterations": 5, + "total_iterations": 82, + "n_passive": 1229, + "warm_start_errors": 73, + "index": 18 + }, + { + "warm_start_error_fraction": 0.03435114503816794, + "reconstruction_s": 0.7063617000003433, + "eval_s": 0.7798271999999997, + "log_likelihood": 4348.438167928539, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18556829999943147, + "n": 1310, + "solve_s": 0.18556829999943147, + "outer_iterations": 30, + "inner_iterations": 6, + "total_iterations": 36, + "n_passive": 1218, + "warm_start_errors": 45, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_adapt_reg_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_adapt_reg_v2026.8.17.1.png new file mode 100644 index 000000000..443dedb48 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_adapt_reg_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_2000_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_2000_v2026.8.17.1.json new file mode 100644 index 000000000..40338847b --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_2000_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "mesh_2000", + "configuration": { + "model": "mesh_2000", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 2000, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_2000", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 4.116752099999758, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.05679611650485437, + "mean_warm_start_error_fraction": 0.056432038834951445, + "p90_warm_start_error_fraction": 0.06019417475728155, + "max_warm_start_error_fraction": 0.062135922330097085, + "median_outer_iterations": 85.0, + "mean_outer_iterations": 84.3, + "p90_outer_iterations": 88.60000000000001, + "max_outer_iterations": 114.0, + "median_inner_iterations": 11.0, + "mean_inner_iterations": 11.25, + "p90_inner_iterations": 13.0, + "max_inner_iterations": 20.0, + "median_total_iterations": 96.0, + "mean_total_iterations": 95.55, + "p90_total_iterations": 100.60000000000001, + "max_total_iterations": 134.0, + "median_n_passive": 1890.5, + "mean_n_passive": 1890.6, + "p90_n_passive": 1895.2, + "max_n_passive": 1898.0, + "median_warm_start_errors": 117.0, + "mean_warm_start_errors": 116.25, + "p90_warm_start_errors": 124.0, + "max_warm_start_errors": 128.0, + "median_solve_s": 1.37628820000009, + "mean_solve_s": 1.411632069999905, + "p90_solve_s": 1.5335996800004068, + "max_solve_s": 2.0416849000002912, + "median_reconstruction_s": 2.254718000000139, + "mean_reconstruction_s": 2.276492499999995, + "p90_reconstruction_s": 2.541147980000005, + "max_reconstruction_s": 3.0187306000007084, + "median_eval_s": 2.613825049999832, + "mean_eval_s": 2.6316371749998324, + "p90_eval_s": 2.9080870900002083, + "max_eval_s": 3.3858933000001343 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.00825242718446602, + "mean_warm_start_error_fraction": 0.010873786407766988, + "p90_warm_start_error_fraction": 0.013446601941747578, + "max_warm_start_error_fraction": 0.05339805825242718, + "median_outer_iterations": 11.0, + "mean_outer_iterations": 13.9, + "p90_outer_iterations": 21.0, + "max_outer_iterations": 75.0, + "median_inner_iterations": 2.0, + "mean_inner_iterations": 2.45, + "p90_inner_iterations": 3.1000000000000014, + "max_inner_iterations": 10.0, + "median_total_iterations": 13.0, + "mean_total_iterations": 16.35, + "p90_total_iterations": 24.1, + "max_total_iterations": 85.0, + "median_n_passive": 1889.5, + "mean_n_passive": 1890.05, + "p90_n_passive": 1895.1, + "max_n_passive": 1898.0, + "median_warm_start_errors": 17.0, + "mean_warm_start_errors": 22.4, + "p90_warm_start_errors": 27.70000000000001, + "max_warm_start_errors": 110.0, + "median_solve_s": 0.3038364500002899, + "mean_solve_s": 0.3694345150001027, + "p90_solve_s": 0.44116445999970894, + "max_solve_s": 1.3467273999995086, + "median_reconstruction_s": 0.9827368999999635, + "mean_reconstruction_s": 1.106437224999945, + "p90_reconstruction_s": 1.44613511999969, + "max_reconstruction_s": 2.2000693999998475, + "median_eval_s": 1.3016384499992455, + "mean_eval_s": 1.4760885749999033, + "p90_eval_s": 1.8900839200001431, + "max_eval_s": 2.5298681000003853 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 7.68226248043455e-16, + "max_abs_reconstruction_deviation": 5.264997327003584e-10 + }, + "iteration_reduction_ratio": 7.384615384615385, + "solve_speedup_ratio": 4.529700764996355, + "eval_speedup_ratio": 2.0081037480118593, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.05339805825242718, + "reconstruction_s": 2.2974661000007472, + "eval_s": 2.6235790999999153, + "log_likelihood": 7168.794400171308, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4192716999996264, + "n": 2060, + "solve_s": 1.4192716999996264, + "outer_iterations": 75, + "inner_iterations": 10, + "total_iterations": 85, + "n_passive": 1889, + "warm_start_errors": 110, + "index": 0 + }, + { + "warm_start_error_fraction": 0.055825242718446605, + "reconstruction_s": 2.2242537000001903, + "eval_s": 2.601048799999262, + "log_likelihood": 7139.382884402501, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2887858999993114, + "n": 2060, + "solve_s": 1.2887858999993114, + "outer_iterations": 81, + "inner_iterations": 11, + "total_iterations": 92, + "n_passive": 1892, + "warm_start_errors": 115, + "index": 1 + }, + { + "warm_start_error_fraction": 0.05825242718446602, + "reconstruction_s": 2.168484100000569, + "eval_s": 2.565273000000161, + "log_likelihood": 7101.162622566544, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3717987999998513, + "n": 2060, + "solve_s": 1.3717987999998513, + "outer_iterations": 83, + "inner_iterations": 13, + "total_iterations": 96, + "n_passive": 1891, + "warm_start_errors": 120, + "index": 2 + }, + { + "warm_start_error_fraction": 0.055825242718446605, + "reconstruction_s": 2.1143466999992597, + "eval_s": 2.4537928999998257, + "log_likelihood": 7144.577737286041, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2658944999993764, + "n": 2060, + "solve_s": 1.2658944999993764, + "outer_iterations": 86, + "inner_iterations": 12, + "total_iterations": 98, + "n_passive": 1891, + "warm_start_errors": 115, + "index": 3 + }, + { + "warm_start_error_fraction": 0.05631067961165048, + "reconstruction_s": 2.3061430999996446, + "eval_s": 2.6588937999995323, + "log_likelihood": 7146.185492536062, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.493154499999946, + "n": 2060, + "solve_s": 1.493154499999946, + "outer_iterations": 88, + "inner_iterations": 9, + "total_iterations": 97, + "n_passive": 1895, + "warm_start_errors": 116, + "index": 4 + }, + { + "warm_start_error_fraction": 0.058737864077669906, + "reconstruction_s": 2.102746699999443, + "eval_s": 2.440457499999866, + "log_likelihood": 7103.540079729251, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2369738999996116, + "n": 2060, + "solve_s": 1.2369738999996116, + "outer_iterations": 79, + "inner_iterations": 11, + "total_iterations": 90, + "n_passive": 1887, + "warm_start_errors": 121, + "index": 5 + }, + { + "warm_start_error_fraction": 0.053883495145631066, + "reconstruction_s": 2.04512319999958, + "eval_s": 2.363967500000399, + "log_likelihood": 7164.00335516612, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3053212999993775, + "n": 2060, + "solve_s": 1.3053212999993775, + "outer_iterations": 85, + "inner_iterations": 12, + "total_iterations": 97, + "n_passive": 1894, + "warm_start_errors": 111, + "index": 6 + }, + { + "warm_start_error_fraction": 0.053883495145631066, + "reconstruction_s": 2.145369999999275, + "eval_s": 2.473915099999431, + "log_likelihood": 7186.687099821456, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3485581999993883, + "n": 2060, + "solve_s": 1.3485581999993883, + "outer_iterations": 82, + "inner_iterations": 10, + "total_iterations": 92, + "n_passive": 1894, + "warm_start_errors": 111, + "index": 7 + }, + { + "warm_start_error_fraction": 0.05145631067961165, + "reconstruction_s": 1.9908771000000343, + "eval_s": 2.323621199999252, + "log_likelihood": 7205.084169644729, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2127086999998937, + "n": 2060, + "solve_s": 1.2127086999998937, + "outer_iterations": 71, + "inner_iterations": 9, + "total_iterations": 80, + "n_passive": 1893, + "warm_start_errors": 106, + "index": 8 + }, + { + "warm_start_error_fraction": 0.05776699029126214, + "reconstruction_s": 1.9577031999997416, + "eval_s": 2.2770341999994343, + "log_likelihood": 7121.357015894441, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.194297000000006, + "n": 2060, + "solve_s": 1.194297000000006, + "outer_iterations": 80, + "inner_iterations": 12, + "total_iterations": 92, + "n_passive": 1888, + "warm_start_errors": 119, + "index": 9 + }, + { + "warm_start_error_fraction": 0.062135922330097085, + "reconstruction_s": 2.248355100000481, + "eval_s": 2.6381919000004928, + "log_likelihood": 7103.33475917487, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.48424630000045, + "n": 2060, + "solve_s": 1.48424630000045, + "outer_iterations": 94, + "inner_iterations": 12, + "total_iterations": 106, + "n_passive": 1888, + "warm_start_errors": 128, + "index": 10 + }, + { + "warm_start_error_fraction": 0.05776699029126214, + "reconstruction_s": 2.287729599999693, + "eval_s": 2.6829649000001154, + "log_likelihood": 7095.903384117048, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3588442999998733, + "n": 2060, + "solve_s": 1.3588442999998733, + "outer_iterations": 86, + "inner_iterations": 11, + "total_iterations": 97, + "n_passive": 1888, + "warm_start_errors": 119, + "index": 11 + }, + { + "warm_start_error_fraction": 0.059223300970873784, + "reconstruction_s": 3.0187306000007084, + "eval_s": 3.3858933000001343, + "log_likelihood": 7084.914486384768, + "n_solves": 1, + "retries": 0, + "solve_s_total": 2.0416849000002912, + "n": 2060, + "solve_s": 2.0416849000002912, + "outer_iterations": 114, + "inner_iterations": 20, + "total_iterations": 134, + "n_passive": 1890, + "warm_start_errors": 122, + "index": 12 + }, + { + "warm_start_error_fraction": 0.06019417475728155, + "reconstruction_s": 2.6338181000000986, + "eval_s": 2.99553190000006, + "log_likelihood": 7100.536977853349, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.573719699999856, + "n": 2060, + "solve_s": 1.573719699999856, + "outer_iterations": 87, + "inner_iterations": 13, + "total_iterations": 100, + "n_passive": 1888, + "warm_start_errors": 124, + "index": 13 + }, + { + "warm_start_error_fraction": 0.05825242718446602, + "reconstruction_s": 2.3510009999999966, + "eval_s": 2.740910999999869, + "log_likelihood": 7074.720794683303, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.529141900000468, + "n": 2060, + "solve_s": 1.529141900000468, + "outer_iterations": 88, + "inner_iterations": 7, + "total_iterations": 95, + "n_passive": 1888, + "warm_start_errors": 120, + "index": 14 + }, + { + "warm_start_error_fraction": 0.05728155339805825, + "reconstruction_s": 2.530851299999995, + "eval_s": 2.8983710000002247, + "log_likelihood": 7090.964713357198, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4844207999994978, + "n": 2060, + "solve_s": 1.4844207999994978, + "outer_iterations": 85, + "inner_iterations": 11, + "total_iterations": 96, + "n_passive": 1880, + "warm_start_errors": 118, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06019417475728155, + "reconstruction_s": 2.3919381999994584, + "eval_s": 2.737845599999673, + "log_likelihood": 7142.194179626887, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.493776300000718, + "n": 2060, + "solve_s": 1.493776300000718, + "outer_iterations": 87, + "inner_iterations": 11, + "total_iterations": 98, + "n_passive": 1890, + "warm_start_errors": 124, + "index": 16 + }, + { + "warm_start_error_fraction": 0.055825242718446605, + "reconstruction_s": 2.3813960000006773, + "eval_s": 2.7219648999998753, + "log_likelihood": 7185.524532632409, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.5263396999998804, + "n": 2060, + "solve_s": 1.5263396999998804, + "outer_iterations": 86, + "inner_iterations": 10, + "total_iterations": 96, + "n_passive": 1898, + "warm_start_errors": 115, + "index": 17 + }, + { + "warm_start_error_fraction": 0.052912621359223304, + "reconstruction_s": 2.261080899999797, + "eval_s": 2.604070999999749, + "log_likelihood": 7213.258628567402, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3807776000003287, + "n": 2060, + "solve_s": 1.3807776000003287, + "outer_iterations": 83, + "inner_iterations": 11, + "total_iterations": 94, + "n_passive": 1897, + "warm_start_errors": 109, + "index": 18 + }, + { + "warm_start_error_fraction": 0.04951456310679612, + "reconstruction_s": 2.0724353000005067, + "eval_s": 2.4454148999993777, + "log_likelihood": 7317.585677415075, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2229254000003493, + "n": 2060, + "solve_s": 1.2229254000003493, + "outer_iterations": 66, + "inner_iterations": 10, + "total_iterations": 76, + "n_passive": 1891, + "warm_start_errors": 102, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.05339805825242718, + "reconstruction_s": 2.2000693999998475, + "eval_s": 2.5298681000003853, + "log_likelihood": 7168.794400171308, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3467273999995086, + "n": 2060, + "solve_s": 1.3467273999995086, + "outer_iterations": 75, + "inner_iterations": 10, + "total_iterations": 85, + "n_passive": 1889, + "warm_start_errors": 110, + "index": 0 + }, + { + "warm_start_error_fraction": 0.007281553398058253, + "reconstruction_s": 0.9164514000003692, + "eval_s": 1.2749473999992915, + "log_likelihood": 7139.382884402503, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28177529999993567, + "n": 2060, + "solve_s": 0.28177529999993567, + "outer_iterations": 11, + "inner_iterations": 1, + "total_iterations": 12, + "n_passive": 1892, + "warm_start_errors": 15, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0048543689320388345, + "reconstruction_s": 0.9071113999998488, + "eval_s": 1.2579052000000956, + "log_likelihood": 7101.162622566546, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25807590000022174, + "n": 2060, + "solve_s": 0.25807590000022174, + "outer_iterations": 6, + "inner_iterations": 2, + "total_iterations": 8, + "n_passive": 1890, + "warm_start_errors": 10, + "index": 2 + }, + { + "warm_start_error_fraction": 0.009223300970873786, + "reconstruction_s": 0.9371467000000848, + "eval_s": 1.253027600000678, + "log_likelihood": 7144.577737286043, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2693040000003748, + "n": 2060, + "solve_s": 0.2693040000003748, + "outer_iterations": 11, + "inner_iterations": 3, + "total_iterations": 14, + "n_passive": 1891, + "warm_start_errors": 19, + "index": 3 + }, + { + "warm_start_error_fraction": 0.003883495145631068, + "reconstruction_s": 0.8866232000000309, + "eval_s": 1.2045911000004708, + "log_likelihood": 7146.185492536064, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23824710000008054, + "n": 2060, + "solve_s": 0.23824710000008054, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1895, + "warm_start_errors": 8, + "index": 4 + }, + { + "warm_start_error_fraction": 0.00825242718446602, + "reconstruction_s": 0.8472850000007384, + "eval_s": 1.1846178999994663, + "log_likelihood": 7103.540079729248, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2608373000002757, + "n": 2060, + "solve_s": 0.2608373000002757, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1886, + "warm_start_errors": 17, + "index": 5 + }, + { + "warm_start_error_fraction": 0.007766990291262136, + "reconstruction_s": 0.982503899999756, + "eval_s": 1.2886033999993742, + "log_likelihood": 7164.003355166121, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26873500000056083, + "n": 2060, + "solve_s": 0.26873500000056083, + "outer_iterations": 12, + "inner_iterations": 2, + "total_iterations": 14, + "n_passive": 1894, + "warm_start_errors": 16, + "index": 6 + }, + { + "warm_start_error_fraction": 0.00825242718446602, + "reconstruction_s": 0.982969900000171, + "eval_s": 1.330869899999925, + "log_likelihood": 7186.687099821453, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31709130000035657, + "n": 2060, + "solve_s": 0.31709130000035657, + "outer_iterations": 8, + "inner_iterations": 2, + "total_iterations": 10, + "n_passive": 1893, + "warm_start_errors": 17, + "index": 7 + }, + { + "warm_start_error_fraction": 0.01116504854368932, + "reconstruction_s": 0.9931353000001764, + "eval_s": 1.3146734999991168, + "log_likelihood": 7205.084169644727, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31072630000016943, + "n": 2060, + "solve_s": 0.31072630000016943, + "outer_iterations": 12, + "inner_iterations": 2, + "total_iterations": 14, + "n_passive": 1892, + "warm_start_errors": 23, + "index": 8 + }, + { + "warm_start_error_fraction": 0.007281553398058253, + "reconstruction_s": 0.9556463999997504, + "eval_s": 1.2857793999992282, + "log_likelihood": 7121.35701589444, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2751331999998001, + "n": 2060, + "solve_s": 0.2751331999998001, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 1887, + "warm_start_errors": 15, + "index": 9 + }, + { + "warm_start_error_fraction": 0.01262135922330097, + "reconstruction_s": 0.9222577999998975, + "eval_s": 1.2593434000000343, + "log_likelihood": 7103.334759174864, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3135331000003134, + "n": 2060, + "solve_s": 0.3135331000003134, + "outer_iterations": 15, + "inner_iterations": 3, + "total_iterations": 18, + "n_passive": 1887, + "warm_start_errors": 26, + "index": 10 + }, + { + "warm_start_error_fraction": 0.01650485436893204, + "reconstruction_s": 0.9572413000005326, + "eval_s": 1.2734300000001895, + "log_likelihood": 7095.90338411705, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.343724000000293, + "n": 2060, + "solve_s": 0.343724000000293, + "outer_iterations": 21, + "inner_iterations": 4, + "total_iterations": 25, + "n_passive": 1887, + "warm_start_errors": 34, + "index": 11 + }, + { + "warm_start_error_fraction": 0.003883495145631068, + "reconstruction_s": 0.8759265999997297, + "eval_s": 1.2571860999996716, + "log_likelihood": 7084.91448638477, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24232759999995324, + "n": 2060, + "solve_s": 0.24232759999995324, + "outer_iterations": 7, + "inner_iterations": 2, + "total_iterations": 9, + "n_passive": 1889, + "warm_start_errors": 8, + "index": 12 + }, + { + "warm_start_error_fraction": 0.002912621359223301, + "reconstruction_s": 1.1666678000001411, + "eval_s": 1.7312628999998196, + "log_likelihood": 7100.536977853344, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2953941000005216, + "n": 2060, + "solve_s": 0.2953941000005216, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1887, + "warm_start_errors": 6, + "index": 13 + }, + { + "warm_start_error_fraction": 0.007281553398058253, + "reconstruction_s": 1.5447050999991916, + "eval_s": 2.034509799999796, + "log_likelihood": 7074.720794683301, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41840979999960837, + "n": 2060, + "solve_s": 0.41840979999960837, + "outer_iterations": 11, + "inner_iterations": 1, + "total_iterations": 12, + "n_passive": 1888, + "warm_start_errors": 15, + "index": 14 + }, + { + "warm_start_error_fraction": 0.011650485436893204, + "reconstruction_s": 1.2515177999994194, + "eval_s": 1.7711522999998124, + "log_likelihood": 7090.964713357199, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43865639999967243, + "n": 2060, + "solve_s": 0.43865639999967243, + "outer_iterations": 12, + "inner_iterations": 2, + "total_iterations": 14, + "n_passive": 1880, + "warm_start_errors": 24, + "index": 15 + }, + { + "warm_start_error_fraction": 0.010194174757281554, + "reconstruction_s": 1.4351828999997451, + "eval_s": 1.8740366000001814, + "log_likelihood": 7142.194179626887, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.46373700000003737, + "n": 2060, + "solve_s": 0.46373700000003737, + "outer_iterations": 18, + "inner_iterations": 3, + "total_iterations": 21, + "n_passive": 1889, + "warm_start_errors": 21, + "index": 16 + }, + { + "warm_start_error_fraction": 0.013106796116504855, + "reconstruction_s": 1.3383484999994835, + "eval_s": 1.6343543000002683, + "log_likelihood": 7185.524532632406, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42148899999938294, + "n": 2060, + "solve_s": 0.42148899999938294, + "outer_iterations": 21, + "inner_iterations": 3, + "total_iterations": 24, + "n_passive": 1898, + "warm_start_errors": 27, + "index": 17 + }, + { + "warm_start_error_fraction": 0.005825242718446602, + "reconstruction_s": 1.0382705000001806, + "eval_s": 1.4148764000001393, + "log_likelihood": 7213.258628567403, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2969466000004104, + "n": 2060, + "solve_s": 0.2969466000004104, + "outer_iterations": 7, + "inner_iterations": 3, + "total_iterations": 10, + "n_passive": 1896, + "warm_start_errors": 12, + "index": 18 + }, + { + "warm_start_error_fraction": 0.012135922330097087, + "reconstruction_s": 0.9896835999998075, + "eval_s": 1.346736200000123, + "log_likelihood": 7317.585677415076, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32781990000057704, + "n": 2060, + "solve_s": 0.32781990000057704, + "outer_iterations": 12, + "inner_iterations": 2, + "total_iterations": 14, + "n_passive": 1891, + "warm_start_errors": 25, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.06407766990291261, + "mean_warm_start_error_fraction": 0.06313106796116505, + "p90_warm_start_error_fraction": 0.08349514563106798, + "max_warm_start_error_fraction": 0.10145631067961165, + "median_outer_iterations": 99.0, + "mean_outer_iterations": 97.0, + "p90_outer_iterations": 120.40000000000003, + "max_outer_iterations": 154.0, + "median_inner_iterations": 15.0, + "mean_inner_iterations": 16.85, + "p90_inner_iterations": 23.200000000000003, + "max_inner_iterations": 46.0, + "median_total_iterations": 113.5, + "mean_total_iterations": 113.85, + "p90_total_iterations": 154.60000000000002, + "max_total_iterations": 179.0, + "median_n_passive": 1875.0, + "mean_n_passive": 1874.4, + "p90_n_passive": 1894.1, + "max_n_passive": 1905.0, + "median_warm_start_errors": 132.0, + "mean_warm_start_errors": 130.05, + "p90_warm_start_errors": 172.00000000000003, + "max_warm_start_errors": 209.0, + "median_solve_s": 1.6012338000000454, + "mean_solve_s": 1.6577870150001217, + "p90_solve_s": 1.8805894399998584, + "max_solve_s": 2.5254463000001124, + "median_reconstruction_s": 2.720346950000021, + "mean_reconstruction_s": 2.6965261450000524, + "p90_reconstruction_s": 3.087620010000137, + "max_reconstruction_s": 3.5225310000005265, + "median_eval_s": 3.1026805499996044, + "mean_eval_s": 3.0942745499999544, + "p90_eval_s": 3.546616619999804, + "max_eval_s": 3.973676399999931 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.06456310679611652, + "mean_warm_start_error_fraction": 0.06570388349514564, + "p90_warm_start_error_fraction": 0.08276699029126214, + "max_warm_start_error_fraction": 0.08932038834951456, + "median_outer_iterations": 108.0, + "mean_outer_iterations": 104.35, + "p90_outer_iterations": 116.30000000000001, + "max_outer_iterations": 160.0, + "median_inner_iterations": 19.0, + "mean_inner_iterations": 21.15, + "p90_inner_iterations": 31.1, + "max_inner_iterations": 40.0, + "median_total_iterations": 128.5, + "mean_total_iterations": 125.5, + "p90_total_iterations": 139.20000000000002, + "max_total_iterations": 200.0, + "median_n_passive": 1875.0, + "mean_n_passive": 1874.2, + "p90_n_passive": 1894.1, + "max_n_passive": 1905.0, + "median_warm_start_errors": 133.0, + "mean_warm_start_errors": 135.35, + "p90_warm_start_errors": 170.5, + "max_warm_start_errors": 184.0, + "median_solve_s": 1.2076707500000339, + "mean_solve_s": 1.2613503949999996, + "p90_solve_s": 1.7917644899996958, + "max_solve_s": 1.8993196000001262, + "median_reconstruction_s": 1.8946382999997695, + "mean_reconstruction_s": 2.014927805000025, + "p90_reconstruction_s": 2.6191002699996715, + "max_reconstruction_s": 2.8405941000000894, + "median_eval_s": 2.2858760999997685, + "mean_eval_s": 2.376647279999952, + "p90_eval_s": 2.979937719999998, + "max_eval_s": 3.2149389999995037 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.2565885654695481e-14, + "max_abs_reconstruction_deviation": 5.2586823784395165e-09 + }, + "iteration_reduction_ratio": 0.8832684824902723, + "solve_speedup_ratio": 1.3258860496538485, + "eval_speedup_ratio": 1.357326650381409, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.0825242718446602, + "reconstruction_s": 2.448574600000029, + "eval_s": 2.779732099999819, + "log_likelihood": 5456.813211559655, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.5582609000002776, + "n": 2060, + "solve_s": 1.5582609000002776, + "outer_iterations": 114, + "inner_iterations": 16, + "total_iterations": 130, + "n_passive": 1837, + "warm_start_errors": 170, + "index": 0 + }, + { + "warm_start_error_fraction": 0.06601941747572816, + "reconstruction_s": 3.5225310000005265, + "eval_s": 3.973676399999931, + "log_likelihood": 6612.666791665338, + "n_solves": 1, + "retries": 0, + "solve_s_total": 2.5254463000001124, + "n": 2060, + "solve_s": 2.5254463000001124, + "outer_iterations": 107, + "inner_iterations": 15, + "total_iterations": 122, + "n_passive": 1875, + "warm_start_errors": 136, + "index": 1 + }, + { + "warm_start_error_fraction": 0.050970873786407765, + "reconstruction_s": 2.7497192000000723, + "eval_s": 3.3039381999997204, + "log_likelihood": 7192.744509428636, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.7771730000004027, + "n": 2060, + "solve_s": 1.7771730000004027, + "outer_iterations": 69, + "inner_iterations": 13, + "total_iterations": 82, + "n_passive": 1894, + "warm_start_errors": 105, + "index": 2 + }, + { + "warm_start_error_fraction": 0.070873786407767, + "reconstruction_s": 3.1125276000002486, + "eval_s": 3.5462079999997513, + "log_likelihood": 6567.040643382352, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.8443717999998626, + "n": 2060, + "solve_s": 1.8443717999998626, + "outer_iterations": 93, + "inner_iterations": 12, + "total_iterations": 105, + "n_passive": 1858, + "warm_start_errors": 146, + "index": 3 + }, + { + "warm_start_error_fraction": 0.053883495145631066, + "reconstruction_s": 2.6883409000001848, + "eval_s": 3.068059500000345, + "log_likelihood": 7024.427612850599, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4770466999998462, + "n": 2060, + "solve_s": 1.4770466999998462, + "outer_iterations": 75, + "inner_iterations": 16, + "total_iterations": 91, + "n_passive": 1889, + "warm_start_errors": 111, + "index": 4 + }, + { + "warm_start_error_fraction": 0.06796116504854369, + "reconstruction_s": 2.8803225999999995, + "eval_s": 3.32901049999964, + "log_likelihood": 6997.2075080046625, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.7718291000001045, + "n": 2060, + "solve_s": 1.7718291000001045, + "outer_iterations": 101, + "inner_iterations": 8, + "total_iterations": 109, + "n_passive": 1885, + "warm_start_errors": 140, + "index": 5 + }, + { + "warm_start_error_fraction": 0.06650485436893204, + "reconstruction_s": 3.0172395999998116, + "eval_s": 3.4798666999995476, + "log_likelihood": 6816.842555209736, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.6237927000001946, + "n": 2060, + "solve_s": 1.6237927000001946, + "outer_iterations": 99, + "inner_iterations": 15, + "total_iterations": 114, + "n_passive": 1875, + "warm_start_errors": 137, + "index": 6 + }, + { + "warm_start_error_fraction": 0.06504854368932039, + "reconstruction_s": 3.017143600000054, + "eval_s": 3.4042891999997664, + "log_likelihood": 6474.862989576887, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.7410622000006697, + "n": 2060, + "solve_s": 1.7410622000006697, + "outer_iterations": 74, + "inner_iterations": 11, + "total_iterations": 85, + "n_passive": 1862, + "warm_start_errors": 134, + "index": 7 + }, + { + "warm_start_error_fraction": 0.06359223300970873, + "reconstruction_s": 2.2343842000000222, + "eval_s": 2.6321266999993895, + "log_likelihood": 6564.892567829057, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4110067000001436, + "n": 2060, + "solve_s": 1.4110067000001436, + "outer_iterations": 96, + "inner_iterations": 17, + "total_iterations": 113, + "n_passive": 1869, + "warm_start_errors": 131, + "index": 8 + }, + { + "warm_start_error_fraction": 0.09223300970873786, + "reconstruction_s": 2.931874799999605, + "eval_s": 3.2681401999998343, + "log_likelihood": 5138.843818919247, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.7355458000001818, + "n": 2060, + "solve_s": 1.7355458000001818, + "outer_iterations": 142, + "inner_iterations": 18, + "total_iterations": 160, + "n_passive": 1850, + "warm_start_errors": 190, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0645631067961165, + "reconstruction_s": 2.6909746999999697, + "eval_s": 3.085081199999877, + "log_likelihood": 6869.659844693891, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.7381030000005921, + "n": 2060, + "solve_s": 1.7381030000005921, + "outer_iterations": 100, + "inner_iterations": 15, + "total_iterations": 115, + "n_passive": 1870, + "warm_start_errors": 133, + "index": 10 + }, + { + "warm_start_error_fraction": 0.10145631067961165, + "reconstruction_s": 3.0848525000001246, + "eval_s": 3.5502942000002804, + "log_likelihood": 5613.838981870756, + "n_solves": 1, + "retries": 0, + "solve_s_total": 2.2065481999998156, + "n": 2060, + "solve_s": 2.2065481999998156, + "outer_iterations": 154, + "inner_iterations": 25, + "total_iterations": 179, + "n_passive": 1850, + "warm_start_errors": 209, + "index": 11 + }, + { + "warm_start_error_fraction": 0.05728155339805825, + "reconstruction_s": 2.7579051000002437, + "eval_s": 3.2158219000002646, + "log_likelihood": 6605.484157728401, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.5510840999995708, + "n": 2060, + "solve_s": 1.5510840999995708, + "outer_iterations": 74, + "inner_iterations": 15, + "total_iterations": 89, + "n_passive": 1865, + "warm_start_errors": 118, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05, + "reconstruction_s": 2.399833700000272, + "eval_s": 2.7541235000007873, + "log_likelihood": 7115.742449288045, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3153157999995528, + "n": 2060, + "solve_s": 1.3153157999995528, + "outer_iterations": 72, + "inner_iterations": 23, + "total_iterations": 95, + "n_passive": 1879, + "warm_start_errors": 103, + "index": 13 + }, + { + "warm_start_error_fraction": 0.04223300970873786, + "reconstruction_s": 2.7830157000007603, + "eval_s": 3.120279899999332, + "log_likelihood": 7157.754505112721, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.8434857000002012, + "n": 2060, + "solve_s": 1.8434857000002012, + "outer_iterations": 108, + "inner_iterations": 46, + "total_iterations": 154, + "n_passive": 1895, + "warm_start_errors": 87, + "index": 14 + }, + { + "warm_start_error_fraction": 0.039805825242718446, + "reconstruction_s": 1.9633509000004778, + "eval_s": 2.313251399999899, + "log_likelihood": 7130.537395088721, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.1686537000005046, + "n": 2060, + "solve_s": 1.1686537000005046, + "outer_iterations": 63, + "inner_iterations": 15, + "total_iterations": 78, + "n_passive": 1894, + "warm_start_errors": 82, + "index": 15 + }, + { + "warm_start_error_fraction": 0.052912621359223304, + "reconstruction_s": 2.2988394999993034, + "eval_s": 2.65389309999955, + "log_likelihood": 7038.376811999479, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3864352999999028, + "n": 2060, + "solve_s": 1.3864352999999028, + "outer_iterations": 80, + "inner_iterations": 13, + "total_iterations": 93, + "n_passive": 1890, + "warm_start_errors": 109, + "index": 16 + }, + { + "warm_start_error_fraction": 0.04563106796116505, + "reconstruction_s": 2.3716406999992614, + "eval_s": 2.7080966000003173, + "log_likelihood": 7398.450536271065, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4687516000003598, + "n": 2060, + "solve_s": 1.4687516000003598, + "outer_iterations": 102, + "inner_iterations": 14, + "total_iterations": 116, + "n_passive": 1905, + "warm_start_errors": 94, + "index": 17 + }, + { + "warm_start_error_fraction": 0.062135922330097085, + "reconstruction_s": 2.292295700000068, + "eval_s": 2.6719297000008737, + "log_likelihood": 6543.698875010002, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4331528000002436, + "n": 2060, + "solve_s": 1.4331528000002436, + "outer_iterations": 99, + "inner_iterations": 15, + "total_iterations": 114, + "n_passive": 1884, + "warm_start_errors": 128, + "index": 18 + }, + { + "warm_start_error_fraction": 0.06699029126213592, + "reconstruction_s": 2.685156300000017, + "eval_s": 3.027672000000166, + "log_likelihood": 6154.717062854226, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.5786748999998963, + "n": 2060, + "solve_s": 1.5786748999998963, + "outer_iterations": 118, + "inner_iterations": 15, + "total_iterations": 133, + "n_passive": 1862, + "warm_start_errors": 138, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.0825242718446602, + "reconstruction_s": 2.8405941000000894, + "eval_s": 3.2149389999995037, + "log_likelihood": 5456.813211559655, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.8081237000005785, + "n": 2060, + "solve_s": 1.8081237000005785, + "outer_iterations": 114, + "inner_iterations": 16, + "total_iterations": 130, + "n_passive": 1837, + "warm_start_errors": 170, + "index": 0 + }, + { + "warm_start_error_fraction": 0.06407766990291262, + "reconstruction_s": 2.145432399999663, + "eval_s": 2.555361100000482, + "log_likelihood": 6612.666791665342, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.489226100000451, + "n": 2060, + "solve_s": 1.489226100000451, + "outer_iterations": 115, + "inner_iterations": 17, + "total_iterations": 132, + "n_passive": 1875, + "warm_start_errors": 132, + "index": 1 + }, + { + "warm_start_error_fraction": 0.07524271844660194, + "reconstruction_s": 2.58334990000003, + "eval_s": 2.9798025000000052, + "log_likelihood": 7192.744509428667, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.7899467999995977, + "n": 2060, + "solve_s": 1.7899467999995977, + "outer_iterations": 113, + "inner_iterations": 19, + "total_iterations": 132, + "n_passive": 1894, + "warm_start_errors": 155, + "index": 2 + }, + { + "warm_start_error_fraction": 0.06310679611650485, + "reconstruction_s": 2.6515837000006286, + "eval_s": 2.9562839000000167, + "log_likelihood": 6567.040643382427, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2818443999994997, + "n": 2060, + "solve_s": 1.2818443999994997, + "outer_iterations": 113, + "inner_iterations": 22, + "total_iterations": 135, + "n_passive": 1858, + "warm_start_errors": 130, + "index": 3 + }, + { + "warm_start_error_fraction": 0.055825242718446605, + "reconstruction_s": 1.8442089999998643, + "eval_s": 2.202365899999677, + "log_likelihood": 7024.427612850613, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2049606000000495, + "n": 2060, + "solve_s": 1.2049606000000495, + "outer_iterations": 103, + "inner_iterations": 18, + "total_iterations": 121, + "n_passive": 1889, + "warm_start_errors": 115, + "index": 4 + }, + { + "warm_start_error_fraction": 0.059223300970873784, + "reconstruction_s": 2.1254649000002246, + "eval_s": 2.556758599999739, + "log_likelihood": 6997.207508004733, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3923162000000957, + "n": 2060, + "solve_s": 1.3923162000000957, + "outer_iterations": 107, + "inner_iterations": 31, + "total_iterations": 138, + "n_passive": 1885, + "warm_start_errors": 122, + "index": 5 + }, + { + "warm_start_error_fraction": 0.06504854368932039, + "reconstruction_s": 2.289653800000451, + "eval_s": 2.6873133999997663, + "log_likelihood": 6816.842555209735, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.442986000000019, + "n": 2060, + "solve_s": 1.442986000000019, + "outer_iterations": 94, + "inner_iterations": 32, + "total_iterations": 126, + "n_passive": 1875, + "warm_start_errors": 134, + "index": 6 + }, + { + "warm_start_error_fraction": 0.08495145631067962, + "reconstruction_s": 1.9651558000005025, + "eval_s": 2.282402799999545, + "log_likelihood": 6474.862989576899, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2103809000000183, + "n": 2060, + "solve_s": 1.2103809000000183, + "outer_iterations": 105, + "inner_iterations": 18, + "total_iterations": 123, + "n_passive": 1862, + "warm_start_errors": 175, + "index": 7 + }, + { + "warm_start_error_fraction": 0.06553398058252427, + "reconstruction_s": 2.05403749999914, + "eval_s": 2.4520001000000775, + "log_likelihood": 6564.892567829055, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4203403000001344, + "n": 2060, + "solve_s": 1.4203403000001344, + "outer_iterations": 128, + "inner_iterations": 22, + "total_iterations": 150, + "n_passive": 1869, + "warm_start_errors": 135, + "index": 8 + }, + { + "warm_start_error_fraction": 0.07718446601941747, + "reconstruction_s": 1.7352003000005425, + "eval_s": 2.0960456000002523, + "log_likelihood": 5138.843818919182, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0473327999998219, + "n": 2060, + "solve_s": 1.0473327999998219, + "outer_iterations": 97, + "inner_iterations": 23, + "total_iterations": 120, + "n_passive": 1850, + "warm_start_errors": 159, + "index": 9 + }, + { + "warm_start_error_fraction": 0.07233009708737864, + "reconstruction_s": 1.8592405999997936, + "eval_s": 2.1933911000005537, + "log_likelihood": 6869.659844693924, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.1466895000003205, + "n": 2060, + "solve_s": 1.1466895000003205, + "outer_iterations": 110, + "inner_iterations": 22, + "total_iterations": 132, + "n_passive": 1869, + "warm_start_errors": 149, + "index": 10 + }, + { + "warm_start_error_fraction": 0.06165048543689321, + "reconstruction_s": 1.7214026999999987, + "eval_s": 2.083181899999545, + "log_likelihood": 5613.838981870751, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0953702999995585, + "n": 2060, + "solve_s": 1.0953702999995585, + "outer_iterations": 109, + "inner_iterations": 18, + "total_iterations": 127, + "n_passive": 1850, + "warm_start_errors": 127, + "index": 11 + }, + { + "warm_start_error_fraction": 0.08932038834951456, + "reconstruction_s": 2.6154909999995652, + "eval_s": 2.9811546999999337, + "log_likelihood": 6605.484157728405, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.8993196000001262, + "n": 2060, + "solve_s": 1.8993196000001262, + "outer_iterations": 160, + "inner_iterations": 40, + "total_iterations": 200, + "n_passive": 1864, + "warm_start_errors": 184, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05, + "reconstruction_s": 1.8632027999992715, + "eval_s": 2.329871099999764, + "log_likelihood": 7115.742449288057, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.1487550000001647, + "n": 2060, + "solve_s": 1.1487550000001647, + "outer_iterations": 85, + "inner_iterations": 11, + "total_iterations": 96, + "n_passive": 1879, + "warm_start_errors": 103, + "index": 13 + }, + { + "warm_start_error_fraction": 0.07475728155339806, + "reconstruction_s": 1.780868200000441, + "eval_s": 2.089549299999817, + "log_likelihood": 7157.754505112758, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0453263999997944, + "n": 2060, + "solve_s": 1.0453263999997944, + "outer_iterations": 111, + "inner_iterations": 25, + "total_iterations": 136, + "n_passive": 1895, + "warm_start_errors": 154, + "index": 14 + }, + { + "warm_start_error_fraction": 0.050970873786407765, + "reconstruction_s": 1.4270587999999407, + "eval_s": 1.752633699999933, + "log_likelihood": 7130.537395088702, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8050852000005762, + "n": 2060, + "solve_s": 0.8050852000005762, + "outer_iterations": 66, + "inner_iterations": 15, + "total_iterations": 81, + "n_passive": 1894, + "warm_start_errors": 105, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06553398058252427, + "reconstruction_s": 1.7438297999997303, + "eval_s": 2.0466319999995903, + "log_likelihood": 7038.376811999514, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.056206999999631, + "n": 2060, + "solve_s": 1.056206999999631, + "outer_iterations": 101, + "inner_iterations": 19, + "total_iterations": 120, + "n_passive": 1889, + "warm_start_errors": 135, + "index": 16 + }, + { + "warm_start_error_fraction": 0.043689320388349516, + "reconstruction_s": 1.497576200000367, + "eval_s": 1.8126573000008648, + "log_likelihood": 7398.450536271069, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.9175348999997368, + "n": 2060, + "solve_s": 0.9175348999997368, + "outer_iterations": 80, + "inner_iterations": 20, + "total_iterations": 100, + "n_passive": 1905, + "warm_start_errors": 90, + "index": 17 + }, + { + "warm_start_error_fraction": 0.05631067961165048, + "reconstruction_s": 1.9260738000002675, + "eval_s": 2.289349399999992, + "log_likelihood": 6543.698875010002, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2387567999994644, + "n": 2060, + "solve_s": 1.2387567999994644, + "outer_iterations": 112, + "inner_iterations": 19, + "total_iterations": 131, + "n_passive": 1883, + "warm_start_errors": 116, + "index": 18 + }, + { + "warm_start_error_fraction": 0.05679611650485437, + "reconstruction_s": 1.6291307999999844, + "eval_s": 1.9712521999999808, + "log_likelihood": 6154.717062854232, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7865054000003511, + "n": 2060, + "solve_s": 0.7865054000003511, + "outer_iterations": 64, + "inner_iterations": 16, + "total_iterations": 80, + "n_passive": 1862, + "warm_start_errors": 117, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_2000_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_2000_v2026.8.17.1.png new file mode 100644 index 000000000..048a51796 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_2000_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_600_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_600_v2026.8.17.1.json new file mode 100644 index 000000000..f0c1e8e55 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_600_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "mesh_600", + "configuration": { + "model": "mesh_600", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 600, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_600", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 2.2148489999999583, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.06515151515151515, + "mean_warm_start_error_fraction": 0.0640151515151515, + "p90_warm_start_error_fraction": 0.0696969696969697, + "max_warm_start_error_fraction": 0.07121212121212121, + "median_outer_iterations": 30.0, + "mean_outer_iterations": 30.25, + "p90_outer_iterations": 34.0, + "max_outer_iterations": 37.0, + "median_inner_iterations": 9.0, + "mean_inner_iterations": 9.0, + "p90_inner_iterations": 10.200000000000003, + "max_inner_iterations": 13.0, + "median_total_iterations": 39.0, + "mean_total_iterations": 39.25, + "p90_total_iterations": 45.1, + "max_total_iterations": 47.0, + "median_n_passive": 589.5, + "mean_n_passive": 589.8, + "p90_n_passive": 593.1, + "max_n_passive": 595.0, + "median_warm_start_errors": 43.0, + "mean_warm_start_errors": 42.25, + "p90_warm_start_errors": 46.0, + "max_warm_start_errors": 47.0, + "median_solve_s": 0.06440835000012157, + "mean_solve_s": 0.06418039500017585, + "p90_solve_s": 0.08028825000064899, + "max_solve_s": 0.087296000000606, + "median_reconstruction_s": 0.49278934999983903, + "mean_reconstruction_s": 0.5118642899999031, + "p90_reconstruction_s": 0.597519299999749, + "max_reconstruction_s": 0.6461193000004641, + "median_eval_s": 0.5044597999999496, + "mean_eval_s": 0.5249521449999974, + "p90_eval_s": 0.6124880199993641, + "max_eval_s": 0.6613743000007162 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.012878787878787878, + "mean_warm_start_error_fraction": 0.014848484848484852, + "p90_warm_start_error_fraction": 0.021515151515151522, + "max_warm_start_error_fraction": 0.06515151515151515, + "median_outer_iterations": 4.5, + "mean_outer_iterations": 6.25, + "p90_outer_iterations": 10.200000000000003, + "max_outer_iterations": 31.0, + "median_inner_iterations": 2.0, + "mean_inner_iterations": 2.1, + "p90_inner_iterations": 3.0, + "max_inner_iterations": 7.0, + "median_total_iterations": 6.0, + "mean_total_iterations": 8.35, + "p90_total_iterations": 12.200000000000003, + "max_total_iterations": 38.0, + "median_n_passive": 589.0, + "mean_n_passive": 589.6, + "p90_n_passive": 593.1, + "max_n_passive": 595.0, + "median_warm_start_errors": 8.5, + "mean_warm_start_errors": 9.8, + "p90_warm_start_errors": 14.200000000000003, + "max_warm_start_errors": 43.0, + "median_solve_s": 0.01012300000002142, + "mean_solve_s": 0.01410101499996017, + "p90_solve_s": 0.017112949999864215, + "max_solve_s": 0.07364749999942433, + "median_reconstruction_s": 0.41513274999988425, + "mean_reconstruction_s": 0.419243945000062, + "p90_reconstruction_s": 0.4431822399998055, + "max_reconstruction_s": 0.46700220000002446, + "median_eval_s": 0.42616585000041596, + "mean_eval_s": 0.4310418499999287, + "p90_eval_s": 0.45531525999958833, + "max_eval_s": 0.47896639999999024 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 7.558480606267272e-16, + "max_abs_reconstruction_deviation": 2.5219832888723204e-09 + }, + "iteration_reduction_ratio": 6.5, + "solve_speedup_ratio": 6.362575323519241, + "eval_speedup_ratio": 1.1837170904225602, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.06515151515151515, + "reconstruction_s": 0.5207119999995484, + "eval_s": 0.5348479000003863, + "log_likelihood": 7265.6964106687765, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07585680000011052, + "n": 660, + "solve_s": 0.07585680000011052, + "outer_iterations": 31, + "inner_iterations": 7, + "total_iterations": 38, + "n_passive": 592, + "warm_start_errors": 43, + "index": 0 + }, + { + "warm_start_error_fraction": 0.06363636363636363, + "reconstruction_s": 0.5352300999993531, + "eval_s": 0.5499823000000106, + "log_likelihood": 7237.652968585449, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05749699999978475, + "n": 660, + "solve_s": 0.05749699999978475, + "outer_iterations": 29, + "inner_iterations": 7, + "total_iterations": 36, + "n_passive": 590, + "warm_start_errors": 42, + "index": 1 + }, + { + "warm_start_error_fraction": 0.06363636363636363, + "reconstruction_s": 0.5941138999996838, + "eval_s": 0.607898600000226, + "log_likelihood": 7199.299776692247, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08441880000009405, + "n": 660, + "solve_s": 0.08441880000009405, + "outer_iterations": 27, + "inner_iterations": 7, + "total_iterations": 34, + "n_passive": 590, + "warm_start_errors": 42, + "index": 2 + }, + { + "warm_start_error_fraction": 0.06515151515151515, + "reconstruction_s": 0.6461193000004641, + "eval_s": 0.6613743000007162, + "log_likelihood": 7239.243398099665, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06842640000013489, + "n": 660, + "solve_s": 0.06842640000013489, + "outer_iterations": 30, + "inner_iterations": 8, + "total_iterations": 38, + "n_passive": 589, + "warm_start_errors": 43, + "index": 3 + }, + { + "warm_start_error_fraction": 0.06515151515151515, + "reconstruction_s": 0.5926082999994833, + "eval_s": 0.6092654999993101, + "log_likelihood": 7243.408648588999, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06940640000084386, + "n": 660, + "solve_s": 0.06940640000084386, + "outer_iterations": 30, + "inner_iterations": 10, + "total_iterations": 40, + "n_passive": 587, + "warm_start_errors": 43, + "index": 4 + }, + { + "warm_start_error_fraction": 0.06818181818181818, + "reconstruction_s": 0.5277624999998807, + "eval_s": 0.5389186000002155, + "log_likelihood": 7201.575670528215, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05359050000060961, + "n": 660, + "solve_s": 0.05359050000060961, + "outer_iterations": 31, + "inner_iterations": 10, + "total_iterations": 41, + "n_passive": 588, + "warm_start_errors": 45, + "index": 5 + }, + { + "warm_start_error_fraction": 0.06666666666666667, + "reconstruction_s": 0.48115839999991294, + "eval_s": 0.49561240000002726, + "log_likelihood": 7261.04934721416, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07803859999967244, + "n": 660, + "solve_s": 0.07803859999967244, + "outer_iterations": 31, + "inner_iterations": 9, + "total_iterations": 40, + "n_passive": 588, + "warm_start_errors": 44, + "index": 6 + }, + { + "warm_start_error_fraction": 0.06212121212121212, + "reconstruction_s": 0.4823711999997613, + "eval_s": 0.4970852999995259, + "log_likelihood": 7282.179088403567, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.062388800000007905, + "n": 660, + "solve_s": 0.062388800000007905, + "outer_iterations": 25, + "inner_iterations": 7, + "total_iterations": 32, + "n_passive": 592, + "warm_start_errors": 41, + "index": 7 + }, + { + "warm_start_error_fraction": 0.05606060606060606, + "reconstruction_s": 0.5265857999993386, + "eval_s": 0.540237199999865, + "log_likelihood": 7305.210393470316, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0476040000003195, + "n": 660, + "solve_s": 0.0476040000003195, + "outer_iterations": 28, + "inner_iterations": 7, + "total_iterations": 35, + "n_passive": 594, + "warm_start_errors": 37, + "index": 8 + }, + { + "warm_start_error_fraction": 0.06818181818181818, + "reconstruction_s": 0.4709680000005392, + "eval_s": 0.4815223999994487, + "log_likelihood": 7219.662912295906, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.052179500000420376, + "n": 660, + "solve_s": 0.052179500000420376, + "outer_iterations": 33, + "inner_iterations": 12, + "total_iterations": 45, + "n_passive": 588, + "warm_start_errors": 45, + "index": 9 + }, + { + "warm_start_error_fraction": 0.07121212121212121, + "reconstruction_s": 0.43925599999965925, + "eval_s": 0.45218860000022687, + "log_likelihood": 7205.3106103905875, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.053885100000115926, + "n": 660, + "solve_s": 0.053885100000115926, + "outer_iterations": 34, + "inner_iterations": 10, + "total_iterations": 44, + "n_passive": 588, + "warm_start_errors": 47, + "index": 10 + }, + { + "warm_start_error_fraction": 0.0696969696969697, + "reconstruction_s": 0.45008489999963786, + "eval_s": 0.4613196999998763, + "log_likelihood": 7190.413903319745, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04838000000017928, + "n": 660, + "solve_s": 0.04838000000017928, + "outer_iterations": 29, + "inner_iterations": 10, + "total_iterations": 39, + "n_passive": 587, + "warm_start_errors": 46, + "index": 11 + }, + { + "warm_start_error_fraction": 0.06666666666666667, + "reconstruction_s": 0.41606080000019574, + "eval_s": 0.42690500000026077, + "log_likelihood": 7179.281256896479, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05221309999978985, + "n": 660, + "solve_s": 0.05221309999978985, + "outer_iterations": 31, + "inner_iterations": 10, + "total_iterations": 41, + "n_passive": 589, + "warm_start_errors": 44, + "index": 12 + }, + { + "warm_start_error_fraction": 0.06363636363636363, + "reconstruction_s": 0.5223002000002452, + "eval_s": 0.5332364000005327, + "log_likelihood": 7195.408164889935, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049297800000204006, + "n": 660, + "solve_s": 0.049297800000204006, + "outer_iterations": 26, + "inner_iterations": 9, + "total_iterations": 35, + "n_passive": 588, + "warm_start_errors": 42, + "index": 13 + }, + { + "warm_start_error_fraction": 0.0696969696969697, + "reconstruction_s": 0.4428121999999348, + "eval_s": 0.455439600000318, + "log_likelihood": 7168.373554502044, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.056123500000467175, + "n": 660, + "solve_s": 0.056123500000467175, + "outer_iterations": 29, + "inner_iterations": 10, + "total_iterations": 39, + "n_passive": 587, + "warm_start_errors": 46, + "index": 14 + }, + { + "warm_start_error_fraction": 0.06515151515151515, + "reconstruction_s": 0.48762320000059844, + "eval_s": 0.5004364999995232, + "log_likelihood": 7182.680123208589, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0665486999996574, + "n": 660, + "solve_s": 0.0665486999996574, + "outer_iterations": 34, + "inner_iterations": 13, + "total_iterations": 47, + "n_passive": 590, + "warm_start_errors": 43, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06212121212121212, + "reconstruction_s": 0.4910504999998011, + "eval_s": 0.5025021000001288, + "log_likelihood": 7236.3458157913155, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07419969999955356, + "n": 660, + "solve_s": 0.07419969999955356, + "outer_iterations": 30, + "inner_iterations": 9, + "total_iterations": 39, + "n_passive": 590, + "warm_start_errors": 41, + "index": 16 + }, + { + "warm_start_error_fraction": 0.05454545454545454, + "reconstruction_s": 0.49452819999987696, + "eval_s": 0.5064174999997704, + "log_likelihood": 7283.328278303767, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07982930000071065, + "n": 660, + "solve_s": 0.07982930000071065, + "outer_iterations": 29, + "inner_iterations": 8, + "total_iterations": 37, + "n_passive": 591, + "warm_start_errors": 36, + "index": 17 + }, + { + "warm_start_error_fraction": 0.05454545454545454, + "reconstruction_s": 0.6281679000003351, + "eval_s": 0.6414906999998493, + "log_likelihood": 7310.159266039755, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.087296000000606, + "n": 660, + "solve_s": 0.087296000000606, + "outer_iterations": 31, + "inner_iterations": 8, + "total_iterations": 39, + "n_passive": 593, + "warm_start_errors": 36, + "index": 18 + }, + { + "warm_start_error_fraction": 0.05909090909090909, + "reconstruction_s": 0.4877723999998125, + "eval_s": 0.5023622999997315, + "log_likelihood": 7416.701460180124, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06642790000023524, + "n": 660, + "solve_s": 0.06642790000023524, + "outer_iterations": 37, + "inner_iterations": 9, + "total_iterations": 46, + "n_passive": 595, + "warm_start_errors": 39, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.06515151515151515, + "reconstruction_s": 0.46700220000002446, + "eval_s": 0.47896639999999024, + "log_likelihood": 7265.6964106687765, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07364749999942433, + "n": 660, + "solve_s": 0.07364749999942433, + "outer_iterations": 31, + "inner_iterations": 7, + "total_iterations": 38, + "n_passive": 592, + "warm_start_errors": 43, + "index": 0 + }, + { + "warm_start_error_fraction": 0.015151515151515152, + "reconstruction_s": 0.42951739999989513, + "eval_s": 0.44289589999971213, + "log_likelihood": 7237.652968585453, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.015041000000564964, + "n": 660, + "solve_s": 0.015041000000564964, + "outer_iterations": 8, + "inner_iterations": 2, + "total_iterations": 10, + "n_passive": 590, + "warm_start_errors": 10, + "index": 1 + }, + { + "warm_start_error_fraction": 0.004545454545454545, + "reconstruction_s": 0.4155700999999681, + "eval_s": 0.42613300000084564, + "log_likelihood": 7199.299776692245, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0068617999995694845, + "n": 660, + "solve_s": 0.0068617999995694845, + "outer_iterations": 1, + "inner_iterations": 2, + "total_iterations": 3, + "n_passive": 589, + "warm_start_errors": 3, + "index": 2 + }, + { + "warm_start_error_fraction": 0.015151515151515152, + "reconstruction_s": 0.4125523000002431, + "eval_s": 0.42389749999983906, + "log_likelihood": 7239.24339809967, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.009816700000556011, + "n": 660, + "solve_s": 0.009816700000556011, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 589, + "warm_start_errors": 10, + "index": 3 + }, + { + "warm_start_error_fraction": 0.006060606060606061, + "reconstruction_s": 0.3873444999999265, + "eval_s": 0.3995524000001751, + "log_likelihood": 7243.408648588998, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.006691699999464618, + "n": 660, + "solve_s": 0.006691699999464618, + "outer_iterations": 1, + "inner_iterations": 2, + "total_iterations": 3, + "n_passive": 587, + "warm_start_errors": 4, + "index": 4 + }, + { + "warm_start_error_fraction": 0.013636363636363636, + "reconstruction_s": 0.39541230000031646, + "eval_s": 0.410899800000152, + "log_likelihood": 7201.575670528214, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.01853359999950044, + "n": 660, + "solve_s": 0.01853359999950044, + "outer_iterations": 7, + "inner_iterations": 3, + "total_iterations": 10, + "n_passive": 588, + "warm_start_errors": 9, + "index": 5 + }, + { + "warm_start_error_fraction": 0.00909090909090909, + "reconstruction_s": 0.4387832000002163, + "eval_s": 0.449797899999794, + "log_likelihood": 7261.0493472141625, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.00851520000014716, + "n": 660, + "solve_s": 0.00851520000014716, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 588, + "warm_start_errors": 6, + "index": 6 + }, + { + "warm_start_error_fraction": 0.01818181818181818, + "reconstruction_s": 0.41459109999959765, + "eval_s": 0.42589309999948455, + "log_likelihood": 7282.179088403567, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.012873699999545352, + "n": 660, + "solve_s": 0.012873699999545352, + "outer_iterations": 9, + "inner_iterations": 2, + "total_iterations": 11, + "n_passive": 592, + "warm_start_errors": 12, + "index": 7 + }, + { + "warm_start_error_fraction": 0.01818181818181818, + "reconstruction_s": 0.3931308999999601, + "eval_s": 0.40567250000003696, + "log_likelihood": 7305.210393470318, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.011500399999931687, + "n": 660, + "solve_s": 0.011500399999931687, + "outer_iterations": 7, + "inner_iterations": 2, + "total_iterations": 9, + "n_passive": 594, + "warm_start_errors": 12, + "index": 8 + }, + { + "warm_start_error_fraction": 0.012121212121212121, + "reconstruction_s": 0.4217746000003899, + "eval_s": 0.43353599999954895, + "log_likelihood": 7219.662912295911, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.00773990000016056, + "n": 660, + "solve_s": 0.00773990000016056, + "outer_iterations": 2, + "inner_iterations": 3, + "total_iterations": 5, + "n_passive": 588, + "warm_start_errors": 8, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0030303030303030303, + "reconstruction_s": 0.4272829000001366, + "eval_s": 0.43866109999999026, + "log_likelihood": 7205.310610390588, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.007892399999946065, + "n": 660, + "solve_s": 0.007892399999946065, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 588, + "warm_start_errors": 2, + "index": 10 + }, + { + "warm_start_error_fraction": 0.016666666666666666, + "reconstruction_s": 0.44254449999971257, + "eval_s": 0.45465559999956895, + "log_likelihood": 7190.413903319746, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.012377100000776409, + "n": 660, + "solve_s": 0.012377100000776409, + "outer_iterations": 5, + "inner_iterations": 2, + "total_iterations": 7, + "n_passive": 587, + "warm_start_errors": 11, + "index": 11 + }, + { + "warm_start_error_fraction": 0.006060606060606061, + "reconstruction_s": 0.41201619999992545, + "eval_s": 0.42270099999950617, + "log_likelihood": 7179.281256896477, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.010198500000115018, + "n": 660, + "solve_s": 0.010198500000115018, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 589, + "warm_start_errors": 4, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0030303030303030303, + "reconstruction_s": 0.41469539999980043, + "eval_s": 0.4261986999999863, + "log_likelihood": 7195.4081648899355, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.006011499999658554, + "n": 660, + "solve_s": 0.006011499999658554, + "outer_iterations": 0, + "inner_iterations": 1, + "total_iterations": 1, + "n_passive": 587, + "warm_start_errors": 2, + "index": 13 + }, + { + "warm_start_error_fraction": 0.010606060606060607, + "reconstruction_s": 0.4018236000001707, + "eval_s": 0.41346989999965444, + "log_likelihood": 7168.373554502043, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.009616100000130245, + "n": 660, + "solve_s": 0.009616100000130245, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 586, + "warm_start_errors": 7, + "index": 14 + }, + { + "warm_start_error_fraction": 0.015151515151515152, + "reconstruction_s": 0.40869430000020657, + "eval_s": 0.42090230000030715, + "log_likelihood": 7182.680123208591, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.016955099999904633, + "n": 660, + "solve_s": 0.016955099999904633, + "outer_iterations": 9, + "inner_iterations": 3, + "total_iterations": 12, + "n_passive": 590, + "warm_start_errors": 10, + "index": 15 + }, + { + "warm_start_error_fraction": 0.010606060606060607, + "reconstruction_s": 0.43739000000005035, + "eval_s": 0.4484087000000727, + "log_likelihood": 7236.345815791315, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.010047499999927823, + "n": 660, + "solve_s": 0.010047499999927823, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 589, + "warm_start_errors": 7, + "index": 16 + }, + { + "warm_start_error_fraction": 0.024242424242424242, + "reconstruction_s": 0.42431930000020657, + "eval_s": 0.43472139999994397, + "log_likelihood": 7283.328278303767, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.013039599999501661, + "n": 660, + "solve_s": 0.013039599999501661, + "outer_iterations": 12, + "inner_iterations": 2, + "total_iterations": 14, + "n_passive": 591, + "warm_start_errors": 16, + "index": 17 + }, + { + "warm_start_error_fraction": 0.00909090909090909, + "reconstruction_s": 0.44892190000064147, + "eval_s": 0.4612521999997625, + "log_likelihood": 7310.159266039751, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.009684200000265264, + "n": 660, + "solve_s": 0.009684200000265264, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 593, + "warm_start_errors": 6, + "index": 18 + }, + { + "warm_start_error_fraction": 0.021212121212121213, + "reconstruction_s": 0.3915121999998519, + "eval_s": 0.4026216000002023, + "log_likelihood": 7416.70146018012, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.014976800000113144, + "n": 660, + "solve_s": 0.014976800000113144, + "outer_iterations": 10, + "inner_iterations": 2, + "total_iterations": 12, + "n_passive": 595, + "warm_start_errors": 14, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.08181818181818182, + "mean_warm_start_error_fraction": 0.0812121212121212, + "p90_warm_start_error_fraction": 0.10318181818181818, + "max_warm_start_error_fraction": 0.11060606060606061, + "median_outer_iterations": 36.5, + "mean_outer_iterations": 37.2, + "p90_outer_iterations": 46.2, + "max_outer_iterations": 49.0, + "median_inner_iterations": 11.0, + "mean_inner_iterations": 11.25, + "p90_inner_iterations": 13.100000000000001, + "max_inner_iterations": 20.0, + "median_total_iterations": 49.0, + "mean_total_iterations": 48.45, + "p90_total_iterations": 59.2, + "max_total_iterations": 66.0, + "median_n_passive": 581.0, + "mean_n_passive": 580.55, + "p90_n_passive": 592.3, + "max_n_passive": 598.0, + "median_warm_start_errors": 54.0, + "mean_warm_start_errors": 53.6, + "p90_warm_start_errors": 68.1, + "max_warm_start_errors": 73.0, + "median_solve_s": 0.06522714999982782, + "mean_solve_s": 0.06457914499987964, + "p90_solve_s": 0.08278718000010486, + "max_solve_s": 0.08821009999974194, + "median_reconstruction_s": 0.48634990000027756, + "mean_reconstruction_s": 0.49559246000003443, + "p90_reconstruction_s": 0.5849953599999026, + "max_reconstruction_s": 0.6338388000003761, + "median_eval_s": 0.4989510000000337, + "mean_eval_s": 0.5077538149999782, + "p90_eval_s": 0.5961970099997416, + "max_eval_s": 0.6447048999998515 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.08181818181818182, + "mean_warm_start_error_fraction": 0.08325757575757577, + "p90_warm_start_error_fraction": 0.10621212121212122, + "max_warm_start_error_fraction": 0.11212121212121212, + "median_outer_iterations": 45.5, + "mean_outer_iterations": 43.25, + "p90_outer_iterations": 56.60000000000001, + "max_outer_iterations": 62.0, + "median_inner_iterations": 10.0, + "mean_inner_iterations": 10.75, + "p90_inner_iterations": 16.1, + "max_inner_iterations": 20.0, + "median_total_iterations": 56.0, + "mean_total_iterations": 54.0, + "p90_total_iterations": 66.9, + "max_total_iterations": 79.0, + "median_n_passive": 581.0, + "mean_n_passive": 580.3, + "p90_n_passive": 592.2, + "max_n_passive": 598.0, + "median_warm_start_errors": 54.0, + "mean_warm_start_errors": 54.95, + "p90_warm_start_errors": 70.1, + "max_warm_start_errors": 74.0, + "median_solve_s": 0.05042935000028592, + "mean_solve_s": 0.06092440500010525, + "p90_solve_s": 0.09228613000059345, + "max_solve_s": 0.12166490000072372, + "median_reconstruction_s": 0.4997601999998551, + "mean_reconstruction_s": 0.5380541150000226, + "p90_reconstruction_s": 0.6804505799998879, + "max_reconstruction_s": 0.8680718999994497, + "median_eval_s": 0.51178525000023, + "mean_eval_s": 0.5532585100000688, + "p90_eval_s": 0.6966636699998162, + "max_eval_s": 0.8906116999996812 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 5.332344637049255e-15, + "max_abs_reconstruction_deviation": 4.6302295331202004e-10 + }, + "iteration_reduction_ratio": 0.875, + "solve_speedup_ratio": 1.293436262800492, + "eval_speedup_ratio": 0.9749225871589879, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.10303030303030303, + "reconstruction_s": 0.5844655999999304, + "eval_s": 0.5954616999997597, + "log_likelihood": 5522.657640132191, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08224830000017391, + "n": 660, + "solve_s": 0.08224830000017391, + "outer_iterations": 46, + "inner_iterations": 20, + "total_iterations": 66, + "n_passive": 561, + "warm_start_errors": 68, + "index": 0 + }, + { + "warm_start_error_fraction": 0.08787878787878788, + "reconstruction_s": 0.5125836999995954, + "eval_s": 0.5306845999994039, + "log_likelihood": 6693.665173051272, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07010360000003857, + "n": 660, + "solve_s": 0.07010360000003857, + "outer_iterations": 43, + "inner_iterations": 11, + "total_iterations": 54, + "n_passive": 579, + "warm_start_errors": 58, + "index": 1 + }, + { + "warm_start_error_fraction": 0.07878787878787878, + "reconstruction_s": 0.46683459999985644, + "eval_s": 0.47742440000001807, + "log_likelihood": 7298.00986515643, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06178170000021055, + "n": 660, + "solve_s": 0.06178170000021055, + "outer_iterations": 43, + "inner_iterations": 13, + "total_iterations": 56, + "n_passive": 598, + "warm_start_errors": 52, + "index": 2 + }, + { + "warm_start_error_fraction": 0.08636363636363636, + "reconstruction_s": 0.4505986000003759, + "eval_s": 0.46130639999955747, + "log_likelihood": 6665.153403438588, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05512589999943884, + "n": 660, + "solve_s": 0.05512589999943884, + "outer_iterations": 37, + "inner_iterations": 10, + "total_iterations": 47, + "n_passive": 573, + "warm_start_errors": 57, + "index": 3 + }, + { + "warm_start_error_fraction": 0.07424242424242425, + "reconstruction_s": 0.5061992000000828, + "eval_s": 0.5178231999998388, + "log_likelihood": 7139.227697585948, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07688659999985248, + "n": 660, + "solve_s": 0.07688659999985248, + "outer_iterations": 36, + "inner_iterations": 8, + "total_iterations": 44, + "n_passive": 583, + "warm_start_errors": 49, + "index": 4 + }, + { + "warm_start_error_fraction": 0.07121212121212121, + "reconstruction_s": 0.49500220000027184, + "eval_s": 0.5071449999995821, + "log_likelihood": 7087.36199677991, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07181250000030559, + "n": 660, + "solve_s": 0.07181250000030559, + "outer_iterations": 33, + "inner_iterations": 11, + "total_iterations": 44, + "n_passive": 586, + "warm_start_errors": 47, + "index": 5 + }, + { + "warm_start_error_fraction": 0.09242424242424242, + "reconstruction_s": 0.589763199999652, + "eval_s": 0.6028147999995781, + "log_likelihood": 6898.332463282088, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0730068000002575, + "n": 660, + "solve_s": 0.0730068000002575, + "outer_iterations": 43, + "inner_iterations": 11, + "total_iterations": 54, + "n_passive": 577, + "warm_start_errors": 61, + "index": 6 + }, + { + "warm_start_error_fraction": 0.08181818181818182, + "reconstruction_s": 0.5479589999995369, + "eval_s": 0.5610055999995893, + "log_likelihood": 6557.512530996008, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05457099999966886, + "n": 660, + "solve_s": 0.05457099999966886, + "outer_iterations": 29, + "inner_iterations": 10, + "total_iterations": 39, + "n_passive": 578, + "warm_start_errors": 54, + "index": 7 + }, + { + "warm_start_error_fraction": 0.08636363636363636, + "reconstruction_s": 0.4436077000000296, + "eval_s": 0.4550085000000763, + "log_likelihood": 6651.91314205308, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05727350000051956, + "n": 660, + "solve_s": 0.05727350000051956, + "outer_iterations": 37, + "inner_iterations": 13, + "total_iterations": 50, + "n_passive": 584, + "warm_start_errors": 57, + "index": 8 + }, + { + "warm_start_error_fraction": 0.11060606060606061, + "reconstruction_s": 0.528242099999261, + "eval_s": 0.5412562999999864, + "log_likelihood": 5194.705357040443, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07067860000006476, + "n": 660, + "solve_s": 0.07067860000006476, + "outer_iterations": 48, + "inner_iterations": 11, + "total_iterations": 59, + "n_passive": 564, + "warm_start_errors": 73, + "index": 9 + }, + { + "warm_start_error_fraction": 0.08636363636363636, + "reconstruction_s": 0.5547165000007226, + "eval_s": 0.5650072999997064, + "log_likelihood": 6957.3135522330085, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08763709999948333, + "n": 660, + "solve_s": 0.08763709999948333, + "outer_iterations": 42, + "inner_iterations": 10, + "total_iterations": 52, + "n_passive": 588, + "warm_start_errors": 57, + "index": 10 + }, + { + "warm_start_error_fraction": 0.10454545454545454, + "reconstruction_s": 0.44007650000003196, + "eval_s": 0.45161540000026434, + "log_likelihood": 5711.186998203406, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0692191999996794, + "n": 660, + "solve_s": 0.0692191999996794, + "outer_iterations": 49, + "inner_iterations": 12, + "total_iterations": 61, + "n_passive": 569, + "warm_start_errors": 69, + "index": 11 + }, + { + "warm_start_error_fraction": 0.07727272727272727, + "reconstruction_s": 0.6338388000003761, + "eval_s": 0.6447048999998515, + "log_likelihood": 6716.925573448715, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08821009999974194, + "n": 660, + "solve_s": 0.08821009999974194, + "outer_iterations": 41, + "inner_iterations": 13, + "total_iterations": 54, + "n_passive": 571, + "warm_start_errors": 51, + "index": 12 + }, + { + "warm_start_error_fraction": 0.08181818181818182, + "reconstruction_s": 0.4490310999999565, + "eval_s": 0.4610309000008783, + "log_likelihood": 7214.782106084835, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06867259999944508, + "n": 660, + "solve_s": 0.06867259999944508, + "outer_iterations": 33, + "inner_iterations": 9, + "total_iterations": 42, + "n_passive": 579, + "warm_start_errors": 54, + "index": 13 + }, + { + "warm_start_error_fraction": 0.05757575757575758, + "reconstruction_s": 0.4776976000002833, + "eval_s": 0.4907570000004853, + "log_likelihood": 7267.3447433642, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05196449999948527, + "n": 660, + "solve_s": 0.05196449999948527, + "outer_iterations": 34, + "inner_iterations": 9, + "total_iterations": 43, + "n_passive": 588, + "warm_start_errors": 38, + "index": 14 + }, + { + "warm_start_error_fraction": 0.06363636363636363, + "reconstruction_s": 0.41372430000046734, + "eval_s": 0.42502280000007886, + "log_likelihood": 7230.116245529631, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.050400200000694895, + "n": 660, + "solve_s": 0.050400200000694895, + "outer_iterations": 27, + "inner_iterations": 13, + "total_iterations": 40, + "n_passive": 584, + "warm_start_errors": 42, + "index": 15 + }, + { + "warm_start_error_fraction": 0.05454545454545454, + "reconstruction_s": 0.4627528999999413, + "eval_s": 0.47663420000026235, + "log_likelihood": 7142.736262071841, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0452897999994093, + "n": 660, + "solve_s": 0.0452897999994093, + "outer_iterations": 22, + "inner_iterations": 8, + "total_iterations": 30, + "n_passive": 595, + "warm_start_errors": 36, + "index": 16 + }, + { + "warm_start_error_fraction": 0.06212121212121212, + "reconstruction_s": 0.42935839999972814, + "eval_s": 0.4417797000005521, + "log_likelihood": 7513.346011977217, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04463549999945826, + "n": 660, + "solve_s": 0.04463549999945826, + "outer_iterations": 30, + "inner_iterations": 6, + "total_iterations": 36, + "n_passive": 592, + "warm_start_errors": 41, + "index": 17 + }, + { + "warm_start_error_fraction": 0.07575757575757576, + "reconstruction_s": 0.42409240000051796, + "eval_s": 0.4365250000000742, + "log_likelihood": 6635.892091532686, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.054043999999521475, + "n": 660, + "solve_s": 0.054043999999521475, + "outer_iterations": 35, + "inner_iterations": 14, + "total_iterations": 49, + "n_passive": 592, + "warm_start_errors": 50, + "index": 18 + }, + { + "warm_start_error_fraction": 0.08787878787878788, + "reconstruction_s": 0.5013048000000708, + "eval_s": 0.5120686000000205, + "log_likelihood": 6223.131095643476, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05802140000014333, + "n": 660, + "solve_s": 0.05802140000014333, + "outer_iterations": 36, + "inner_iterations": 13, + "total_iterations": 49, + "n_passive": 570, + "warm_start_errors": 58, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.10303030303030303, + "reconstruction_s": 0.5646908999997322, + "eval_s": 0.5754691000001912, + "log_likelihood": 5522.657640132191, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0764657999998235, + "n": 660, + "solve_s": 0.0764657999998235, + "outer_iterations": 46, + "inner_iterations": 20, + "total_iterations": 66, + "n_passive": 561, + "warm_start_errors": 68, + "index": 0 + }, + { + "warm_start_error_fraction": 0.07272727272727272, + "reconstruction_s": 0.40145229999961884, + "eval_s": 0.41225159999976313, + "log_likelihood": 6693.665173051271, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03947580000021844, + "n": 660, + "solve_s": 0.03947580000021844, + "outer_iterations": 43, + "inner_iterations": 6, + "total_iterations": 49, + "n_passive": 579, + "warm_start_errors": 48, + "index": 1 + }, + { + "warm_start_error_fraction": 0.10757575757575757, + "reconstruction_s": 0.4484535000001415, + "eval_s": 0.4609606000003623, + "log_likelihood": 7298.009865156454, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07215850000011415, + "n": 660, + "solve_s": 0.07215850000011415, + "outer_iterations": 62, + "inner_iterations": 13, + "total_iterations": 75, + "n_passive": 598, + "warm_start_errors": 71, + "index": 2 + }, + { + "warm_start_error_fraction": 0.10454545454545454, + "reconstruction_s": 0.5294276000004174, + "eval_s": 0.541524300000674, + "log_likelihood": 6665.15340343858, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05516119999992952, + "n": 660, + "solve_s": 0.05516119999992952, + "outer_iterations": 56, + "inner_iterations": 9, + "total_iterations": 65, + "n_passive": 573, + "warm_start_errors": 69, + "index": 3 + }, + { + "warm_start_error_fraction": 0.07575757575757576, + "reconstruction_s": 0.5639937000005375, + "eval_s": 0.5851959999999963, + "log_likelihood": 7139.227697585951, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0892176000006657, + "n": 660, + "solve_s": 0.0892176000006657, + "outer_iterations": 43, + "inner_iterations": 7, + "total_iterations": 50, + "n_passive": 583, + "warm_start_errors": 50, + "index": 4 + }, + { + "warm_start_error_fraction": 0.09242424242424242, + "reconstruction_s": 0.6883406999995714, + "eval_s": 0.7119471999994857, + "log_likelihood": 7087.361996779923, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11990289999994275, + "n": 660, + "solve_s": 0.11990289999994275, + "outer_iterations": 45, + "inner_iterations": 10, + "total_iterations": 55, + "n_passive": 586, + "warm_start_errors": 61, + "index": 5 + }, + { + "warm_start_error_fraction": 0.0803030303030303, + "reconstruction_s": 0.8680718999994497, + "eval_s": 0.8906116999996812, + "log_likelihood": 6898.332463282095, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07540760000028968, + "n": 660, + "solve_s": 0.07540760000028968, + "outer_iterations": 44, + "inner_iterations": 16, + "total_iterations": 60, + "n_passive": 577, + "warm_start_errors": 53, + "index": 6 + }, + { + "warm_start_error_fraction": 0.10606060606060606, + "reconstruction_s": 0.5884454000006372, + "eval_s": 0.6095083000000159, + "log_likelihood": 6557.512530996015, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07142779999958293, + "n": 660, + "solve_s": 0.07142779999958293, + "outer_iterations": 48, + "inner_iterations": 10, + "total_iterations": 58, + "n_passive": 577, + "warm_start_errors": 70, + "index": 7 + }, + { + "warm_start_error_fraction": 0.08333333333333333, + "reconstruction_s": 0.479551000000356, + "eval_s": 0.4914974000002985, + "log_likelihood": 6651.913142053045, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04569750000064232, + "n": 660, + "solve_s": 0.04569750000064232, + "outer_iterations": 48, + "inner_iterations": 9, + "total_iterations": 57, + "n_passive": 584, + "warm_start_errors": 55, + "index": 8 + }, + { + "warm_start_error_fraction": 0.08484848484848485, + "reconstruction_s": 0.4967538999999306, + "eval_s": 0.5097247000003335, + "log_likelihood": 5194.705357040445, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.045272100000147475, + "n": 660, + "solve_s": 0.045272100000147475, + "outer_iterations": 43, + "inner_iterations": 11, + "total_iterations": 54, + "n_passive": 564, + "warm_start_errors": 56, + "index": 9 + }, + { + "warm_start_error_fraction": 0.07424242424242425, + "reconstruction_s": 0.436123100000259, + "eval_s": 0.4473606000001382, + "log_likelihood": 6957.313552233003, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03657960000055027, + "n": 660, + "solve_s": 0.03657960000055027, + "outer_iterations": 47, + "inner_iterations": 8, + "total_iterations": 55, + "n_passive": 587, + "warm_start_errors": 49, + "index": 10 + }, + { + "warm_start_error_fraction": 0.07272727272727272, + "reconstruction_s": 0.48598839999976917, + "eval_s": 0.4973820000004707, + "log_likelihood": 5711.186998203406, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03476909999972122, + "n": 660, + "solve_s": 0.03476909999972122, + "outer_iterations": 31, + "inner_iterations": 12, + "total_iterations": 43, + "n_passive": 569, + "warm_start_errors": 48, + "index": 11 + }, + { + "warm_start_error_fraction": 0.11212121212121212, + "reconstruction_s": 0.5027664999997796, + "eval_s": 0.5138458000001265, + "log_likelihood": 6716.92557344873, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06140080000022863, + "n": 660, + "solve_s": 0.06140080000022863, + "outer_iterations": 62, + "inner_iterations": 17, + "total_iterations": 79, + "n_passive": 571, + "warm_start_errors": 74, + "index": 12 + }, + { + "warm_start_error_fraction": 0.051515151515151514, + "reconstruction_s": 0.39603950000037, + "eval_s": 0.40777650000018184, + "log_likelihood": 7214.782106084846, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.023814199999833363, + "n": 660, + "solve_s": 0.023814199999833363, + "outer_iterations": 25, + "inner_iterations": 6, + "total_iterations": 31, + "n_passive": 579, + "warm_start_errors": 34, + "index": 13 + }, + { + "warm_start_error_fraction": 0.09242424242424242, + "reconstruction_s": 0.43419229999926756, + "eval_s": 0.44626359999983833, + "log_likelihood": 7267.344743364202, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.036863399999674584, + "n": 660, + "solve_s": 0.036863399999674584, + "outer_iterations": 47, + "inner_iterations": 13, + "total_iterations": 60, + "n_passive": 588, + "warm_start_errors": 61, + "index": 14 + }, + { + "warm_start_error_fraction": 0.05, + "reconstruction_s": 0.45735180000065156, + "eval_s": 0.4727271000001565, + "log_likelihood": 7230.1162455296235, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03910790000009001, + "n": 660, + "solve_s": 0.03910790000009001, + "outer_iterations": 24, + "inner_iterations": 6, + "total_iterations": 30, + "n_passive": 583, + "warm_start_errors": 33, + "index": 15 + }, + { + "warm_start_error_fraction": 0.07727272727272727, + "reconstruction_s": 0.4735655000004044, + "eval_s": 0.4847669999999198, + "log_likelihood": 7142.736262071829, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04464000000007218, + "n": 660, + "solve_s": 0.04464000000007218, + "outer_iterations": 48, + "inner_iterations": 12, + "total_iterations": 60, + "n_passive": 594, + "warm_start_errors": 51, + "index": 16 + }, + { + "warm_start_error_fraction": 0.05757575757575758, + "reconstruction_s": 0.6027893999998923, + "eval_s": 0.6225545999996029, + "log_likelihood": 7513.346011977215, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08582429999933083, + "n": 660, + "solve_s": 0.08582429999933083, + "outer_iterations": 29, + "inner_iterations": 13, + "total_iterations": 42, + "n_passive": 592, + "warm_start_errors": 38, + "index": 17 + }, + { + "warm_start_error_fraction": 0.08636363636363636, + "reconstruction_s": 0.663510999999744, + "eval_s": 0.6888366000002861, + "log_likelihood": 6635.892091532675, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.12166490000072372, + "n": 660, + "solve_s": 0.12166490000072372, + "outer_iterations": 51, + "inner_iterations": 9, + "total_iterations": 60, + "n_passive": 591, + "warm_start_errors": 57, + "index": 18 + }, + { + "warm_start_error_fraction": 0.0803030303030303, + "reconstruction_s": 0.679573899999923, + "eval_s": 0.6949654999998529, + "log_likelihood": 6223.131095643479, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04363710000052379, + "n": 660, + "solve_s": 0.04363710000052379, + "outer_iterations": 23, + "inner_iterations": 8, + "total_iterations": 31, + "n_passive": 570, + "warm_start_errors": 53, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_600_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_600_v2026.8.17.1.png new file mode 100644 index 000000000..a70ac7000 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_mesh_600_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_edge_zeroing_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_edge_zeroing_v2026.8.17.1.json new file mode 100644 index 000000000..2074be637 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_edge_zeroing_v2026.8.17.1.json @@ -0,0 +1,1602 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "no_edge_zeroing", + "configuration": { + "model": "no_edge_zeroing", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true, + "use_edge_zeroed_pixels": false + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 2.6319791000005353, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.05648854961832061, + "mean_warm_start_error_fraction": 0.056832061068702276, + "p90_warm_start_error_fraction": 0.06038167938931298, + "max_warm_start_error_fraction": 0.06412213740458016, + "median_outer_iterations": 58.5, + "mean_outer_iterations": 63.75, + "p90_outer_iterations": 69.0, + "max_outer_iterations": 177.0, + "median_inner_iterations": 11.5, + "mean_inner_iterations": 12.5, + "p90_inner_iterations": 15.200000000000003, + "max_inner_iterations": 29.0, + "median_total_iterations": 70.0, + "mean_total_iterations": 76.25, + "p90_total_iterations": 82.2, + "max_total_iterations": 206.0, + "median_n_passive": 1189.0, + "mean_n_passive": 1190.75, + "p90_n_passive": 1197.3, + "max_n_passive": 1200.0, + "median_warm_start_errors": 74.0, + "mean_warm_start_errors": 74.45, + "p90_warm_start_errors": 79.1, + "max_warm_start_errors": 84.0, + "median_solve_s": 0.35877055000037217, + "mean_solve_s": 0.41856836500005556, + "p90_solve_s": 0.6572881199997938, + "max_solve_s": 0.8691275999999561, + "median_reconstruction_s": 0.9175662499997088, + "mean_reconstruction_s": 0.9855948999998873, + "p90_reconstruction_s": 1.2352166800003945, + "max_reconstruction_s": 1.4538301999991745, + "median_eval_s": 0.9925639500002035, + "mean_eval_s": 1.0608727150000505, + "p90_eval_s": 1.3188605600000303, + "max_eval_s": 1.5340866000005917 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.006870229007633588, + "mean_warm_start_error_fraction": 0.009160305343511452, + "p90_warm_start_error_fraction": 0.010763358778625956, + "max_warm_start_error_fraction": 0.05877862595419847, + "median_outer_iterations": 4.5, + "mean_outer_iterations": 9.6, + "p90_outer_iterations": 16.900000000000013, + "max_outer_iterations": 69.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 2.35, + "p90_inner_iterations": 3.1000000000000014, + "max_inner_iterations": 15.0, + "median_total_iterations": 7.0, + "mean_total_iterations": 11.95, + "p90_total_iterations": 19.100000000000016, + "max_total_iterations": 84.0, + "median_n_passive": 1189.0, + "mean_n_passive": 1190.5, + "p90_n_passive": 1196.4, + "max_n_passive": 1200.0, + "median_warm_start_errors": 9.0, + "mean_warm_start_errors": 12.0, + "p90_warm_start_errors": 14.100000000000001, + "max_warm_start_errors": 77.0, + "median_solve_s": 0.05726010000034876, + "mean_solve_s": 0.09001494499998444, + "p90_solve_s": 0.10307960999944057, + "max_solve_s": 0.5925379999998768, + "median_reconstruction_s": 0.558291950000239, + "mean_reconstruction_s": 0.5919487900000149, + "p90_reconstruction_s": 0.6758607099998699, + "max_reconstruction_s": 1.1522171999995408, + "median_eval_s": 0.6319744000002174, + "mean_eval_s": 0.6623967250000533, + "p90_eval_s": 0.7493358900001112, + "max_eval_s": 1.220811400000457 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.1137321561171996e-14, + "max_abs_reconstruction_deviation": 4.839854513688513e-09 + }, + "iteration_reduction_ratio": 10.0, + "solve_speedup_ratio": 6.265629120420449, + "eval_speedup_ratio": 1.5705761973900556, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 1.1319098999993003, + "eval_s": 1.195651199999702, + "log_likelihood": 7206.2076831241275, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.514708900000187, + "n": 1310, + "solve_s": 0.514708900000187, + "outer_iterations": 69, + "inner_iterations": 15, + "total_iterations": 84, + "n_passive": 1195, + "warm_start_errors": 77, + "index": 0 + }, + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 0.9484987000005276, + "eval_s": 1.0345753999999943, + "log_likelihood": 7180.432957367257, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4523418999997375, + "n": 1310, + "solve_s": 0.4523418999997375, + "outer_iterations": 61, + "inner_iterations": 11, + "total_iterations": 72, + "n_passive": 1187, + "warm_start_errors": 77, + "index": 1 + }, + { + "warm_start_error_fraction": 0.05801526717557252, + "reconstruction_s": 0.9911308000000645, + "eval_s": 1.0586673999996492, + "log_likelihood": 7143.994889035799, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4436433000000761, + "n": 1310, + "solve_s": 0.4436433000000761, + "outer_iterations": 59, + "inner_iterations": 12, + "total_iterations": 71, + "n_passive": 1188, + "warm_start_errors": 76, + "index": 2 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 0.9101522999999361, + "eval_s": 0.9761140999999043, + "log_likelihood": 7182.878722012542, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3597421000004033, + "n": 1310, + "solve_s": 0.3597421000004033, + "outer_iterations": 68, + "inner_iterations": 12, + "total_iterations": 80, + "n_passive": 1190, + "warm_start_errors": 79, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.7996285999997781, + "eval_s": 0.8609360000000379, + "log_likelihood": 7186.1758717889115, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29685389999940526, + "n": 1310, + "solve_s": 0.29685389999940526, + "outer_iterations": 56, + "inner_iterations": 9, + "total_iterations": 65, + "n_passive": 1189, + "warm_start_errors": 72, + "index": 4 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.8133096000001387, + "eval_s": 0.8756649999995716, + "log_likelihood": 7145.024669580045, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35299380000014935, + "n": 1310, + "solve_s": 0.35299380000014935, + "outer_iterations": 63, + "inner_iterations": 15, + "total_iterations": 78, + "n_passive": 1189, + "warm_start_errors": 75, + "index": 5 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.8527179000002434, + "eval_s": 0.9406723000001875, + "log_likelihood": 7202.126666486725, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3683567000007315, + "n": 1310, + "solve_s": 0.3683567000007315, + "outer_iterations": 58, + "inner_iterations": 11, + "total_iterations": 69, + "n_passive": 1191, + "warm_start_errors": 74, + "index": 6 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.8278140000002168, + "eval_s": 0.9047435000002224, + "log_likelihood": 7222.820219978847, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35779900000034104, + "n": 1310, + "solve_s": 0.35779900000034104, + "outer_iterations": 62, + "inner_iterations": 13, + "total_iterations": 75, + "n_passive": 1194, + "warm_start_errors": 73, + "index": 7 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.8939951000002111, + "eval_s": 0.9647522000004756, + "log_likelihood": 7245.156725065453, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33490029999938997, + "n": 1310, + "solve_s": 0.33490029999938997, + "outer_iterations": 49, + "inner_iterations": 17, + "total_iterations": 66, + "n_passive": 1186, + "warm_start_errors": 74, + "index": 8 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.8875738999995519, + "eval_s": 0.9645328000005975, + "log_likelihood": 7163.555514905283, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.292460900000151, + "n": 1310, + "solve_s": 0.292460900000151, + "outer_iterations": 43, + "inner_iterations": 12, + "total_iterations": 55, + "n_passive": 1186, + "warm_start_errors": 69, + "index": 9 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.8291306999999506, + "eval_s": 0.8924762000006012, + "log_likelihood": 7149.741003169525, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28444149999995716, + "n": 1310, + "solve_s": 0.28444149999995716, + "outer_iterations": 49, + "inner_iterations": 9, + "total_iterations": 58, + "n_passive": 1186, + "warm_start_errors": 74, + "index": 10 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.8582799999994677, + "eval_s": 0.9189962999998897, + "log_likelihood": 7135.510856830086, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29141819999949803, + "n": 1310, + "solve_s": 0.29141819999949803, + "outer_iterations": 52, + "inner_iterations": 8, + "total_iterations": 60, + "n_passive": 1187, + "warm_start_errors": 74, + "index": 11 + }, + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 0.8825905000003331, + "eval_s": 0.9545911999994132, + "log_likelihood": 7124.427599165532, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3414022000006298, + "n": 1310, + "solve_s": 0.3414022000006298, + "outer_iterations": 54, + "inner_iterations": 8, + "total_iterations": 62, + "n_passive": 1187, + "warm_start_errors": 77, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.9813709000000017, + "eval_s": 1.0625188000003618, + "log_likelihood": 7138.787138439643, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3388167000002795, + "n": 1310, + "solve_s": 0.3388167000002795, + "outer_iterations": 51, + "inner_iterations": 10, + "total_iterations": 61, + "n_passive": 1189, + "warm_start_errors": 69, + "index": 13 + }, + { + "warm_start_error_fraction": 0.06412213740458016, + "reconstruction_s": 1.347348399999646, + "eval_s": 1.4337127999997392, + "log_likelihood": 7112.5359410066085, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8691275999999561, + "n": 1310, + "solve_s": 0.8691275999999561, + "outer_iterations": 177, + "inner_iterations": 29, + "total_iterations": 206, + "n_passive": 1189, + "warm_start_errors": 84, + "index": 14 + }, + { + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 0.995086199999605, + "eval_s": 1.0958457999995517, + "log_likelihood": 7123.2058733607955, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4075690999998187, + "n": 1310, + "solve_s": 0.4075690999998187, + "outer_iterations": 69, + "inner_iterations": 11, + "total_iterations": 80, + "n_passive": 1200, + "warm_start_errors": 80, + "index": 15 + }, + { + "warm_start_error_fraction": 0.05801526717557252, + "reconstruction_s": 1.4538301999991745, + "eval_s": 1.5340866000005917, + "log_likelihood": 7175.679342224557, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.655146799999784, + "n": 1310, + "solve_s": 0.655146799999784, + "outer_iterations": 68, + "inner_iterations": 14, + "total_iterations": 82, + "n_passive": 1197, + "warm_start_errors": 76, + "index": 16 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 1.2227576000004774, + "eval_s": 1.3060992000000624, + "log_likelihood": 7224.232752132497, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6765599999998813, + "n": 1310, + "solve_s": 0.6765599999998813, + "outer_iterations": 60, + "inner_iterations": 11, + "total_iterations": 71, + "n_passive": 1192, + "warm_start_errors": 72, + "index": 17 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 1.1597924999996394, + "eval_s": 1.233803699999953, + "log_likelihood": 7250.149755309723, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3965100000004895, + "n": 1310, + "solve_s": 0.3965100000004895, + "outer_iterations": 53, + "inner_iterations": 11, + "total_iterations": 64, + "n_passive": 1193, + "warm_start_errors": 69, + "index": 18 + }, + { + "warm_start_error_fraction": 0.051908396946564885, + "reconstruction_s": 0.9249801999994816, + "eval_s": 1.0090138000005027, + "log_likelihood": 7349.56988625817, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3365744000002451, + "n": 1310, + "solve_s": 0.3365744000002451, + "outer_iterations": 54, + "inner_iterations": 12, + "total_iterations": 66, + "n_passive": 1200, + "warm_start_errors": 68, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 1.1522171999995408, + "eval_s": 1.220811400000457, + "log_likelihood": 7206.2076831241275, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5925379999998768, + "n": 1310, + "solve_s": 0.5925379999998768, + "outer_iterations": 69, + "inner_iterations": 15, + "total_iterations": 84, + "n_passive": 1195, + "warm_start_errors": 77, + "index": 0 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.5532087999999931, + "eval_s": 0.6266805000004751, + "log_likelihood": 7180.432957367257, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05861810000078549, + "n": 1310, + "solve_s": 0.05861810000078549, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1187, + "warm_start_errors": 8, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0015267175572519084, + "reconstruction_s": 0.5671043000002101, + "eval_s": 0.6451382999994166, + "log_likelihood": 7143.9948890357955, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04619280000042636, + "n": 1310, + "solve_s": 0.04619280000042636, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1187, + "warm_start_errors": 2, + "index": 2 + }, + { + "warm_start_error_fraction": 0.008396946564885497, + "reconstruction_s": 0.5724102000003768, + "eval_s": 0.6474371000003885, + "log_likelihood": 7182.878722012543, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0675586999996085, + "n": 1310, + "solve_s": 0.0675586999996085, + "outer_iterations": 9, + "inner_iterations": 3, + "total_iterations": 12, + "n_passive": 1190, + "warm_start_errors": 11, + "index": 3 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.5624731999996584, + "eval_s": 0.6372682999999597, + "log_likelihood": 7186.1758717889115, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06234409999979107, + "n": 1310, + "solve_s": 0.06234409999979107, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 1189, + "warm_start_errors": 9, + "index": 4 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.6389512000005197, + "eval_s": 0.7200453000004927, + "log_likelihood": 7145.024669580045, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05998029999955179, + "n": 1310, + "solve_s": 0.05998029999955179, + "outer_iterations": 4, + "inner_iterations": 3, + "total_iterations": 7, + "n_passive": 1188, + "warm_start_errors": 9, + "index": 5 + }, + { + "warm_start_error_fraction": 0.0053435114503816794, + "reconstruction_s": 0.6734409999999116, + "eval_s": 0.7482374000001073, + "log_likelihood": 7202.126666486729, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05206530000032217, + "n": 1310, + "solve_s": 0.05206530000032217, + "outer_iterations": 5, + "inner_iterations": 2, + "total_iterations": 7, + "n_passive": 1191, + "warm_start_errors": 7, + "index": 6 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 0.5011107999998785, + "eval_s": 0.5641828999996505, + "log_likelihood": 7222.820219978841, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06309750000036729, + "n": 1310, + "solve_s": 0.06309750000036729, + "outer_iterations": 10, + "inner_iterations": 2, + "total_iterations": 12, + "n_passive": 1194, + "warm_start_errors": 13, + "index": 7 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.4880710999996154, + "eval_s": 0.5504985999996279, + "log_likelihood": 7245.156725065451, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04307489999973768, + "n": 1310, + "solve_s": 0.04307489999973768, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1186, + "warm_start_errors": 10, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0015267175572519084, + "reconstruction_s": 0.5058378000003358, + "eval_s": 0.5666633999999249, + "log_likelihood": 7163.55551490529, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.044011899999532034, + "n": 1310, + "solve_s": 0.044011899999532034, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1186, + "warm_start_errors": 2, + "index": 9 + }, + { + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 0.47891140000047017, + "eval_s": 0.5404939999998533, + "log_likelihood": 7149.7410031695235, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04446490000009362, + "n": 1310, + "solve_s": 0.04446490000009362, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1186, + "warm_start_errors": 6, + "index": 10 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.5415046000007351, + "eval_s": 0.6250992000004771, + "log_likelihood": 7135.510856830093, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07241900000008172, + "n": 1310, + "solve_s": 0.07241900000008172, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1187, + "warm_start_errors": 9, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 0.590050899999369, + "eval_s": 0.6676853999997547, + "log_likelihood": 7124.427599165536, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.051813299999594165, + "n": 1310, + "solve_s": 0.051813299999594165, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1187, + "warm_start_errors": 4, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 0.6040137999998478, + "eval_s": 0.6749155000006795, + "log_likelihood": 7138.787138439637, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049294100000224716, + "n": 1310, + "solve_s": 0.049294100000224716, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1188, + "warm_start_errors": 3, + "index": 13 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 0.5016832999999679, + "eval_s": 0.5633834999998726, + "log_likelihood": 7112.535941006625, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.055902099999912025, + "n": 1310, + "solve_s": 0.055902099999912025, + "outer_iterations": 8, + "inner_iterations": 3, + "total_iterations": 11, + "n_passive": 1189, + "warm_start_errors": 13, + "index": 14 + }, + { + "warm_start_error_fraction": 0.011450381679389313, + "reconstruction_s": 0.5541107000008196, + "eval_s": 0.6185993999997663, + "log_likelihood": 7123.205873360791, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09361709999939194, + "n": 1310, + "solve_s": 0.09361709999939194, + "outer_iterations": 16, + "inner_iterations": 2, + "total_iterations": 18, + "n_passive": 1200, + "warm_start_errors": 15, + "index": 15 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.5837894999995115, + "eval_s": 0.6527864999998201, + "log_likelihood": 7175.679342224557, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05159029999958875, + "n": 1310, + "solve_s": 0.05159029999958875, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 1196, + "warm_start_errors": 10, + "index": 16 + }, + { + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 0.6976380999994944, + "eval_s": 0.7592223000001468, + "log_likelihood": 7224.2327521324605, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1882421999998769, + "n": 1310, + "solve_s": 0.1882421999998769, + "outer_iterations": 25, + "inner_iterations": 4, + "total_iterations": 29, + "n_passive": 1192, + "warm_start_errors": 14, + "index": 17 + }, + { + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 0.5370278999998845, + "eval_s": 0.5998898999996527, + "log_likelihood": 7250.14975530976, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04419040000084351, + "n": 1310, + "solve_s": 0.04419040000084351, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1192, + "warm_start_errors": 6, + "index": 18 + }, + { + "warm_start_error_fraction": 0.00916030534351145, + "reconstruction_s": 0.5354200000001583, + "eval_s": 0.6188956000005419, + "log_likelihood": 7349.569886258088, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05928390000008221, + "n": 1310, + "solve_s": 0.05928390000008221, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 1200, + "warm_start_errors": 12, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.0732824427480916, + "mean_warm_start_error_fraction": 0.07412213740458015, + "p90_warm_start_error_fraction": 0.09618320610687024, + "max_warm_start_error_fraction": 0.10610687022900764, + "median_outer_iterations": 70.0, + "mean_outer_iterations": 70.3, + "p90_outer_iterations": 92.30000000000001, + "max_outer_iterations": 106.0, + "median_inner_iterations": 17.0, + "mean_inner_iterations": 15.65, + "p90_inner_iterations": 20.200000000000003, + "max_inner_iterations": 28.0, + "median_total_iterations": 86.0, + "mean_total_iterations": 85.95, + "p90_total_iterations": 110.50000000000003, + "max_total_iterations": 134.0, + "median_n_passive": 1177.0, + "mean_n_passive": 1176.0, + "p90_n_passive": 1194.1, + "max_n_passive": 1196.0, + "median_warm_start_errors": 96.0, + "mean_warm_start_errors": 97.1, + "p90_warm_start_errors": 126.00000000000001, + "max_warm_start_errors": 139.0, + "median_solve_s": 0.5226786499997615, + "mean_solve_s": 0.5387889099999938, + "p90_solve_s": 0.7862128700001451, + "max_solve_s": 0.8799480999996376, + "median_reconstruction_s": 1.180768899999748, + "mean_reconstruction_s": 1.2031900949999454, + "p90_reconstruction_s": 1.5174891800003933, + "max_reconstruction_s": 1.600774100000308, + "median_eval_s": 1.2630002500000046, + "mean_eval_s": 1.292762035000169, + "p90_eval_s": 1.6360031100005472, + "max_eval_s": 1.7174493000002258 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.07328244274809159, + "mean_warm_start_error_fraction": 0.07561068702290076, + "p90_warm_start_error_fraction": 0.09809160305343512, + "max_warm_start_error_fraction": 0.10610687022900764, + "median_outer_iterations": 80.5, + "mean_outer_iterations": 77.0, + "p90_outer_iterations": 100.60000000000001, + "max_outer_iterations": 122.0, + "median_inner_iterations": 14.5, + "mean_inner_iterations": 15.45, + "p90_inner_iterations": 20.400000000000006, + "max_inner_iterations": 28.0, + "median_total_iterations": 94.0, + "mean_total_iterations": 92.45, + "p90_total_iterations": 120.50000000000003, + "max_total_iterations": 146.0, + "median_n_passive": 1177.0, + "mean_n_passive": 1175.7, + "p90_n_passive": 1194.1, + "max_n_passive": 1195.0, + "median_warm_start_errors": 96.0, + "mean_warm_start_errors": 99.05, + "p90_warm_start_errors": 128.5, + "max_warm_start_errors": 139.0, + "median_solve_s": 0.32435025000040696, + "mean_solve_s": 0.336861854999961, + "p90_solve_s": 0.40667269000041417, + "max_solve_s": 0.8253371000000698, + "median_reconstruction_s": 0.8093666499999017, + "mean_reconstruction_s": 0.875369555000043, + "p90_reconstruction_s": 1.1225778899996839, + "max_reconstruction_s": 1.7354615999993257, + "median_eval_s": 0.8732388999997056, + "mean_eval_s": 0.9470141449999119, + "p90_eval_s": 1.2258293399992908, + "max_eval_s": 1.8093591999995624 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 8.576436264303573e-15, + "max_abs_reconstruction_deviation": 1.2760601464378851e-09 + }, + "iteration_reduction_ratio": 0.9148936170212766, + "solve_speedup_ratio": 1.611463687785367, + "eval_speedup_ratio": 1.446339884767422, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.10610687022900764, + "reconstruction_s": 1.0694388999991133, + "eval_s": 1.1353159999998752, + "log_likelihood": 5471.7973737752245, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.53111810000064, + "n": 1310, + "solve_s": 0.53111810000064, + "outer_iterations": 106, + "inner_iterations": 28, + "total_iterations": 134, + "n_passive": 1140, + "warm_start_errors": 139, + "index": 0 + }, + { + "warm_start_error_fraction": 0.07862595419847328, + "reconstruction_s": 1.0141176000006453, + "eval_s": 1.0769338000000062, + "log_likelihood": 6654.532108619627, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4424100999995062, + "n": 1310, + "solve_s": 0.4424100999995062, + "outer_iterations": 71, + "inner_iterations": 18, + "total_iterations": 89, + "n_passive": 1167, + "warm_start_errors": 103, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0633587786259542, + "reconstruction_s": 0.9345229000000472, + "eval_s": 1.003352100000484, + "log_likelihood": 7226.971211085075, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.36127230000056443, + "n": 1310, + "solve_s": 0.36127230000056443, + "outer_iterations": 56, + "inner_iterations": 11, + "total_iterations": 67, + "n_passive": 1195, + "warm_start_errors": 83, + "index": 2 + }, + { + "warm_start_error_fraction": 0.0900763358778626, + "reconstruction_s": 1.0061226999996506, + "eval_s": 1.0737341000003653, + "log_likelihood": 6607.394360361324, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39560379999966244, + "n": 1310, + "solve_s": 0.39560379999966244, + "outer_iterations": 75, + "inner_iterations": 12, + "total_iterations": 87, + "n_passive": 1160, + "warm_start_errors": 118, + "index": 3 + }, + { + "warm_start_error_fraction": 0.06564885496183206, + "reconstruction_s": 0.9941299000001891, + "eval_s": 1.070907199999965, + "log_likelihood": 7059.915357499807, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4321158999991894, + "n": 1310, + "solve_s": 0.4321158999991894, + "outer_iterations": 66, + "inner_iterations": 11, + "total_iterations": 77, + "n_passive": 1183, + "warm_start_errors": 86, + "index": 4 + }, + { + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 1.208215200000268, + "eval_s": 1.3085202999991452, + "log_likelihood": 7022.075373233347, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6046263000007457, + "n": 1310, + "solve_s": 0.6046263000007457, + "outer_iterations": 71, + "inner_iterations": 15, + "total_iterations": 86, + "n_passive": 1187, + "warm_start_errors": 87, + "index": 5 + }, + { + "warm_start_error_fraction": 0.07633587786259542, + "reconstruction_s": 1.4026508000006288, + "eval_s": 1.5075014000003648, + "log_likelihood": 6851.779927925216, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7777734000001146, + "n": 1310, + "solve_s": 0.7777734000001146, + "outer_iterations": 69, + "inner_iterations": 17, + "total_iterations": 86, + "n_passive": 1178, + "warm_start_errors": 100, + "index": 6 + }, + { + "warm_start_error_fraction": 0.07480916030534351, + "reconstruction_s": 1.5180029000002833, + "eval_s": 1.6552676999999676, + "log_likelihood": 6489.95669202, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5954559000001609, + "n": 1310, + "solve_s": 0.5954559000001609, + "outer_iterations": 65, + "inner_iterations": 19, + "total_iterations": 84, + "n_passive": 1174, + "warm_start_errors": 98, + "index": 7 + }, + { + "warm_start_error_fraction": 0.07480916030534351, + "reconstruction_s": 1.0890393999998196, + "eval_s": 1.154534500000409, + "log_likelihood": 6595.61511794514, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3808134999999311, + "n": 1310, + "solve_s": 0.3808134999999311, + "outer_iterations": 58, + "inner_iterations": 17, + "total_iterations": 75, + "n_passive": 1173, + "warm_start_errors": 98, + "index": 8 + }, + { + "warm_start_error_fraction": 0.10305343511450382, + "reconstruction_s": 1.2199014999996507, + "eval_s": 1.2904344000007768, + "log_likelihood": 5179.313174140179, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.59070950000023, + "n": 1310, + "solve_s": 0.59070950000023, + "outer_iterations": 91, + "inner_iterations": 18, + "total_iterations": 109, + "n_passive": 1154, + "warm_start_errors": 135, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0717557251908397, + "reconstruction_s": 1.1918017999996664, + "eval_s": 1.3013746000005995, + "log_likelihood": 6904.463092361186, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5275978999998188, + "n": 1310, + "solve_s": 0.5275978999998188, + "outer_iterations": 74, + "inner_iterations": 12, + "total_iterations": 86, + "n_passive": 1189, + "warm_start_errors": 94, + "index": 10 + }, + { + "warm_start_error_fraction": 0.0816793893129771, + "reconstruction_s": 1.1722553999998127, + "eval_s": 1.2551271000002089, + "log_likelihood": 5670.821967236286, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.45610450000003766, + "n": 1310, + "solve_s": 0.45610450000003766, + "outer_iterations": 61, + "inner_iterations": 17, + "total_iterations": 78, + "n_passive": 1153, + "warm_start_errors": 107, + "index": 11 + }, + { + "warm_start_error_fraction": 0.07022900763358779, + "reconstruction_s": 1.4384273999994548, + "eval_s": 1.5433724000004077, + "log_likelihood": 6645.924461980561, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6698171000007278, + "n": 1310, + "solve_s": 0.6698171000007278, + "outer_iterations": 64, + "inner_iterations": 22, + "total_iterations": 86, + "n_passive": 1158, + "warm_start_errors": 92, + "index": 12 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 1.0808876999999484, + "eval_s": 1.1460301999995863, + "log_likelihood": 7151.479245469516, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30577999999968597, + "n": 1310, + "solve_s": 0.30577999999968597, + "outer_iterations": 39, + "inner_iterations": 11, + "total_iterations": 50, + "n_passive": 1176, + "warm_start_errors": 73, + "index": 13 + }, + { + "warm_start_error_fraction": 0.051908396946564885, + "reconstruction_s": 1.028137099999185, + "eval_s": 1.1142021000005116, + "log_likelihood": 7191.732883022228, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4802116999999271, + "n": 1310, + "solve_s": 0.4802116999999271, + "outer_iterations": 74, + "inner_iterations": 17, + "total_iterations": 91, + "n_passive": 1193, + "warm_start_errors": 68, + "index": 14 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 1.5174321000004056, + "eval_s": 1.6338626000006116, + "log_likelihood": 7160.20116161501, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8799480999996376, + "n": 1310, + "solve_s": 0.8799480999996376, + "outer_iterations": 80, + "inner_iterations": 19, + "total_iterations": 99, + "n_passive": 1196, + "warm_start_errors": 73, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06870229007633588, + "reconstruction_s": 1.2713052999997672, + "eval_s": 1.398732800000289, + "log_likelihood": 7076.978400817326, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5177593999997043, + "n": 1310, + "solve_s": 0.5177593999997043, + "outer_iterations": 51, + "inner_iterations": 7, + "total_iterations": 58, + "n_passive": 1192, + "warm_start_errors": 90, + "index": 16 + }, + { + "warm_start_error_fraction": 0.05343511450381679, + "reconstruction_s": 1.1173568000003797, + "eval_s": 1.1977146999997785, + "log_likelihood": 7435.3671852353855, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4026274999996531, + "n": 1310, + "solve_s": 0.4026274999996531, + "outer_iterations": 43, + "inner_iterations": 7, + "total_iterations": 50, + "n_passive": 1194, + "warm_start_errors": 70, + "index": 17 + }, + { + "warm_start_error_fraction": 0.07862595419847328, + "reconstruction_s": 1.1892823999996835, + "eval_s": 1.2708733999998003, + "log_likelihood": 6574.8371202406925, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5618650999995225, + "n": 1310, + "solve_s": 0.5618650999995225, + "outer_iterations": 88, + "inner_iterations": 15, + "total_iterations": 103, + "n_passive": 1188, + "warm_start_errors": 103, + "index": 18 + }, + { + "warm_start_error_fraction": 0.09541984732824428, + "reconstruction_s": 1.600774100000308, + "eval_s": 1.7174493000002258, + "log_likelihood": 6171.152613875354, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8621681000004173, + "n": 1310, + "solve_s": 0.8621681000004173, + "outer_iterations": 104, + "inner_iterations": 20, + "total_iterations": 124, + "n_passive": 1170, + "warm_start_errors": 125, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.10610687022900764, + "reconstruction_s": 1.7354615999993257, + "eval_s": 1.8093591999995624, + "log_likelihood": 5471.7973737752245, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8253371000000698, + "n": 1310, + "solve_s": 0.8253371000000698, + "outer_iterations": 106, + "inner_iterations": 28, + "total_iterations": 134, + "n_passive": 1140, + "warm_start_errors": 139, + "index": 0 + }, + { + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 0.8552139000003081, + "eval_s": 0.9176611000002595, + "log_likelihood": 6654.5321086196445, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3578760999998849, + "n": 1310, + "solve_s": 0.3578760999998849, + "outer_iterations": 100, + "inner_iterations": 19, + "total_iterations": 119, + "n_passive": 1167, + "warm_start_errors": 115, + "index": 1 + }, + { + "warm_start_error_fraction": 0.08244274809160305, + "reconstruction_s": 0.8542305999999371, + "eval_s": 0.9174653999998554, + "log_likelihood": 7226.971211085047, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3762783000001946, + "n": 1310, + "solve_s": 0.3762783000001946, + "outer_iterations": 91, + "inner_iterations": 20, + "total_iterations": 111, + "n_passive": 1195, + "warm_start_errors": 108, + "index": 2 + }, + { + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 0.9196972999998252, + "eval_s": 0.9838020999995933, + "log_likelihood": 6607.394360361362, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3905400000003283, + "n": 1310, + "solve_s": 0.3905400000003283, + "outer_iterations": 97, + "inner_iterations": 19, + "total_iterations": 116, + "n_passive": 1160, + "warm_start_errors": 115, + "index": 3 + }, + { + "warm_start_error_fraction": 0.06946564885496183, + "reconstruction_s": 0.871152799999436, + "eval_s": 0.9422368000005008, + "log_likelihood": 7059.915357499809, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33083740000074613, + "n": 1310, + "solve_s": 0.33083740000074613, + "outer_iterations": 80, + "inner_iterations": 14, + "total_iterations": 94, + "n_passive": 1183, + "warm_start_errors": 91, + "index": 4 + }, + { + "warm_start_error_fraction": 0.06259541984732825, + "reconstruction_s": 0.8176262999995743, + "eval_s": 0.8793146999996679, + "log_likelihood": 7022.075373233368, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3178631000000678, + "n": 1310, + "solve_s": 0.3178631000000678, + "outer_iterations": 81, + "inner_iterations": 13, + "total_iterations": 94, + "n_passive": 1187, + "warm_start_errors": 82, + "index": 5 + }, + { + "warm_start_error_fraction": 0.07557251908396946, + "reconstruction_s": 0.8011070000002292, + "eval_s": 0.8671630999997433, + "log_likelihood": 6851.7799279252395, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23991369999930612, + "n": 1310, + "solve_s": 0.23991369999930612, + "outer_iterations": 55, + "inner_iterations": 16, + "total_iterations": 71, + "n_passive": 1178, + "warm_start_errors": 99, + "index": 6 + }, + { + "warm_start_error_fraction": 0.10152671755725191, + "reconstruction_s": 0.786661400000412, + "eval_s": 0.8520950000001903, + "log_likelihood": 6489.956692020017, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34355149999919377, + "n": 1310, + "solve_s": 0.34355149999919377, + "outer_iterations": 97, + "inner_iterations": 18, + "total_iterations": 115, + "n_passive": 1173, + "warm_start_errors": 133, + "index": 7 + }, + { + "warm_start_error_fraction": 0.08091603053435115, + "reconstruction_s": 0.8605425000005198, + "eval_s": 0.9318610000000263, + "log_likelihood": 6595.615117945145, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3834046000001763, + "n": 1310, + "solve_s": 0.3834046000001763, + "outer_iterations": 82, + "inner_iterations": 17, + "total_iterations": 99, + "n_passive": 1173, + "warm_start_errors": 106, + "index": 8 + }, + { + "warm_start_error_fraction": 0.083206106870229, + "reconstruction_s": 1.1223518999995576, + "eval_s": 1.2461858999995457, + "log_likelihood": 5179.313174140187, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39266809999935504, + "n": 1310, + "solve_s": 0.39266809999935504, + "outer_iterations": 59, + "inner_iterations": 12, + "total_iterations": 71, + "n_passive": 1154, + "warm_start_errors": 109, + "index": 9 + }, + { + "warm_start_error_fraction": 0.07938931297709924, + "reconstruction_s": 1.124611800000821, + "eval_s": 1.2235674999992625, + "log_likelihood": 6904.463092361187, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4220905000001949, + "n": 1310, + "solve_s": 0.4220905000001949, + "outer_iterations": 87, + "inner_iterations": 16, + "total_iterations": 103, + "n_passive": 1188, + "warm_start_errors": 104, + "index": 10 + }, + { + "warm_start_error_fraction": 0.07099236641221374, + "reconstruction_s": 0.7524676000002728, + "eval_s": 0.823221999999987, + "log_likelihood": 5670.821967236286, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23436399999991409, + "n": 1310, + "solve_s": 0.23436399999991409, + "outer_iterations": 57, + "inner_iterations": 12, + "total_iterations": 69, + "n_passive": 1153, + "warm_start_errors": 93, + "index": 11 + }, + { + "warm_start_error_fraction": 0.09770992366412214, + "reconstruction_s": 0.9314054000005854, + "eval_s": 0.9927501000001939, + "log_likelihood": 6645.9244619805595, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4049596000004385, + "n": 1310, + "solve_s": 0.4049596000004385, + "outer_iterations": 122, + "inner_iterations": 24, + "total_iterations": 146, + "n_passive": 1157, + "warm_start_errors": 128, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.7689614999999321, + "eval_s": 0.8336953999996695, + "log_likelihood": 7151.479245469577, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3069824999993216, + "n": 1310, + "solve_s": 0.3069824999993216, + "outer_iterations": 56, + "inner_iterations": 8, + "total_iterations": 64, + "n_passive": 1176, + "warm_start_errors": 75, + "index": 13 + }, + { + "warm_start_error_fraction": 0.06793893129770992, + "reconstruction_s": 0.7541143999997075, + "eval_s": 0.8184542999997575, + "log_likelihood": 7191.732883022249, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24673190000066825, + "n": 1310, + "solve_s": 0.24673190000066825, + "outer_iterations": 71, + "inner_iterations": 11, + "total_iterations": 82, + "n_passive": 1193, + "warm_start_errors": 89, + "index": 14 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 0.6964693999998417, + "eval_s": 0.7720077000003585, + "log_likelihood": 7160.201161614967, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22538319999966916, + "n": 1310, + "solve_s": 0.22538319999966916, + "outer_iterations": 46, + "inner_iterations": 12, + "total_iterations": 58, + "n_passive": 1195, + "warm_start_errors": 66, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06412213740458016, + "reconstruction_s": 0.7523261000005732, + "eval_s": 0.8172311000007539, + "log_likelihood": 7076.978400817332, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24751290000040171, + "n": 1310, + "solve_s": 0.24751290000040171, + "outer_iterations": 70, + "inner_iterations": 15, + "total_iterations": 85, + "n_passive": 1191, + "warm_start_errors": 84, + "index": 16 + }, + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 0.6546877999999197, + "eval_s": 0.7375579999998081, + "log_likelihood": 7435.367185235396, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2003380999994988, + "n": 1310, + "solve_s": 0.2003380999994988, + "outer_iterations": 44, + "inner_iterations": 13, + "total_iterations": 57, + "n_passive": 1194, + "warm_start_errors": 67, + "index": 17 + }, + { + "warm_start_error_fraction": 0.06946564885496183, + "reconstruction_s": 0.7986227999999755, + "eval_s": 0.8637203999996927, + "log_likelihood": 6574.837120240636, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3000173000000359, + "n": 1310, + "solve_s": 0.3000173000000359, + "outer_iterations": 88, + "inner_iterations": 13, + "total_iterations": 101, + "n_passive": 1187, + "warm_start_errors": 91, + "index": 18 + }, + { + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 0.6496790000001056, + "eval_s": 0.710932099999809, + "log_likelihood": 6171.1526138753525, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.19058719999975438, + "n": 1310, + "solve_s": 0.19058719999975438, + "outer_iterations": 51, + "inner_iterations": 9, + "total_iterations": 60, + "n_passive": 1170, + "warm_start_errors": 87, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_edge_zeroing_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_edge_zeroing_v2026.8.17.1.png new file mode 100644 index 000000000..cea548797 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_edge_zeroing_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_lens_light_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_lens_light_v2026.8.17.1.json new file mode 100644 index 000000000..ad86dc009 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_lens_light_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "no_lens_light", + "configuration": { + "model": "no_lens_light", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 7, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "none", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 3.121683500000472, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.07719999999999999, + "mean_warm_start_error_fraction": 0.07604000000000001, + "p90_warm_start_error_fraction": 0.0816, + "max_warm_start_error_fraction": 0.084, + "median_outer_iterations": 53.0, + "mean_outer_iterations": 51.95, + "p90_outer_iterations": 58.1, + "max_outer_iterations": 59.0, + "median_inner_iterations": 6.0, + "mean_inner_iterations": 5.95, + "p90_inner_iterations": 8.0, + "max_inner_iterations": 8.0, + "median_total_iterations": 59.0, + "mean_total_iterations": 57.9, + "p90_total_iterations": 64.1, + "max_total_iterations": 67.0, + "median_n_passive": 1093.0, + "mean_n_passive": 1092.85, + "p90_n_passive": 1097.0, + "max_n_passive": 1098.0, + "median_warm_start_errors": 96.5, + "mean_warm_start_errors": 95.05, + "p90_warm_start_errors": 102.0, + "max_warm_start_errors": 105.0, + "median_solve_s": 0.2887706500000604, + "mean_solve_s": 0.32181939000001875, + "p90_solve_s": 0.47283842000033477, + "max_solve_s": 0.5194222999998601, + "median_reconstruction_s": 0.5001204499994856, + "mean_reconstruction_s": 0.5417841200000566, + "p90_reconstruction_s": 0.6946061400004511, + "max_reconstruction_s": 0.8067454000001817, + "median_eval_s": 0.5602694500003054, + "mean_eval_s": 0.6052585250001357, + "p90_eval_s": 0.7917062999998962, + "max_eval_s": 0.879464800000278 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.014, + "mean_warm_start_error_fraction": 0.017679999999999998, + "p90_warm_start_error_fraction": 0.02568, + "max_warm_start_error_fraction": 0.0776, + "median_outer_iterations": 11.0, + "mean_outer_iterations": 14.0, + "p90_outer_iterations": 23.000000000000014, + "max_outer_iterations": 51.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 2.75, + "p90_inner_iterations": 8.0, + "max_inner_iterations": 8.0, + "median_total_iterations": 14.5, + "mean_total_iterations": 16.75, + "p90_total_iterations": 27.000000000000014, + "max_total_iterations": 59.0, + "median_n_passive": 1093.0, + "mean_n_passive": 1092.85, + "p90_n_passive": 1097.0, + "max_n_passive": 1098.0, + "median_warm_start_errors": 17.5, + "mean_warm_start_errors": 22.1, + "p90_warm_start_errors": 32.1, + "max_warm_start_errors": 97.0, + "median_solve_s": 0.06617945000016334, + "mean_solve_s": 0.0873324500000308, + "p90_solve_s": 0.11534309000035137, + "max_solve_s": 0.347581099999843, + "median_reconstruction_s": 0.2541452000004938, + "mean_reconstruction_s": 0.2752632650000123, + "p90_reconstruction_s": 0.31315219000052824, + "max_reconstruction_s": 0.617518499999278, + "median_eval_s": 0.3240693000002466, + "mean_eval_s": 0.34025442999995903, + "p90_eval_s": 0.3794697899998937, + "max_eval_s": 0.6789730000000418 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 4.489662249638087e-16, + "max_abs_reconstruction_deviation": 3.9168668308775523e-13 + }, + "iteration_reduction_ratio": 4.068965517241379, + "solve_speedup_ratio": 4.3634489256007365, + "eval_speedup_ratio": 1.728856914246055, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.0776, + "reconstruction_s": 0.7238312999998016, + "eval_s": 0.8005875000008018, + "log_likelihood": -129609.02569551891, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4694556000004013, + "n": 1250, + "solve_s": 0.4694556000004013, + "outer_iterations": 51, + "inner_iterations": 8, + "total_iterations": 59, + "n_passive": 1089, + "warm_start_errors": 97, + "index": 0 + }, + { + "warm_start_error_fraction": 0.076, + "reconstruction_s": 0.5889477999999144, + "eval_s": 0.6506067000000257, + "log_likelihood": -129699.31956064393, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3344568000002255, + "n": 1250, + "solve_s": 0.3344568000002255, + "outer_iterations": 58, + "inner_iterations": 5, + "total_iterations": 63, + "n_passive": 1098, + "warm_start_errors": 95, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0736, + "reconstruction_s": 0.49285560000043915, + "eval_s": 0.5587181000000783, + "log_likelihood": -129529.30661006982, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26950849999957427, + "n": 1250, + "solve_s": 0.26950849999957427, + "outer_iterations": 43, + "inner_iterations": 5, + "total_iterations": 48, + "n_passive": 1085, + "warm_start_errors": 92, + "index": 2 + }, + { + "warm_start_error_fraction": 0.0792, + "reconstruction_s": 0.6558025000003909, + "eval_s": 0.7321763999998439, + "log_likelihood": -129695.3094546835, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42219810000005964, + "n": 1250, + "solve_s": 0.42219810000005964, + "outer_iterations": 58, + "inner_iterations": 6, + "total_iterations": 64, + "n_passive": 1095, + "warm_start_errors": 99, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0816, + "reconstruction_s": 0.8067454000001817, + "eval_s": 0.879464800000278, + "log_likelihood": -129678.67586882139, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5194222999998601, + "n": 1250, + "solve_s": 0.5194222999998601, + "outer_iterations": 59, + "inner_iterations": 8, + "total_iterations": 67, + "n_passive": 1097, + "warm_start_errors": 102, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0776, + "reconstruction_s": 0.48958239999956277, + "eval_s": 0.5421513999999661, + "log_likelihood": -129648.19551439429, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2687242000001788, + "n": 1250, + "solve_s": 0.2687242000001788, + "outer_iterations": 52, + "inner_iterations": 6, + "total_iterations": 58, + "n_passive": 1089, + "warm_start_errors": 97, + "index": 5 + }, + { + "warm_start_error_fraction": 0.0752, + "reconstruction_s": 0.4443375000000742, + "eval_s": 0.4979714999999487, + "log_likelihood": -129775.20756388904, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2565058999998655, + "n": 1250, + "solve_s": 0.2565058999998655, + "outer_iterations": 56, + "inner_iterations": 4, + "total_iterations": 60, + "n_passive": 1095, + "warm_start_errors": 94, + "index": 6 + }, + { + "warm_start_error_fraction": 0.0768, + "reconstruction_s": 0.45883930000036344, + "eval_s": 0.520460799999455, + "log_likelihood": -129690.2356166174, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26480629999969096, + "n": 1250, + "solve_s": 0.26480629999969096, + "outer_iterations": 54, + "inner_iterations": 5, + "total_iterations": 59, + "n_passive": 1091, + "warm_start_errors": 96, + "index": 7 + }, + { + "warm_start_error_fraction": 0.084, + "reconstruction_s": 0.5756846000003861, + "eval_s": 0.6289059000000634, + "log_likelihood": -129740.33995866956, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.295726299999842, + "n": 1250, + "solve_s": 0.295726299999842, + "outer_iterations": 57, + "inner_iterations": 8, + "total_iterations": 65, + "n_passive": 1092, + "warm_start_errors": 105, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0784, + "reconstruction_s": 0.44243460000052437, + "eval_s": 0.4957235000001674, + "log_likelihood": -129777.64460292362, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25711450000017066, + "n": 1250, + "solve_s": 0.25711450000017066, + "outer_iterations": 59, + "inner_iterations": 4, + "total_iterations": 63, + "n_passive": 1096, + "warm_start_errors": 98, + "index": 9 + }, + { + "warm_start_error_fraction": 0.08, + "reconstruction_s": 0.47367430000031163, + "eval_s": 0.5314736999998786, + "log_likelihood": -129717.79437197777, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27771990000019287, + "n": 1250, + "solve_s": 0.27771990000019287, + "outer_iterations": 55, + "inner_iterations": 6, + "total_iterations": 61, + "n_passive": 1093, + "warm_start_errors": 100, + "index": 10 + }, + { + "warm_start_error_fraction": 0.076, + "reconstruction_s": 0.4740410999993401, + "eval_s": 0.5321652000002359, + "log_likelihood": -129885.34335599234, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2721358999997392, + "n": 1250, + "solve_s": 0.2721358999997392, + "outer_iterations": 52, + "inner_iterations": 6, + "total_iterations": 58, + "n_passive": 1097, + "warm_start_errors": 95, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0784, + "reconstruction_s": 0.50417059999927, + "eval_s": 0.5614567000002353, + "log_likelihood": -129700.20993040738, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29610560000037367, + "n": 1250, + "solve_s": 0.29610560000037367, + "outer_iterations": 58, + "inner_iterations": 5, + "total_iterations": 63, + "n_passive": 1094, + "warm_start_errors": 98, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0816, + "reconstruction_s": 0.526162000000113, + "eval_s": 0.5898942000003444, + "log_likelihood": -129657.11679745495, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.322320000000218, + "n": 1250, + "solve_s": 0.322320000000218, + "outer_iterations": 52, + "inner_iterations": 7, + "total_iterations": 59, + "n_passive": 1087, + "warm_start_errors": 102, + "index": 13 + }, + { + "warm_start_error_fraction": 0.08, + "reconstruction_s": 0.517595700000129, + "eval_s": 0.5800723000002108, + "log_likelihood": -129688.93184612735, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3119334000002709, + "n": 1250, + "solve_s": 0.3119334000002709, + "outer_iterations": 55, + "inner_iterations": 5, + "total_iterations": 60, + "n_passive": 1092, + "warm_start_errors": 100, + "index": 14 + }, + { + "warm_start_error_fraction": 0.0768, + "reconstruction_s": 0.4960702999997011, + "eval_s": 0.5590822000003755, + "log_likelihood": -129652.62933419808, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2818150000002788, + "n": 1250, + "solve_s": 0.2818150000002788, + "outer_iterations": 47, + "inner_iterations": 5, + "total_iterations": 52, + "n_passive": 1091, + "warm_start_errors": 96, + "index": 15 + }, + { + "warm_start_error_fraction": 0.072, + "reconstruction_s": 0.48015850000047067, + "eval_s": 0.5382428000002619, + "log_likelihood": -129638.18165859999, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2703996999998708, + "n": 1250, + "solve_s": 0.2703996999998708, + "outer_iterations": 49, + "inner_iterations": 6, + "total_iterations": 55, + "n_passive": 1093, + "warm_start_errors": 90, + "index": 16 + }, + { + "warm_start_error_fraction": 0.064, + "reconstruction_s": 0.4334054999999353, + "eval_s": 0.48773590000018885, + "log_likelihood": -129604.23634512539, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23690999999962514, + "n": 1250, + "solve_s": 0.23690999999962514, + "outer_iterations": 41, + "inner_iterations": 7, + "total_iterations": 48, + "n_passive": 1093, + "warm_start_errors": 80, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0688, + "reconstruction_s": 0.6913589000005231, + "eval_s": 0.7907194999997955, + "log_likelihood": -129471.11061521282, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5032837999997355, + "n": 1250, + "solve_s": 0.5032837999997355, + "outer_iterations": 41, + "inner_iterations": 7, + "total_iterations": 48, + "n_passive": 1094, + "warm_start_errors": 86, + "index": 18 + }, + { + "warm_start_error_fraction": 0.0632, + "reconstruction_s": 0.5599844999997003, + "eval_s": 0.6275614000005589, + "log_likelihood": -129610.1817402064, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30584600000020146, + "n": 1250, + "solve_s": 0.30584600000020146, + "outer_iterations": 42, + "inner_iterations": 6, + "total_iterations": 48, + "n_passive": 1096, + "warm_start_errors": 79, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.0776, + "reconstruction_s": 0.617518499999278, + "eval_s": 0.6789730000000418, + "log_likelihood": -129609.02569551891, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.347581099999843, + "n": 1250, + "solve_s": 0.347581099999843, + "outer_iterations": 51, + "inner_iterations": 8, + "total_iterations": 59, + "n_passive": 1089, + "warm_start_errors": 97, + "index": 0 + }, + { + "warm_start_error_fraction": 0.0104, + "reconstruction_s": 0.23200719999931607, + "eval_s": 0.29336479999983567, + "log_likelihood": -129699.31956064393, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05729539999992994, + "n": 1250, + "solve_s": 0.05729539999992994, + "outer_iterations": 11, + "inner_iterations": 1, + "total_iterations": 12, + "n_passive": 1098, + "warm_start_errors": 13, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0264, + "reconstruction_s": 0.24116260000027978, + "eval_s": 0.29564310000023397, + "log_likelihood": -129529.30661006985, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08651360000021668, + "n": 1250, + "solve_s": 0.08651360000021668, + "outer_iterations": 18, + "inner_iterations": 8, + "total_iterations": 26, + "n_passive": 1085, + "warm_start_errors": 33, + "index": 2 + }, + { + "warm_start_error_fraction": 0.0224, + "reconstruction_s": 0.27851409999948373, + "eval_s": 0.3337597999998252, + "log_likelihood": -129695.3094546835, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11777119999987917, + "n": 1250, + "solve_s": 0.11777119999987917, + "outer_iterations": 32, + "inner_iterations": 4, + "total_iterations": 36, + "n_passive": 1095, + "warm_start_errors": 28, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0064, + "reconstruction_s": 0.20537859999967623, + "eval_s": 0.2624651999994967, + "log_likelihood": -129678.67586882139, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04728660000000673, + "n": 1250, + "solve_s": 0.04728660000000673, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1097, + "warm_start_errors": 8, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0096, + "reconstruction_s": 0.2177211999996871, + "eval_s": 0.3197217000006276, + "log_likelihood": -129648.19551439423, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.045943500000248605, + "n": 1250, + "solve_s": 0.045943500000248605, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1089, + "warm_start_errors": 12, + "index": 5 + }, + { + "warm_start_error_fraction": 0.0064, + "reconstruction_s": 0.2755403999999544, + "eval_s": 0.3397239999994781, + "log_likelihood": -129775.20756388904, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.057681399999637506, + "n": 1250, + "solve_s": 0.057681399999637506, + "outer_iterations": 8, + "inner_iterations": 1, + "total_iterations": 9, + "n_passive": 1095, + "warm_start_errors": 8, + "index": 6 + }, + { + "warm_start_error_fraction": 0.008, + "reconstruction_s": 0.22170360000018263, + "eval_s": 0.2759034000000611, + "log_likelihood": -129690.2356166174, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.045385900000837864, + "n": 1250, + "solve_s": 0.045385900000837864, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1091, + "warm_start_errors": 10, + "index": 7 + }, + { + "warm_start_error_fraction": 0.0168, + "reconstruction_s": 0.2140903999998045, + "eval_s": 0.2689626000001226, + "log_likelihood": -129740.33995866956, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0594517999998061, + "n": 1250, + "solve_s": 0.0594517999998061, + "outer_iterations": 11, + "inner_iterations": 6, + "total_iterations": 17, + "n_passive": 1092, + "warm_start_errors": 21, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0128, + "reconstruction_s": 0.3011189999997441, + "eval_s": 0.36383589999968535, + "log_likelihood": -129777.64460292362, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1140242000001308, + "n": 1250, + "solve_s": 0.1140242000001308, + "outer_iterations": 19, + "inner_iterations": 1, + "total_iterations": 20, + "n_passive": 1096, + "warm_start_errors": 16, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0168, + "reconstruction_s": 0.22531460000027437, + "eval_s": 0.29241210000054707, + "log_likelihood": -129717.79437197777, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05882689999998547, + "n": 1250, + "solve_s": 0.05882689999998547, + "outer_iterations": 12, + "inner_iterations": 1, + "total_iterations": 13, + "n_passive": 1093, + "warm_start_errors": 21, + "index": 10 + }, + { + "warm_start_error_fraction": 0.0256, + "reconstruction_s": 0.3085752000006323, + "eval_s": 0.3773615999998583, + "log_likelihood": -129885.34335599234, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11507330000040383, + "n": 1250, + "solve_s": 0.11507330000040383, + "outer_iterations": 22, + "inner_iterations": 2, + "total_iterations": 24, + "n_passive": 1097, + "warm_start_errors": 32, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0232, + "reconstruction_s": 0.2671278000007078, + "eval_s": 0.32841689999986556, + "log_likelihood": -129700.2099304074, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07730429999992339, + "n": 1250, + "solve_s": 0.07730429999992339, + "outer_iterations": 15, + "inner_iterations": 1, + "total_iterations": 16, + "n_passive": 1094, + "warm_start_errors": 29, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0152, + "reconstruction_s": 0.22431460000007064, + "eval_s": 0.2858560000004218, + "log_likelihood": -129657.11679745495, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05410639999990963, + "n": 1250, + "solve_s": 0.05410639999990963, + "outer_iterations": 9, + "inner_iterations": 2, + "total_iterations": 11, + "n_passive": 1087, + "warm_start_errors": 19, + "index": 13 + }, + { + "warm_start_error_fraction": 0.012, + "reconstruction_s": 0.23670149999998102, + "eval_s": 0.2971263999997973, + "log_likelihood": -129688.93184612735, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06276400000024296, + "n": 1250, + "solve_s": 0.06276400000024296, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 1092, + "warm_start_errors": 15, + "index": 14 + }, + { + "warm_start_error_fraction": 0.012, + "reconstruction_s": 0.32050690000050963, + "eval_s": 0.39357269999982236, + "log_likelihood": -129652.62933419808, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08086220000041067, + "n": 1250, + "solve_s": 0.08086220000041067, + "outer_iterations": 8, + "inner_iterations": 8, + "total_iterations": 16, + "n_passive": 1091, + "warm_start_errors": 15, + "index": 15 + }, + { + "warm_start_error_fraction": 0.0032, + "reconstruction_s": 0.23305570000047737, + "eval_s": 0.2976594000001569, + "log_likelihood": -129638.18165859999, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0447876999996879, + "n": 1250, + "solve_s": 0.0447876999996879, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1093, + "warm_start_errors": 4, + "index": 16 + }, + { + "warm_start_error_fraction": 0.0176, + "reconstruction_s": 0.2926484999998138, + "eval_s": 0.36920829999962734, + "log_likelihood": -129604.23634512539, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09739399999943998, + "n": 1250, + "solve_s": 0.09739399999943998, + "outer_iterations": 15, + "inner_iterations": 2, + "total_iterations": 17, + "n_passive": 1093, + "warm_start_errors": 22, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0104, + "reconstruction_s": 0.27992989999984275, + "eval_s": 0.3532188999997743, + "log_likelihood": -129471.11061521282, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06959490000008373, + "n": 1250, + "solve_s": 0.06959490000008373, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 1094, + "warm_start_errors": 13, + "index": 18 + }, + { + "warm_start_error_fraction": 0.0208, + "reconstruction_s": 0.31233500000053027, + "eval_s": 0.3779027999999016, + "log_likelihood": -129610.18174020643, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10700059999999212, + "n": 1250, + "solve_s": 0.10700059999999212, + "outer_iterations": 16, + "inner_iterations": 4, + "total_iterations": 20, + "n_passive": 1096, + "warm_start_errors": 26, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.0728, + "mean_warm_start_error_fraction": 0.07404, + "p90_warm_start_error_fraction": 0.08272, + "max_warm_start_error_fraction": 0.0856, + "median_outer_iterations": 46.0, + "mean_outer_iterations": 47.6, + "p90_outer_iterations": 54.1, + "max_outer_iterations": 63.0, + "median_inner_iterations": 6.0, + "mean_inner_iterations": 6.8, + "p90_inner_iterations": 10.0, + "max_inner_iterations": 10.0, + "median_total_iterations": 53.0, + "mean_total_iterations": 54.4, + "p90_total_iterations": 62.300000000000004, + "max_total_iterations": 69.0, + "median_n_passive": 1089.0, + "mean_n_passive": 1089.0, + "p90_n_passive": 1096.1, + "max_n_passive": 1102.0, + "median_warm_start_errors": 91.0, + "mean_warm_start_errors": 92.55, + "p90_warm_start_errors": 103.4, + "max_warm_start_errors": 107.0, + "median_solve_s": 0.29989354999997886, + "mean_solve_s": 0.2975961749999897, + "p90_solve_s": 0.35974447000016885, + "max_solve_s": 0.385971900000186, + "median_reconstruction_s": 0.5013754999999946, + "mean_reconstruction_s": 0.5150652549999905, + "p90_reconstruction_s": 0.6235751100000926, + "max_reconstruction_s": 0.6784432000004017, + "median_eval_s": 0.5578394999997727, + "mean_eval_s": 0.5766761650001172, + "p90_eval_s": 0.6942265500003487, + "max_eval_s": 0.7432906000003641 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.09359999999999999, + "mean_warm_start_error_fraction": 0.09124, + "p90_warm_start_error_fraction": 0.11960000000000001, + "max_warm_start_error_fraction": 0.1288, + "median_outer_iterations": 81.0, + "mean_outer_iterations": 79.0, + "p90_outer_iterations": 110.60000000000001, + "max_outer_iterations": 130.0, + "median_inner_iterations": 18.0, + "mean_inner_iterations": 18.65, + "p90_inner_iterations": 29.1, + "max_inner_iterations": 35.0, + "median_total_iterations": 101.0, + "mean_total_iterations": 97.65, + "p90_total_iterations": 135.70000000000002, + "max_total_iterations": 160.0, + "median_n_passive": 1089.0, + "mean_n_passive": 1089.0, + "p90_n_passive": 1096.1, + "max_n_passive": 1102.0, + "median_warm_start_errors": 117.0, + "mean_warm_start_errors": 114.05, + "p90_warm_start_errors": 149.5, + "max_warm_start_errors": 161.0, + "median_solve_s": 0.30561909999960335, + "mean_solve_s": 0.34948163499993823, + "p90_solve_s": 0.5615258300003916, + "max_solve_s": 0.5803835999995499, + "median_reconstruction_s": 0.5065591000002314, + "mean_reconstruction_s": 0.549159735000012, + "p90_reconstruction_s": 0.7865864800001873, + "max_reconstruction_s": 0.8215820999994321, + "median_eval_s": 0.5785744499999055, + "mean_eval_s": 0.6223885600000812, + "p90_eval_s": 0.8660315900005117, + "max_eval_s": 0.9296457000000373 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 2.248747083709884e-16, + "max_abs_reconstruction_deviation": 3.821387650759789e-13 + }, + "iteration_reduction_ratio": 0.5247524752475248, + "solve_speedup_ratio": 0.9812657324112534, + "eval_speedup_ratio": 0.9641620019685692, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.0712, + "reconstruction_s": 0.5977852999994866, + "eval_s": 0.6875665000006848, + "log_likelihood": -129564.36026544767, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34503380000023753, + "n": 1250, + "solve_s": 0.34503380000023753, + "outer_iterations": 46, + "inner_iterations": 6, + "total_iterations": 52, + "n_passive": 1087, + "warm_start_errors": 89, + "index": 0 + }, + { + "warm_start_error_fraction": 0.0688, + "reconstruction_s": 0.6222010000001319, + "eval_s": 0.6915034000003288, + "log_likelihood": -129848.2631441436, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3373520000004646, + "n": 1250, + "solve_s": 0.3373520000004646, + "outer_iterations": 46, + "inner_iterations": 6, + "total_iterations": 52, + "n_passive": 1092, + "warm_start_errors": 86, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0776, + "reconstruction_s": 0.5734979999997449, + "eval_s": 0.6400516000003336, + "log_likelihood": -129718.56187866037, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3418740999995862, + "n": 1250, + "solve_s": 0.3418740999995862, + "outer_iterations": 50, + "inner_iterations": 10, + "total_iterations": 60, + "n_passive": 1088, + "warm_start_errors": 97, + "index": 2 + }, + { + "warm_start_error_fraction": 0.0648, + "reconstruction_s": 0.6359420999997383, + "eval_s": 0.718734900000527, + "log_likelihood": -129422.42668179245, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3742674999994051, + "n": 1250, + "solve_s": 0.3742674999994051, + "outer_iterations": 40, + "inner_iterations": 8, + "total_iterations": 48, + "n_passive": 1097, + "warm_start_errors": 81, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0744, + "reconstruction_s": 0.6784432000004017, + "eval_s": 0.7432906000003641, + "log_likelihood": -129734.9802740046, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.385971900000186, + "n": 1250, + "solve_s": 0.385971900000186, + "outer_iterations": 43, + "inner_iterations": 9, + "total_iterations": 52, + "n_passive": 1087, + "warm_start_errors": 93, + "index": 4 + }, + { + "warm_start_error_fraction": 0.076, + "reconstruction_s": 0.5435317000001305, + "eval_s": 0.6006451000002926, + "log_likelihood": -129373.88118618593, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3247993000004499, + "n": 1250, + "solve_s": 0.3247993000004499, + "outer_iterations": 55, + "inner_iterations": 10, + "total_iterations": 65, + "n_passive": 1090, + "warm_start_errors": 95, + "index": 5 + }, + { + "warm_start_error_fraction": 0.0696, + "reconstruction_s": 0.4551973999996335, + "eval_s": 0.5090885999998136, + "log_likelihood": -129588.4016840255, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25586569999995845, + "n": 1250, + "solve_s": 0.25586569999995845, + "outer_iterations": 46, + "inner_iterations": 9, + "total_iterations": 55, + "n_passive": 1091, + "warm_start_errors": 87, + "index": 6 + }, + { + "warm_start_error_fraction": 0.0632, + "reconstruction_s": 0.5590363999999681, + "eval_s": 0.6344069999995554, + "log_likelihood": -129708.61797646474, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35813080000025366, + "n": 1250, + "solve_s": 0.35813080000025366, + "outer_iterations": 42, + "inner_iterations": 5, + "total_iterations": 47, + "n_passive": 1081, + "warm_start_errors": 79, + "index": 7 + }, + { + "warm_start_error_fraction": 0.0712, + "reconstruction_s": 0.4400876999998218, + "eval_s": 0.5087627000002612, + "log_likelihood": -129484.31698745192, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2292254000003595, + "n": 1250, + "solve_s": 0.2292254000003595, + "outer_iterations": 46, + "inner_iterations": 6, + "total_iterations": 52, + "n_passive": 1091, + "warm_start_errors": 89, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0704, + "reconstruction_s": 0.4516153000004124, + "eval_s": 0.502400100000159, + "log_likelihood": -129865.04775899838, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25373750000017026, + "n": 1250, + "solve_s": 0.25373750000017026, + "outer_iterations": 53, + "inner_iterations": 4, + "total_iterations": 57, + "n_passive": 1095, + "warm_start_errors": 88, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0856, + "reconstruction_s": 0.44020850000015344, + "eval_s": 0.4961637000005794, + "log_likelihood": -130295.60230149562, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26059989999976096, + "n": 1250, + "solve_s": 0.26059989999976096, + "outer_iterations": 44, + "inner_iterations": 5, + "total_iterations": 49, + "n_passive": 1076, + "warm_start_errors": 107, + "index": 10 + }, + { + "warm_start_error_fraction": 0.0784, + "reconstruction_s": 0.4850817000005918, + "eval_s": 0.544488900000033, + "log_likelihood": -130354.48301552088, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.291807599999629, + "n": 1250, + "solve_s": 0.291807599999629, + "outer_iterations": 51, + "inner_iterations": 8, + "total_iterations": 59, + "n_passive": 1087, + "warm_start_errors": 98, + "index": 11 + }, + { + "warm_start_error_fraction": 0.072, + "reconstruction_s": 0.5022961000004216, + "eval_s": 0.5583040999999866, + "log_likelihood": -129855.85151620256, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3079795000003287, + "n": 1250, + "solve_s": 0.3079795000003287, + "outer_iterations": 46, + "inner_iterations": 8, + "total_iterations": 54, + "n_passive": 1096, + "warm_start_errors": 90, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0704, + "reconstruction_s": 0.43523110000023735, + "eval_s": 0.48955189999924187, + "log_likelihood": -129431.99052371125, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23909900000035123, + "n": 1250, + "solve_s": 0.23909900000035123, + "outer_iterations": 47, + "inner_iterations": 5, + "total_iterations": 52, + "n_passive": 1102, + "warm_start_errors": 88, + "index": 13 + }, + { + "warm_start_error_fraction": 0.0696, + "reconstruction_s": 0.4675968000001376, + "eval_s": 0.5236746999999013, + "log_likelihood": -130070.94431280644, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26949770000010176, + "n": 1250, + "solve_s": 0.26949770000010176, + "outer_iterations": 41, + "inner_iterations": 6, + "total_iterations": 47, + "n_passive": 1085, + "warm_start_errors": 87, + "index": 14 + }, + { + "warm_start_error_fraction": 0.0856, + "reconstruction_s": 0.5004548999995677, + "eval_s": 0.5582728999997926, + "log_likelihood": -129641.43417540003, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3089927999999418, + "n": 1250, + "solve_s": 0.3089927999999418, + "outer_iterations": 63, + "inner_iterations": 6, + "total_iterations": 69, + "n_passive": 1092, + "warm_start_errors": 107, + "index": 15 + }, + { + "warm_start_error_fraction": 0.0736, + "reconstruction_s": 0.45659880000039266, + "eval_s": 0.5089168000004065, + "log_likelihood": -129497.30659052849, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.253885099999934, + "n": 1250, + "solve_s": 0.253885099999934, + "outer_iterations": 54, + "inner_iterations": 5, + "total_iterations": 59, + "n_passive": 1095, + "warm_start_errors": 92, + "index": 16 + }, + { + "warm_start_error_fraction": 0.0784, + "reconstruction_s": 0.5202140000001236, + "eval_s": 0.5763279000002512, + "log_likelihood": -130380.93923304827, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3114676999994117, + "n": 1250, + "solve_s": 0.3114676999994117, + "outer_iterations": 52, + "inner_iterations": 10, + "total_iterations": 62, + "n_passive": 1077, + "warm_start_errors": 98, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0824, + "reconstruction_s": 0.5042484999994485, + "eval_s": 0.5574060999997528, + "log_likelihood": -130340.70370161903, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2732780999995157, + "n": 1250, + "solve_s": 0.2732780999995157, + "outer_iterations": 48, + "inner_iterations": 7, + "total_iterations": 55, + "n_passive": 1084, + "warm_start_errors": 103, + "index": 18 + }, + { + "warm_start_error_fraction": 0.0776, + "reconstruction_s": 0.4320365999992646, + "eval_s": 0.48396580000007816, + "log_likelihood": -130210.31695258693, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22905809999974736, + "n": 1250, + "solve_s": 0.22905809999974736, + "outer_iterations": 39, + "inner_iterations": 3, + "total_iterations": 42, + "n_passive": 1087, + "warm_start_errors": 97, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.0712, + "reconstruction_s": 0.4121260000001712, + "eval_s": 0.46351290000075096, + "log_likelihood": -129564.36026544767, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22967700000026525, + "n": 1250, + "solve_s": 0.22967700000026525, + "outer_iterations": 46, + "inner_iterations": 6, + "total_iterations": 52, + "n_passive": 1087, + "warm_start_errors": 89, + "index": 0 + }, + { + "warm_start_error_fraction": 0.0904, + "reconstruction_s": 0.4404230000000098, + "eval_s": 0.4952070999997886, + "log_likelihood": -129848.2631441436, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29380219999984547, + "n": 1250, + "solve_s": 0.29380219999984547, + "outer_iterations": 78, + "inner_iterations": 15, + "total_iterations": 93, + "n_passive": 1092, + "warm_start_errors": 113, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0576, + "reconstruction_s": 0.33672999999998865, + "eval_s": 0.3906115999998292, + "log_likelihood": -129718.56187866037, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.16507739999997284, + "n": 1250, + "solve_s": 0.16507739999997284, + "outer_iterations": 52, + "inner_iterations": 7, + "total_iterations": 59, + "n_passive": 1088, + "warm_start_errors": 72, + "index": 2 + }, + { + "warm_start_error_fraction": 0.0968, + "reconstruction_s": 0.41950829999950656, + "eval_s": 0.4775299000002633, + "log_likelihood": -129422.42668179242, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27520169999934296, + "n": 1250, + "solve_s": 0.27520169999934296, + "outer_iterations": 82, + "inner_iterations": 24, + "total_iterations": 106, + "n_passive": 1097, + "warm_start_errors": 121, + "index": 3 + }, + { + "warm_start_error_fraction": 0.1056, + "reconstruction_s": 0.42866400000002614, + "eval_s": 0.4790909000003012, + "log_likelihood": -129734.9802740046, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2552843000003122, + "n": 1250, + "solve_s": 0.2552843000003122, + "outer_iterations": 87, + "inner_iterations": 14, + "total_iterations": 101, + "n_passive": 1087, + "warm_start_errors": 132, + "index": 4 + }, + { + "warm_start_error_fraction": 0.1048, + "reconstruction_s": 0.4902357999999367, + "eval_s": 0.5489008000004105, + "log_likelihood": -129373.88118618593, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.343766200000573, + "n": 1250, + "solve_s": 0.343766200000573, + "outer_iterations": 105, + "inner_iterations": 29, + "total_iterations": 134, + "n_passive": 1090, + "warm_start_errors": 131, + "index": 5 + }, + { + "warm_start_error_fraction": 0.1032, + "reconstruction_s": 0.3797816000005696, + "eval_s": 0.42977669999982027, + "log_likelihood": -129588.4016840255, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22142580000036105, + "n": 1250, + "solve_s": 0.22142580000036105, + "outer_iterations": 83, + "inner_iterations": 18, + "total_iterations": 101, + "n_passive": 1091, + "warm_start_errors": 129, + "index": 6 + }, + { + "warm_start_error_fraction": 0.088, + "reconstruction_s": 0.4683481000001848, + "eval_s": 0.5598958000000493, + "log_likelihood": -129708.61797646474, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31743599999936123, + "n": 1250, + "solve_s": 0.31743599999936123, + "outer_iterations": 75, + "inner_iterations": 29, + "total_iterations": 104, + "n_passive": 1081, + "warm_start_errors": 110, + "index": 7 + }, + { + "warm_start_error_fraction": 0.0864, + "reconstruction_s": 0.4943696000000273, + "eval_s": 0.5555341999997836, + "log_likelihood": -129484.31698745192, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27527890000055777, + "n": 1250, + "solve_s": 0.27527890000055777, + "outer_iterations": 73, + "inner_iterations": 23, + "total_iterations": 96, + "n_passive": 1091, + "warm_start_errors": 108, + "index": 8 + }, + { + "warm_start_error_fraction": 0.088, + "reconstruction_s": 0.6207643999996435, + "eval_s": 0.700162399999499, + "log_likelihood": -129865.04775899838, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41335799999978917, + "n": 1250, + "solve_s": 0.41335799999978917, + "outer_iterations": 83, + "inner_iterations": 18, + "total_iterations": 101, + "n_passive": 1095, + "warm_start_errors": 110, + "index": 9 + }, + { + "warm_start_error_fraction": 0.1192, + "reconstruction_s": 0.795293799999854, + "eval_s": 0.9296457000000373, + "log_likelihood": -130295.60230149564, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5494699999999284, + "n": 1250, + "solve_s": 0.5494699999999284, + "outer_iterations": 81, + "inner_iterations": 28, + "total_iterations": 109, + "n_passive": 1076, + "warm_start_errors": 149, + "index": 10 + }, + { + "warm_start_error_fraction": 0.036, + "reconstruction_s": 0.5283767999999327, + "eval_s": 0.6291511999997965, + "log_likelihood": -130354.48301552088, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2170890999996118, + "n": 1250, + "solve_s": 0.2170890999996118, + "outer_iterations": 34, + "inner_iterations": 3, + "total_iterations": 37, + "n_passive": 1087, + "warm_start_errors": 45, + "index": 11 + }, + { + "warm_start_error_fraction": 0.1288, + "reconstruction_s": 0.8215820999994321, + "eval_s": 0.9009803000008105, + "log_likelihood": -129855.85151620256, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5803835999995499, + "n": 1250, + "solve_s": 0.5803835999995499, + "outer_iterations": 130, + "inner_iterations": 30, + "total_iterations": 160, + "n_passive": 1096, + "warm_start_errors": 161, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0992, + "reconstruction_s": 0.5288137999996252, + "eval_s": 0.5975468000006003, + "log_likelihood": -129431.99052371125, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3235201999996207, + "n": 1250, + "solve_s": 0.3235201999996207, + "outer_iterations": 81, + "inner_iterations": 24, + "total_iterations": 105, + "n_passive": 1102, + "warm_start_errors": 124, + "index": 13 + }, + { + "warm_start_error_fraction": 0.1176, + "reconstruction_s": 0.7856190000002243, + "eval_s": 0.8621484000004784, + "log_likelihood": -130070.94431280644, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5759828000000198, + "n": 1250, + "solve_s": 0.5759828000000198, + "outer_iterations": 98, + "inner_iterations": 26, + "total_iterations": 124, + "n_passive": 1085, + "warm_start_errors": 147, + "index": 14 + }, + { + "warm_start_error_fraction": 0.1192, + "reconstruction_s": 0.7773409999999785, + "eval_s": 0.8469620999994731, + "log_likelihood": -129641.43417540003, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5599195000004329, + "n": 1250, + "solve_s": 0.5599195000004329, + "outer_iterations": 110, + "inner_iterations": 18, + "total_iterations": 128, + "n_passive": 1092, + "warm_start_errors": 149, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06, + "reconstruction_s": 0.5187486000004355, + "eval_s": 0.5972530999997616, + "log_likelihood": -129497.30659052852, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2744626999992761, + "n": 1250, + "solve_s": 0.2744626999992761, + "outer_iterations": 53, + "inner_iterations": 11, + "total_iterations": 64, + "n_passive": 1095, + "warm_start_errors": 75, + "index": 16 + }, + { + "warm_start_error_fraction": 0.1232, + "reconstruction_s": 0.751854000000094, + "eval_s": 0.8213007000003927, + "log_likelihood": -130380.93923304827, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5480164999999033, + "n": 1250, + "solve_s": 0.5480164999999033, + "outer_iterations": 116, + "inner_iterations": 35, + "total_iterations": 151, + "n_passive": 1077, + "warm_start_errors": 154, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0712, + "reconstruction_s": 0.5239138000006278, + "eval_s": 0.6098156999996718, + "log_likelihood": -130340.70370161903, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33660150000014255, + "n": 1250, + "solve_s": 0.33660150000014255, + "outer_iterations": 63, + "inner_iterations": 9, + "total_iterations": 72, + "n_passive": 1084, + "warm_start_errors": 89, + "index": 18 + }, + { + "warm_start_error_fraction": 0.0584, + "reconstruction_s": 0.4607009999999718, + "eval_s": 0.552744900000107, + "log_likelihood": -130210.31695258693, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23387929999989865, + "n": 1250, + "solve_s": 0.23387929999989865, + "outer_iterations": 50, + "inner_iterations": 6, + "total_iterations": 56, + "n_passive": 1087, + "warm_start_errors": 73, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_lens_light_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_lens_light_v2026.8.17.1.png new file mode 100644 index 000000000..2a4d321d1 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_no_lens_light_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_powerlaw_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_powerlaw_v2026.8.17.1.json new file mode 100644 index 000000000..089a35006 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_powerlaw_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "powerlaw", + "configuration": { + "model": "powerlaw", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 12, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 2.409054200000355, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.05267175572519084, + "mean_warm_start_error_fraction": 0.05259541984732824, + "p90_warm_start_error_fraction": 0.05656488549618321, + "max_warm_start_error_fraction": 0.05877862595419847, + "median_outer_iterations": 51.0, + "mean_outer_iterations": 50.9, + "p90_outer_iterations": 56.0, + "max_outer_iterations": 61.0, + "median_inner_iterations": 8.5, + "mean_inner_iterations": 9.35, + "p90_inner_iterations": 12.100000000000001, + "max_inner_iterations": 16.0, + "median_total_iterations": 59.5, + "mean_total_iterations": 60.25, + "p90_total_iterations": 64.70000000000002, + "max_total_iterations": 72.0, + "median_n_passive": 1192.5, + "mean_n_passive": 1191.8, + "p90_n_passive": 1195.0, + "max_n_passive": 1196.0, + "median_warm_start_errors": 69.0, + "mean_warm_start_errors": 68.9, + "p90_warm_start_errors": 74.1, + "max_warm_start_errors": 77.0, + "median_solve_s": 0.2809268000000884, + "mean_solve_s": 0.31946282999988396, + "p90_solve_s": 0.4227402999992592, + "max_solve_s": 0.4698150000003807, + "median_reconstruction_s": 0.8122832499998367, + "mean_reconstruction_s": 0.8671019299998989, + "p90_reconstruction_s": 1.0089759000001322, + "max_reconstruction_s": 1.097037000000455, + "median_eval_s": 0.8715384500001164, + "mean_eval_s": 0.9373648149999554, + "p90_eval_s": 1.100120849999803, + "max_eval_s": 1.1713545000002341 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.007251908396946565, + "mean_warm_start_error_fraction": 0.011603053435114504, + "p90_warm_start_error_fraction": 0.020610687022900764, + "max_warm_start_error_fraction": 0.05725190839694656, + "median_outer_iterations": 6.5, + "mean_outer_iterations": 9.45, + "p90_outer_iterations": 15.300000000000004, + "max_outer_iterations": 56.0, + "median_inner_iterations": 1.5, + "mean_inner_iterations": 2.35, + "p90_inner_iterations": 3.0, + "max_inner_iterations": 16.0, + "median_total_iterations": 7.5, + "mean_total_iterations": 11.8, + "p90_total_iterations": 18.300000000000004, + "max_total_iterations": 72.0, + "median_n_passive": 1192.0, + "mean_n_passive": 1191.5, + "p90_n_passive": 1195.0, + "max_n_passive": 1196.0, + "median_warm_start_errors": 9.5, + "mean_warm_start_errors": 15.2, + "p90_warm_start_errors": 27.0, + "max_warm_start_errors": 75.0, + "median_solve_s": 0.05067090000011376, + "mean_solve_s": 0.07010652999997546, + "p90_solve_s": 0.07728580000066361, + "max_solve_s": 0.354048000000148, + "median_reconstruction_s": 0.5033244500004912, + "mean_reconstruction_s": 0.5225149349999356, + "p90_reconstruction_s": 0.5644009299998288, + "max_reconstruction_s": 0.8551245000007839, + "median_eval_s": 0.5629947000002176, + "mean_eval_s": 0.5826528949999101, + "p90_eval_s": 0.6250519800003531, + "max_eval_s": 0.9140880000004472 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 2.639114591626331e-15, + "max_abs_reconstruction_deviation": 9.080594054466928e-10 + }, + "iteration_reduction_ratio": 7.933333333333334, + "solve_speedup_ratio": 5.544144666849369, + "eval_speedup_ratio": 1.5480402390995505, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.8355319000002055, + "eval_s": 0.894199800000024, + "log_likelihood": 7206.667312644546, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3455015999998068, + "n": 1310, + "solve_s": 0.3455015999998068, + "outer_iterations": 56, + "inner_iterations": 16, + "total_iterations": 72, + "n_passive": 1195, + "warm_start_errors": 75, + "index": 0 + }, + { + "warm_start_error_fraction": 0.05343511450381679, + "reconstruction_s": 0.8076472999991893, + "eval_s": 0.8649898000003304, + "log_likelihood": 7281.888036193719, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2737682999995741, + "n": 1310, + "solve_s": 0.2737682999995741, + "outer_iterations": 47, + "inner_iterations": 11, + "total_iterations": 58, + "n_passive": 1192, + "warm_start_errors": 70, + "index": 1 + }, + { + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 0.7997877000007065, + "eval_s": 0.8577982999995584, + "log_likelihood": 7298.633894265651, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2802579999997761, + "n": 1310, + "solve_s": 0.2802579999997761, + "outer_iterations": 53, + "inner_iterations": 11, + "total_iterations": 64, + "n_passive": 1195, + "warm_start_errors": 71, + "index": 2 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.733198099999754, + "eval_s": 0.8147256000002017, + "log_likelihood": 7304.956499717503, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27666669999962323, + "n": 1310, + "solve_s": 0.27666669999962323, + "outer_iterations": 51, + "inner_iterations": 6, + "total_iterations": 57, + "n_passive": 1190, + "warm_start_errors": 69, + "index": 3 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 0.7667694000001575, + "eval_s": 0.8299292000001515, + "log_likelihood": 7282.497371888521, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25068000000010215, + "n": 1310, + "solve_s": 0.25068000000010215, + "outer_iterations": 47, + "inner_iterations": 9, + "total_iterations": 56, + "n_passive": 1191, + "warm_start_errors": 65, + "index": 4 + }, + { + "warm_start_error_fraction": 0.04885496183206107, + "reconstruction_s": 1.097037000000455, + "eval_s": 1.1713545000002341, + "log_likelihood": 7299.637789188988, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4698150000003807, + "n": 1310, + "solve_s": 0.4698150000003807, + "outer_iterations": 46, + "inner_iterations": 6, + "total_iterations": 52, + "n_passive": 1193, + "warm_start_errors": 64, + "index": 5 + }, + { + "warm_start_error_fraction": 0.04885496183206107, + "reconstruction_s": 0.9887351999996099, + "eval_s": 1.1244267000001855, + "log_likelihood": 7310.091486134535, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33761630000026344, + "n": 1310, + "solve_s": 0.33761630000026344, + "outer_iterations": 49, + "inner_iterations": 8, + "total_iterations": 57, + "n_passive": 1194, + "warm_start_errors": 64, + "index": 6 + }, + { + "warm_start_error_fraction": 0.051908396946564885, + "reconstruction_s": 1.007684000000154, + "eval_s": 1.0674423000000388, + "log_likelihood": 7269.709712378224, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2638536999993448, + "n": 1310, + "solve_s": 0.2638536999993448, + "outer_iterations": 51, + "inner_iterations": 7, + "total_iterations": 58, + "n_passive": 1194, + "warm_start_errors": 68, + "index": 7 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.9906277999998565, + "eval_s": 1.0627642999997988, + "log_likelihood": 7301.631133537557, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39551230000051874, + "n": 1310, + "solve_s": 0.39551230000051874, + "outer_iterations": 56, + "inner_iterations": 7, + "total_iterations": 63, + "n_passive": 1194, + "warm_start_errors": 72, + "index": 8 + }, + { + "warm_start_error_fraction": 0.04732824427480916, + "reconstruction_s": 1.0206029999999373, + "eval_s": 1.0974201999997604, + "log_likelihood": 7282.354580196819, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.45556959999976243, + "n": 1310, + "solve_s": 0.45556959999976243, + "outer_iterations": 53, + "inner_iterations": 7, + "total_iterations": 60, + "n_passive": 1196, + "warm_start_errors": 62, + "index": 9 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.9279712999996264, + "eval_s": 0.9869810000000143, + "log_likelihood": 7237.044119960575, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25526479999916774, + "n": 1310, + "solve_s": 0.25526479999916774, + "outer_iterations": 52, + "inner_iterations": 8, + "total_iterations": 60, + "n_passive": 1191, + "warm_start_errors": 69, + "index": 10 + }, + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 0.9507074000002831, + "eval_s": 1.042501999999331, + "log_likelihood": 7229.863940154917, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41909259999920323, + "n": 1310, + "solve_s": 0.41909259999920323, + "outer_iterations": 50, + "inner_iterations": 8, + "total_iterations": 58, + "n_passive": 1190, + "warm_start_errors": 67, + "index": 11 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 0.9346038999992743, + "eval_s": 1.0291859999997541, + "log_likelihood": 7247.520349852948, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37869469999986904, + "n": 1310, + "solve_s": 0.37869469999986904, + "outer_iterations": 52, + "inner_iterations": 7, + "total_iterations": 59, + "n_passive": 1194, + "warm_start_errors": 65, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.7866014000001087, + "eval_s": 0.854628999999477, + "log_likelihood": 7236.912150104546, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.308878700000605, + "n": 1310, + "solve_s": 0.308878700000605, + "outer_iterations": 51, + "inner_iterations": 12, + "total_iterations": 63, + "n_passive": 1190, + "warm_start_errors": 72, + "index": 13 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.8128851000001305, + "eval_s": 0.8707617000000027, + "log_likelihood": 7294.558956604778, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2674683000004734, + "n": 1310, + "solve_s": 0.2674683000004734, + "outer_iterations": 50, + "inner_iterations": 11, + "total_iterations": 61, + "n_passive": 1189, + "warm_start_errors": 69, + "index": 14 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 0.7733215999996901, + "eval_s": 0.8314603999997416, + "log_likelihood": 7284.153512593673, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24626269999953365, + "n": 1310, + "solve_s": 0.24626269999953365, + "outer_iterations": 47, + "inner_iterations": 13, + "total_iterations": 60, + "n_passive": 1187, + "warm_start_errors": 66, + "index": 15 + }, + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 0.74081600000045, + "eval_s": 0.8002551999998104, + "log_likelihood": 7289.109754353289, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2813845999999103, + "n": 1310, + "solve_s": 0.2813845999999103, + "outer_iterations": 52, + "inner_iterations": 10, + "total_iterations": 62, + "n_passive": 1188, + "warm_start_errors": 67, + "index": 16 + }, + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 0.8116813999995429, + "eval_s": 0.8723152000002301, + "log_likelihood": 7335.410793773801, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.277641699999549, + "n": 1310, + "solve_s": 0.277641699999549, + "outer_iterations": 50, + "inner_iterations": 8, + "total_iterations": 58, + "n_passive": 1193, + "warm_start_errors": 77, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.7674738999994588, + "eval_s": 0.8267062000004444, + "log_likelihood": 7275.098804871538, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28046900000026653, + "n": 1310, + "solve_s": 0.28046900000026653, + "outer_iterations": 61, + "inner_iterations": 10, + "total_iterations": 71, + "n_passive": 1193, + "warm_start_errors": 72, + "index": 18 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.788355199999387, + "eval_s": 0.8474489000000176, + "log_likelihood": 7253.413620692649, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32485799999994924, + "n": 1310, + "solve_s": 0.32485799999994924, + "outer_iterations": 44, + "inner_iterations": 12, + "total_iterations": 56, + "n_passive": 1187, + "warm_start_errors": 74, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.8551245000007839, + "eval_s": 0.9140880000004472, + "log_likelihood": 7206.667312644546, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.354048000000148, + "n": 1310, + "solve_s": 0.354048000000148, + "outer_iterations": 56, + "inner_iterations": 16, + "total_iterations": 72, + "n_passive": 1195, + "warm_start_errors": 75, + "index": 0 + }, + { + "warm_start_error_fraction": 0.01450381679389313, + "reconstruction_s": 0.5660158000000592, + "eval_s": 0.622739499999625, + "log_likelihood": 7281.888036193732, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06510439999965456, + "n": 1310, + "solve_s": 0.06510439999965456, + "outer_iterations": 8, + "inner_iterations": 1, + "total_iterations": 9, + "n_passive": 1192, + "warm_start_errors": 19, + "index": 1 + }, + { + "warm_start_error_fraction": 0.008396946564885497, + "reconstruction_s": 0.5282003000002078, + "eval_s": 0.5860644000003958, + "log_likelihood": 7298.633894265668, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04836249999971187, + "n": 1310, + "solve_s": 0.04836249999971187, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 1195, + "warm_start_errors": 11, + "index": 2 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.47176979999949253, + "eval_s": 0.5299647000001642, + "log_likelihood": 7304.956499717496, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05060909999974683, + "n": 1310, + "solve_s": 0.05060909999974683, + "outer_iterations": 2, + "inner_iterations": 2, + "total_iterations": 4, + "n_passive": 1190, + "warm_start_errors": 9, + "index": 3 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 0.49355430000014167, + "eval_s": 0.5563957999993363, + "log_likelihood": 7282.497371888529, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05784419999963575, + "n": 1310, + "solve_s": 0.05784419999963575, + "outer_iterations": 8, + "inner_iterations": 3, + "total_iterations": 11, + "n_passive": 1191, + "warm_start_errors": 13, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0015267175572519084, + "reconstruction_s": 0.5146521999995457, + "eval_s": 0.5784475000000384, + "log_likelihood": 7299.637789188985, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.041246400000090944, + "n": 1310, + "solve_s": 0.041246400000090944, + "outer_iterations": 2, + "inner_iterations": 0, + "total_iterations": 2, + "n_passive": 1193, + "warm_start_errors": 2, + "index": 5 + }, + { + "warm_start_error_fraction": 0.0053435114503816794, + "reconstruction_s": 0.5295563999998194, + "eval_s": 0.5946484999994937, + "log_likelihood": 7310.091486134531, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0484759999999369, + "n": 1310, + "solve_s": 0.0484759999999369, + "outer_iterations": 4, + "inner_iterations": 2, + "total_iterations": 6, + "n_passive": 1194, + "warm_start_errors": 7, + "index": 6 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.5639390999995157, + "eval_s": 0.6243749000004755, + "log_likelihood": 7269.709712378222, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.043099399999846355, + "n": 1310, + "solve_s": 0.043099399999846355, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1194, + "warm_start_errors": 8, + "index": 7 + }, + { + "warm_start_error_fraction": 0.012213740458015267, + "reconstruction_s": 0.4883975000002465, + "eval_s": 0.5484561000002941, + "log_likelihood": 7301.631133537547, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0685273999997662, + "n": 1310, + "solve_s": 0.0685273999997662, + "outer_iterations": 11, + "inner_iterations": 2, + "total_iterations": 13, + "n_passive": 1194, + "warm_start_errors": 16, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.47491660000014235, + "eval_s": 0.5353720000002795, + "log_likelihood": 7282.3545801968285, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05073270000048069, + "n": 1310, + "solve_s": 0.05073270000048069, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1196, + "warm_start_errors": 8, + "index": 9 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.49975570000060543, + "eval_s": 0.5586930999998003, + "log_likelihood": 7237.044119960594, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04522919999999431, + "n": 1310, + "solve_s": 0.04522919999999431, + "outer_iterations": 3, + "inner_iterations": 3, + "total_iterations": 6, + "n_passive": 1190, + "warm_start_errors": 10, + "index": 10 + }, + { + "warm_start_error_fraction": 0.003816793893129771, + "reconstruction_s": 0.4679329000000507, + "eval_s": 0.5290906999998697, + "log_likelihood": 7229.863940154921, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03853179999987333, + "n": 1310, + "solve_s": 0.03853179999987333, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1189, + "warm_start_errors": 5, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0053435114503816794, + "reconstruction_s": 0.4674548999992112, + "eval_s": 0.5248913999994329, + "log_likelihood": 7247.520349852948, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.050737999999910244, + "n": 1310, + "solve_s": 0.050737999999910244, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 1194, + "warm_start_errors": 7, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.4806381999997029, + "eval_s": 0.5415001000001212, + "log_likelihood": 7236.912150104545, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04980340000020078, + "n": 1310, + "solve_s": 0.04980340000020078, + "outer_iterations": 4, + "inner_iterations": 2, + "total_iterations": 6, + "n_passive": 1190, + "warm_start_errors": 8, + "index": 13 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.506893200000377, + "eval_s": 0.5672963000006348, + "log_likelihood": 7294.558956604779, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04688110000006418, + "n": 1310, + "solve_s": 0.04688110000006418, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1188, + "warm_start_errors": 8, + "index": 14 + }, + { + "warm_start_error_fraction": 0.013740458015267175, + "reconstruction_s": 0.529450299999553, + "eval_s": 0.5881470999993326, + "log_likelihood": 7284.1535125936725, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0669938999999431, + "n": 1310, + "solve_s": 0.0669938999999431, + "outer_iterations": 13, + "inner_iterations": 1, + "total_iterations": 14, + "n_passive": 1186, + "warm_start_errors": 18, + "index": 15 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.4542825999997149, + "eval_s": 0.5123629999998229, + "log_likelihood": 7289.109754353289, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0450327999997171, + "n": 1310, + "solve_s": 0.0450327999997171, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 1188, + "warm_start_errors": 8, + "index": 16 + }, + { + "warm_start_error_fraction": 0.020610687022900764, + "reconstruction_s": 0.5642214999998032, + "eval_s": 0.6311456999992515, + "log_likelihood": 7335.410793773803, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09238960000038787, + "n": 1310, + "solve_s": 0.09238960000038787, + "outer_iterations": 18, + "inner_iterations": 3, + "total_iterations": 21, + "n_passive": 1193, + "warm_start_errors": 27, + "index": 17 + }, + { + "warm_start_error_fraction": 0.020610687022900764, + "reconstruction_s": 0.520437099999981, + "eval_s": 0.5785820000000967, + "log_likelihood": 7275.098804871543, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07560760000069422, + "n": 1310, + "solve_s": 0.07560760000069422, + "outer_iterations": 15, + "inner_iterations": 3, + "total_iterations": 18, + "n_passive": 1192, + "warm_start_errors": 27, + "index": 18 + }, + { + "warm_start_error_fraction": 0.013740458015267175, + "reconstruction_s": 0.47310579999975744, + "eval_s": 0.5307970999992904, + "log_likelihood": 7253.413620692643, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06287309999970603, + "n": 1310, + "solve_s": 0.06287309999970603, + "outer_iterations": 11, + "inner_iterations": 2, + "total_iterations": 13, + "n_passive": 1186, + "warm_start_errors": 18, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.08206106870229007, + "mean_warm_start_error_fraction": 0.07812977099236641, + "p90_warm_start_error_fraction": 0.0917557251908397, + "max_warm_start_error_fraction": 0.09923664122137404, + "median_outer_iterations": 77.5, + "mean_outer_iterations": 77.65, + "p90_outer_iterations": 100.0, + "max_outer_iterations": 109.0, + "median_inner_iterations": 15.0, + "mean_inner_iterations": 15.7, + "p90_inner_iterations": 20.0, + "max_inner_iterations": 21.0, + "median_total_iterations": 92.5, + "mean_total_iterations": 93.35, + "p90_total_iterations": 116.5, + "max_total_iterations": 129.0, + "median_n_passive": 1171.0, + "mean_n_passive": 1170.7, + "p90_n_passive": 1195.6, + "max_n_passive": 1206.0, + "median_warm_start_errors": 107.5, + "mean_warm_start_errors": 102.35, + "p90_warm_start_errors": 120.2, + "max_warm_start_errors": 130.0, + "median_solve_s": 0.4263105500003803, + "mean_solve_s": 0.46184209500002, + "p90_solve_s": 0.5754263200000427, + "max_solve_s": 0.8416353000002346, + "median_reconstruction_s": 0.9181746500003101, + "mean_reconstruction_s": 1.0667479599999752, + "p90_reconstruction_s": 1.6177789999994276, + "max_reconstruction_s": 1.9233327000001736, + "median_eval_s": 0.9863654499995391, + "mean_eval_s": 1.1374379499999576, + "p90_eval_s": 1.700894559999506, + "max_eval_s": 2.027672700000039 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.07290076335877863, + "mean_warm_start_error_fraction": 0.07309160305343511, + "p90_warm_start_error_fraction": 0.08793893129770992, + "max_warm_start_error_fraction": 0.09312977099236641, + "median_outer_iterations": 72.5, + "mean_outer_iterations": 70.6, + "p90_outer_iterations": 95.0, + "max_outer_iterations": 97.0, + "median_inner_iterations": 15.5, + "mean_inner_iterations": 16.35, + "p90_inner_iterations": 22.1, + "max_inner_iterations": 31.0, + "median_total_iterations": 85.0, + "mean_total_iterations": 86.95, + "p90_total_iterations": 111.30000000000001, + "max_total_iterations": 117.0, + "median_n_passive": 1171.0, + "mean_n_passive": 1170.35, + "p90_n_passive": 1194.6, + "max_n_passive": 1206.0, + "median_warm_start_errors": 95.5, + "mean_warm_start_errors": 95.75, + "p90_warm_start_errors": 115.2, + "max_warm_start_errors": 122.0, + "median_solve_s": 0.2782363500000429, + "mean_solve_s": 0.2902334850000443, + "p90_solve_s": 0.34917846999969715, + "max_solve_s": 0.48234279999996943, + "median_reconstruction_s": 0.7347446499998114, + "mean_reconstruction_s": 0.785360854999999, + "p90_reconstruction_s": 0.905746619999991, + "max_reconstruction_s": 1.2550971999999092, + "median_eval_s": 0.8073192500000914, + "mean_eval_s": 0.8536996150000504, + "p90_eval_s": 0.9941023000003044, + "max_eval_s": 1.3383673999996972 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 7.819192497728983e-15, + "max_abs_reconstruction_deviation": 2.1014203710478796e-09 + }, + "iteration_reduction_ratio": 1.088235294117647, + "solve_speedup_ratio": 1.532188551209483, + "eval_speedup_ratio": 1.2217786829676456, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.09312977099236641, + "reconstruction_s": 0.9163434000001871, + "eval_s": 0.9735986000005141, + "log_likelihood": 5474.7818880446775, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3876057999996192, + "n": 1310, + "solve_s": 0.3876057999996192, + "outer_iterations": 80, + "inner_iterations": 14, + "total_iterations": 94, + "n_passive": 1147, + "warm_start_errors": 122, + "index": 0 + }, + { + "warm_start_error_fraction": 0.083206106870229, + "reconstruction_s": 0.9018866999995225, + "eval_s": 0.9674087000003055, + "log_likelihood": 6651.6017674647865, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37842689999979484, + "n": 1310, + "solve_s": 0.37842689999979484, + "outer_iterations": 76, + "inner_iterations": 10, + "total_iterations": 86, + "n_passive": 1175, + "warm_start_errors": 109, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.8044630000003963, + "eval_s": 0.8678552000001218, + "log_likelihood": 6959.465181308695, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3088954000004378, + "n": 1310, + "solve_s": 0.3088954000004378, + "outer_iterations": 53, + "inner_iterations": 15, + "total_iterations": 68, + "n_passive": 1182, + "warm_start_errors": 72, + "index": 2 + }, + { + "warm_start_error_fraction": 0.08854961832061069, + "reconstruction_s": 0.8878968999997596, + "eval_s": 0.943561399999453, + "log_likelihood": 5649.68519400597, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3818013999998584, + "n": 1310, + "solve_s": 0.3818013999998584, + "outer_iterations": 80, + "inner_iterations": 20, + "total_iterations": 100, + "n_passive": 1155, + "warm_start_errors": 116, + "index": 3 + }, + { + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 0.8813823999998931, + "eval_s": 0.9418078000007881, + "log_likelihood": 5641.747255878448, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42239490000065416, + "n": 1310, + "solve_s": 0.42239490000065416, + "outer_iterations": 96, + "inner_iterations": 20, + "total_iterations": 116, + "n_passive": 1155, + "warm_start_errors": 115, + "index": 4 + }, + { + "warm_start_error_fraction": 0.07251908396946564, + "reconstruction_s": 0.8243379000005007, + "eval_s": 0.8866846999999325, + "log_likelihood": 6616.036284677204, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35844820000056643, + "n": 1310, + "solve_s": 0.35844820000056643, + "outer_iterations": 78, + "inner_iterations": 17, + "total_iterations": 95, + "n_passive": 1188, + "warm_start_errors": 95, + "index": 5 + }, + { + "warm_start_error_fraction": 0.08549618320610687, + "reconstruction_s": 0.9155651000000944, + "eval_s": 0.9741260000000693, + "log_likelihood": 5162.318835242547, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43089450000024954, + "n": 1310, + "solve_s": 0.43089450000024954, + "outer_iterations": 79, + "inner_iterations": 11, + "total_iterations": 90, + "n_passive": 1140, + "warm_start_errors": 112, + "index": 6 + }, + { + "warm_start_error_fraction": 0.08091603053435115, + "reconstruction_s": 0.8379791999996087, + "eval_s": 0.894470100000035, + "log_likelihood": 6601.171141456004, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3794579999994312, + "n": 1310, + "solve_s": 0.3794579999994312, + "outer_iterations": 73, + "inner_iterations": 18, + "total_iterations": 91, + "n_passive": 1162, + "warm_start_errors": 106, + "index": 7 + }, + { + "warm_start_error_fraction": 0.07709923664122137, + "reconstruction_s": 0.8063007000000653, + "eval_s": 0.8648233000003529, + "log_likelihood": 6459.009966554709, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34024090000002616, + "n": 1310, + "solve_s": 0.34024090000002616, + "outer_iterations": 76, + "inner_iterations": 15, + "total_iterations": 91, + "n_passive": 1177, + "warm_start_errors": 101, + "index": 8 + }, + { + "warm_start_error_fraction": 0.08931297709923663, + "reconstruction_s": 1.0486556000005294, + "eval_s": 1.1277506000005815, + "log_likelihood": 6015.149303135446, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5300391000000673, + "n": 1310, + "solve_s": 0.5300391000000673, + "outer_iterations": 81, + "inner_iterations": 14, + "total_iterations": 95, + "n_passive": 1159, + "warm_start_errors": 117, + "index": 9 + }, + { + "warm_start_error_fraction": 0.07099236641221374, + "reconstruction_s": 1.1339800999994623, + "eval_s": 1.202025899999171, + "log_likelihood": 6561.970298321282, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4302262000001065, + "n": 1310, + "solve_s": 0.4302262000001065, + "outer_iterations": 69, + "inner_iterations": 15, + "total_iterations": 84, + "n_passive": 1178, + "warm_start_errors": 93, + "index": 10 + }, + { + "warm_start_error_fraction": 0.07022900763358779, + "reconstruction_s": 1.0664354000000458, + "eval_s": 1.1330802999991647, + "log_likelihood": 6663.5232188930695, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5035939999997936, + "n": 1310, + "solve_s": 0.5035939999997936, + "outer_iterations": 109, + "inner_iterations": 20, + "total_iterations": 129, + "n_passive": 1185, + "warm_start_errors": 92, + "index": 11 + }, + { + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 0.920005900000433, + "eval_s": 0.9856776999995418, + "log_likelihood": 6974.799510194029, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30216569999993226, + "n": 1310, + "solve_s": 0.30216569999993226, + "outer_iterations": 51, + "inner_iterations": 15, + "total_iterations": 66, + "n_passive": 1195, + "warm_start_errors": 71, + "index": 12 + }, + { + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 0.89457289999973, + "eval_s": 0.9870531999995364, + "log_likelihood": 7124.350389709652, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4150755000000572, + "n": 1310, + "solve_s": 0.4150755000000572, + "outer_iterations": 64, + "inner_iterations": 14, + "total_iterations": 78, + "n_passive": 1206, + "warm_start_errors": 88, + "index": 13 + }, + { + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 1.7131772000002456, + "eval_s": 1.8397264000004725, + "log_likelihood": 6065.174878849797, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8416353000002346, + "n": 1310, + "solve_s": 0.8416353000002346, + "outer_iterations": 83, + "inner_iterations": 19, + "total_iterations": 102, + "n_passive": 1157, + "warm_start_errors": 115, + "index": 14 + }, + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 1.9233327000001736, + "eval_s": 2.027672700000039, + "log_likelihood": 6953.282640800198, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7773028000001432, + "n": 1310, + "solve_s": 0.7773028000001432, + "outer_iterations": 65, + "inner_iterations": 18, + "total_iterations": 83, + "n_passive": 1201, + "warm_start_errors": 77, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06564885496183206, + "reconstruction_s": 1.6071791999993366, + "eval_s": 1.6854687999993985, + "log_likelihood": 7211.572234127649, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.47966989999986254, + "n": 1310, + "solve_s": 0.47966989999986254, + "outer_iterations": 63, + "inner_iterations": 9, + "total_iterations": 72, + "n_passive": 1191, + "warm_start_errors": 86, + "index": 16 + }, + { + "warm_start_error_fraction": 0.09923664122137404, + "reconstruction_s": 1.1477849000002607, + "eval_s": 1.2126296000005823, + "log_likelihood": 4956.342487811883, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5031374999998661, + "n": 1310, + "solve_s": 0.5031374999998661, + "outer_iterations": 100, + "inner_iterations": 11, + "total_iterations": 111, + "n_passive": 1146, + "warm_start_errors": 130, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0916030534351145, + "reconstruction_s": 1.0055771999996068, + "eval_s": 1.0670002999995631, + "log_likelihood": 5693.019363548328, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5128342999996676, + "n": 1310, + "solve_s": 0.5128342999996676, + "outer_iterations": 100, + "inner_iterations": 21, + "total_iterations": 121, + "n_passive": 1167, + "warm_start_errors": 120, + "index": 18 + }, + { + "warm_start_error_fraction": 0.08396946564885496, + "reconstruction_s": 1.0981027999996513, + "eval_s": 1.1663376999995307, + "log_likelihood": 5762.328830101757, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5529956000000311, + "n": 1310, + "solve_s": 0.5529956000000311, + "outer_iterations": 77, + "inner_iterations": 18, + "total_iterations": 95, + "n_passive": 1148, + "warm_start_errors": 110, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.09312977099236641, + "reconstruction_s": 1.0213949999997567, + "eval_s": 1.0878535000001648, + "log_likelihood": 5474.7818880446775, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4635790000002089, + "n": 1310, + "solve_s": 0.4635790000002089, + "outer_iterations": 80, + "inner_iterations": 14, + "total_iterations": 94, + "n_passive": 1147, + "warm_start_errors": 122, + "index": 0 + }, + { + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 0.7439722999997684, + "eval_s": 0.803247800000463, + "log_likelihood": 6651.601767464806, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30111459999989165, + "n": 1310, + "solve_s": 0.30111459999989165, + "outer_iterations": 74, + "inner_iterations": 13, + "total_iterations": 87, + "n_passive": 1175, + "warm_start_errors": 88, + "index": 1 + }, + { + "warm_start_error_fraction": 0.06946564885496183, + "reconstruction_s": 0.7250944000006712, + "eval_s": 0.7858264000005875, + "log_likelihood": 6959.465181308719, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29671719999987545, + "n": 1310, + "solve_s": 0.29671719999987545, + "outer_iterations": 77, + "inner_iterations": 31, + "total_iterations": 108, + "n_passive": 1182, + "warm_start_errors": 91, + "index": 2 + }, + { + "warm_start_error_fraction": 0.07938931297709924, + "reconstruction_s": 0.7006664000000455, + "eval_s": 0.7663747999995394, + "log_likelihood": 5649.685194005982, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2577245000002222, + "n": 1310, + "solve_s": 0.2577245000002222, + "outer_iterations": 52, + "inner_iterations": 21, + "total_iterations": 73, + "n_passive": 1154, + "warm_start_errors": 104, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0633587786259542, + "reconstruction_s": 0.7345968999998149, + "eval_s": 0.8067301000000953, + "log_likelihood": 5641.747255878469, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26480780000019877, + "n": 1310, + "solve_s": 0.26480780000019877, + "outer_iterations": 57, + "inner_iterations": 13, + "total_iterations": 70, + "n_passive": 1155, + "warm_start_errors": 83, + "index": 4 + }, + { + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 1.2550971999999092, + "eval_s": 1.3383673999996972, + "log_likelihood": 6616.036284677211, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.48234279999996943, + "n": 1310, + "solve_s": 0.48234279999996943, + "outer_iterations": 97, + "inner_iterations": 14, + "total_iterations": 111, + "n_passive": 1188, + "warm_start_errors": 115, + "index": 5 + }, + { + "warm_start_error_fraction": 0.07633587786259542, + "reconstruction_s": 0.8928968000000168, + "eval_s": 0.9836855000003197, + "log_likelihood": 5162.318835242547, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3212143999999171, + "n": 1310, + "solve_s": 0.3212143999999171, + "outer_iterations": 53, + "inner_iterations": 9, + "total_iterations": 62, + "n_passive": 1140, + "warm_start_errors": 100, + "index": 6 + }, + { + "warm_start_error_fraction": 0.08015267175572519, + "reconstruction_s": 0.8630586999997831, + "eval_s": 0.9347858000000997, + "log_likelihood": 6601.171141456006, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3324891000002026, + "n": 1310, + "solve_s": 0.3324891000002026, + "outer_iterations": 81, + "inner_iterations": 21, + "total_iterations": 102, + "n_passive": 1161, + "warm_start_errors": 105, + "index": 7 + }, + { + "warm_start_error_fraction": 0.08244274809160305, + "reconstruction_s": 0.8246870000002673, + "eval_s": 0.8878218999998353, + "log_likelihood": 6459.009966554712, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3364672999996401, + "n": 1310, + "solve_s": 0.3364672999996401, + "outer_iterations": 95, + "inner_iterations": 19, + "total_iterations": 114, + "n_passive": 1177, + "warm_start_errors": 108, + "index": 8 + }, + { + "warm_start_error_fraction": 0.08091603053435115, + "reconstruction_s": 0.7265684000003603, + "eval_s": 0.8079084000000876, + "log_likelihood": 6015.149303135448, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29060759999993024, + "n": 1310, + "solve_s": 0.29060759999993024, + "outer_iterations": 65, + "inner_iterations": 18, + "total_iterations": 83, + "n_passive": 1159, + "warm_start_errors": 106, + "index": 9 + }, + { + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.7214671000001545, + "eval_s": 0.7795159000006606, + "log_likelihood": 6561.970298321256, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21705759999986185, + "n": 1310, + "solve_s": 0.21705759999986185, + "outer_iterations": 62, + "inner_iterations": 11, + "total_iterations": 73, + "n_passive": 1177, + "warm_start_errors": 74, + "index": 10 + }, + { + "warm_start_error_fraction": 0.07404580152671755, + "reconstruction_s": 0.7706705000000511, + "eval_s": 0.8342887999997401, + "log_likelihood": 6663.5232188930695, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29206729999987147, + "n": 1310, + "solve_s": 0.29206729999987147, + "outer_iterations": 82, + "inner_iterations": 18, + "total_iterations": 100, + "n_passive": 1184, + "warm_start_errors": 97, + "index": 11 + }, + { + "warm_start_error_fraction": 0.05801526717557252, + "reconstruction_s": 0.738101099999767, + "eval_s": 0.8141115999997055, + "log_likelihood": 6974.79951019406, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25415090000024065, + "n": 1310, + "solve_s": 0.25415090000024065, + "outer_iterations": 72, + "inner_iterations": 10, + "total_iterations": 82, + "n_passive": 1194, + "warm_start_errors": 76, + "index": 12 + }, + { + "warm_start_error_fraction": 0.059541984732824425, + "reconstruction_s": 0.7300915999994686, + "eval_s": 0.7924017000004824, + "log_likelihood": 7124.35038970967, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23140439999951923, + "n": 1310, + "solve_s": 0.23140439999951923, + "outer_iterations": 60, + "inner_iterations": 15, + "total_iterations": 75, + "n_passive": 1206, + "warm_start_errors": 78, + "index": 13 + }, + { + "warm_start_error_fraction": 0.08931297709923663, + "reconstruction_s": 0.7140065000003233, + "eval_s": 0.779603799999677, + "log_likelihood": 6065.174878849797, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2658651000001555, + "n": 1310, + "solve_s": 0.2658651000001555, + "outer_iterations": 73, + "inner_iterations": 23, + "total_iterations": 96, + "n_passive": 1157, + "warm_start_errors": 117, + "index": 14 + }, + { + "warm_start_error_fraction": 0.083206106870229, + "reconstruction_s": 0.8127936000000773, + "eval_s": 0.8780507000001307, + "log_likelihood": 6953.282640800189, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3262697999998636, + "n": 1310, + "solve_s": 0.3262697999998636, + "outer_iterations": 95, + "inner_iterations": 22, + "total_iterations": 117, + "n_passive": 1200, + "warm_start_errors": 109, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06259541984732825, + "reconstruction_s": 0.7348923999998078, + "eval_s": 0.811990600000172, + "log_likelihood": 7211.572234127592, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24470000000019354, + "n": 1310, + "solve_s": 0.24470000000019354, + "outer_iterations": 62, + "inner_iterations": 15, + "total_iterations": 77, + "n_passive": 1190, + "warm_start_errors": 82, + "index": 16 + }, + { + "warm_start_error_fraction": 0.0717557251908397, + "reconstruction_s": 0.6959763999993811, + "eval_s": 0.7566652999994403, + "log_likelihood": 4956.342487811893, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2147398000006433, + "n": 1310, + "solve_s": 0.2147398000006433, + "outer_iterations": 61, + "inner_iterations": 16, + "total_iterations": 77, + "n_passive": 1146, + "warm_start_errors": 94, + "index": 17 + }, + { + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 0.6996988999999303, + "eval_s": 0.7585767999999007, + "log_likelihood": 5693.019363548336, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25837490000049, + "n": 1310, + "solve_s": 0.25837490000049, + "outer_iterations": 78, + "inner_iterations": 17, + "total_iterations": 95, + "n_passive": 1167, + "warm_start_errors": 87, + "index": 18 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 0.6014859000006254, + "eval_s": 0.6661855000002106, + "log_likelihood": 5762.328830101752, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.15297559999999066, + "n": 1310, + "solve_s": 0.15297559999999066, + "outer_iterations": 36, + "inner_iterations": 7, + "total_iterations": 43, + "n_passive": 1148, + "warm_start_errors": 79, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_powerlaw_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_powerlaw_v2026.8.17.1.png new file mode 100644 index 000000000..b6abad972 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_powerlaw_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_rectangular_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_rectangular_v2026.8.17.1.json new file mode 100644 index 000000000..fd4fd206b --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_rectangular_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "rectangular", + "configuration": { + "model": "rectangular", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "rectangular_uniform_35x35", + "regularization": "constant", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 3.58859239999947, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.4412532637075718, + "mean_warm_start_error_fraction": 0.4352915578764144, + "p90_warm_start_error_fraction": 0.44386422976501305, + "max_warm_start_error_fraction": 0.44473455178416016, + "median_outer_iterations": 560.0, + "mean_outer_iterations": 555.65, + "p90_outer_iterations": 569.2, + "max_outer_iterations": 586.0, + "median_inner_iterations": 39.5, + "mean_inner_iterations": 39.8, + "p90_inner_iterations": 43.1, + "max_inner_iterations": 47.0, + "median_total_iterations": 598.5, + "mean_total_iterations": 595.45, + "p90_total_iterations": 612.6, + "max_total_iterations": 630.0, + "median_n_passive": 694.0, + "mean_n_passive": 695.6, + "p90_n_passive": 698.0, + "max_n_passive": 716.0, + "median_warm_start_errors": 507.0, + "mean_warm_start_errors": 500.15, + "p90_warm_start_errors": 510.0, + "max_warm_start_errors": 511.0, + "median_solve_s": 0.7269897999999557, + "mean_solve_s": 0.7681787799998346, + "p90_solve_s": 1.048063690000072, + "max_solve_s": 1.210635800000091, + "median_reconstruction_s": 1.2947465500001272, + "mean_reconstruction_s": 1.349882050000042, + "p90_reconstruction_s": 1.651244040000529, + "max_reconstruction_s": 1.840900099999999, + "median_eval_s": 1.366945050000595, + "mean_eval_s": 1.4305006300000969, + "p90_eval_s": 1.7369451599997774, + "max_eval_s": 1.9378489000000627 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.009573542210617928, + "mean_warm_start_error_fraction": 0.03237597911227154, + "p90_warm_start_error_fraction": 0.024717145343777204, + "max_warm_start_error_fraction": 0.4360313315926893, + "median_outer_iterations": 6.0, + "mean_outer_iterations": 36.45, + "p90_outer_iterations": 32.1, + "max_outer_iterations": 564.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 3.9, + "p90_inner_iterations": 7.100000000000001, + "max_inner_iterations": 38.0, + "median_total_iterations": 7.0, + "mean_total_iterations": 40.35, + "p90_total_iterations": 40.0, + "max_total_iterations": 602.0, + "median_n_passive": 694.0, + "mean_n_passive": 695.35, + "p90_n_passive": 698.0, + "max_n_passive": 716.0, + "median_warm_start_errors": 11.0, + "mean_warm_start_errors": 37.2, + "p90_warm_start_errors": 28.400000000000006, + "max_warm_start_errors": 501.0, + "median_solve_s": 0.0352649499996005, + "mean_solve_s": 0.0907001799999307, + "p90_solve_s": 0.09787602000051267, + "max_solve_s": 0.98008530000061, + "median_reconstruction_s": 0.6218995000003815, + "mean_reconstruction_s": 0.6672516950001409, + "p90_reconstruction_s": 0.7125608700001067, + "max_reconstruction_s": 1.6924641000005067, + "median_eval_s": 0.7051923999997598, + "mean_eval_s": 0.7615823799998452, + "p90_eval_s": 0.8495136799993817, + "max_eval_s": 1.7748816999992414 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.2467229493127918e-15, + "max_abs_reconstruction_deviation": 6.691056597674105e-10 + }, + "iteration_reduction_ratio": 85.5, + "solve_speedup_ratio": 20.615080980072037, + "eval_speedup_ratio": 1.938400144415993, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.4360313315926893, + "reconstruction_s": 1.03695450000032, + "eval_s": 1.0972437000000355, + "log_likelihood": 5903.965466512353, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6151569999992716, + "n": 1149, + "solve_s": 0.6151569999992716, + "outer_iterations": 564, + "inner_iterations": 38, + "total_iterations": 602, + "n_passive": 697, + "warm_start_errors": 501, + "index": 0 + }, + { + "warm_start_error_fraction": 0.4386422976501306, + "reconstruction_s": 1.3128452999999354, + "eval_s": 1.3716584000003422, + "log_likelihood": 5873.066646242587, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6615018000002237, + "n": 1149, + "solve_s": 0.6615018000002237, + "outer_iterations": 565, + "inner_iterations": 43, + "total_iterations": 608, + "n_passive": 692, + "warm_start_errors": 504, + "index": 1 + }, + { + "warm_start_error_fraction": 0.4412532637075718, + "reconstruction_s": 1.1194185000003927, + "eval_s": 1.1873185000004014, + "log_likelihood": 5835.883954470541, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6384078999999474, + "n": 1149, + "solve_s": 0.6384078999999474, + "outer_iterations": 571, + "inner_iterations": 47, + "total_iterations": 618, + "n_passive": 694, + "warm_start_errors": 507, + "index": 2 + }, + { + "warm_start_error_fraction": 0.4421235857267189, + "reconstruction_s": 1.6213349999998172, + "eval_s": 1.7025045000000318, + "log_likelihood": 5882.697983982538, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0463957000001756, + "n": 1149, + "solve_s": 1.0463957000001756, + "outer_iterations": 566, + "inner_iterations": 40, + "total_iterations": 606, + "n_passive": 698, + "warm_start_errors": 508, + "index": 3 + }, + { + "warm_start_error_fraction": 0.44299390774586594, + "reconstruction_s": 1.840900099999999, + "eval_s": 1.9378489000000627, + "log_likelihood": 5883.145650134455, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.210635800000091, + "n": 1149, + "solve_s": 1.210635800000091, + "outer_iterations": 586, + "inner_iterations": 44, + "total_iterations": 630, + "n_passive": 698, + "warm_start_errors": 509, + "index": 4 + }, + { + "warm_start_error_fraction": 0.4412532637075718, + "reconstruction_s": 1.3836903999999777, + "eval_s": 1.4571216000003915, + "log_likelihood": 5836.0661590404015, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8821931999991648, + "n": 1149, + "solve_s": 0.8821931999991648, + "outer_iterations": 525, + "inner_iterations": 39, + "total_iterations": 564, + "n_passive": 694, + "warm_start_errors": 507, + "index": 5 + }, + { + "warm_start_error_fraction": 0.43777197563098347, + "reconstruction_s": 1.242889600000126, + "eval_s": 1.3437268999996377, + "log_likelihood": 5893.619666630966, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6323511999999027, + "n": 1149, + "solve_s": 0.6323511999999027, + "outer_iterations": 512, + "inner_iterations": 36, + "total_iterations": 548, + "n_passive": 697, + "warm_start_errors": 503, + "index": 6 + }, + { + "warm_start_error_fraction": 0.4281984334203655, + "reconstruction_s": 1.648718500000541, + "eval_s": 1.732270399999834, + "log_likelihood": 5917.666489099418, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.063075599999138, + "n": 1149, + "solve_s": 1.063075599999138, + "outer_iterations": 558, + "inner_iterations": 41, + "total_iterations": 599, + "n_passive": 694, + "warm_start_errors": 492, + "index": 7 + }, + { + "warm_start_error_fraction": 0.41601392515230634, + "reconstruction_s": 1.6739739000004192, + "eval_s": 1.7790179999992688, + "log_likelihood": 5930.193628581959, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8857410999999047, + "n": 1149, + "solve_s": 0.8857410999999047, + "outer_iterations": 552, + "inner_iterations": 41, + "total_iterations": 593, + "n_passive": 696, + "warm_start_errors": 478, + "index": 8 + }, + { + "warm_start_error_fraction": 0.44386422976501305, + "reconstruction_s": 1.636532100000295, + "eval_s": 1.7024432000007437, + "log_likelihood": 5852.094253598394, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8801760999995167, + "n": 1149, + "solve_s": 0.8801760999995167, + "outer_iterations": 559, + "inner_iterations": 38, + "total_iterations": 597, + "n_passive": 694, + "warm_start_errors": 510, + "index": 9 + }, + { + "warm_start_error_fraction": 0.44299390774586594, + "reconstruction_s": 1.2055225999993127, + "eval_s": 1.2815150000005815, + "log_likelihood": 5833.121448837093, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6878804999996646, + "n": 1149, + "solve_s": 0.6878804999996646, + "outer_iterations": 524, + "inner_iterations": 33, + "total_iterations": 557, + "n_passive": 694, + "warm_start_errors": 509, + "index": 10 + }, + { + "warm_start_error_fraction": 0.44299390774586594, + "reconstruction_s": 1.0674134000000777, + "eval_s": 1.1299315999995088, + "log_likelihood": 5831.220857042912, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.554019000000153, + "n": 1149, + "solve_s": 0.554019000000153, + "outer_iterations": 566, + "inner_iterations": 43, + "total_iterations": 609, + "n_passive": 692, + "warm_start_errors": 509, + "index": 11 + }, + { + "warm_start_error_fraction": 0.4421235857267189, + "reconstruction_s": 1.228240200000073, + "eval_s": 1.3124932000000626, + "log_likelihood": 5822.522100957884, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8009770999997272, + "n": 1149, + "solve_s": 0.8009770999997272, + "outer_iterations": 559, + "inner_iterations": 38, + "total_iterations": 597, + "n_passive": 691, + "warm_start_errors": 508, + "index": 12 + }, + { + "warm_start_error_fraction": 0.44386422976501305, + "reconstruction_s": 1.2558060999999725, + "eval_s": 1.3309853000000658, + "log_likelihood": 5835.50189744867, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6064952999995512, + "n": 1149, + "solve_s": 0.6064952999995512, + "outer_iterations": 565, + "inner_iterations": 42, + "total_iterations": 607, + "n_passive": 693, + "warm_start_errors": 510, + "index": 13 + }, + { + "warm_start_error_fraction": 0.44473455178416016, + "reconstruction_s": 1.1818458999996437, + "eval_s": 1.2639940999997634, + "log_likelihood": 5811.408577960397, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6430577000000994, + "n": 1149, + "solve_s": 0.6430577000000994, + "outer_iterations": 561, + "inner_iterations": 37, + "total_iterations": 598, + "n_passive": 693, + "warm_start_errors": 511, + "index": 14 + }, + { + "warm_start_error_fraction": 0.4386422976501306, + "reconstruction_s": 1.276647800000319, + "eval_s": 1.362231700000848, + "log_likelihood": 5827.525400179178, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.745940800000426, + "n": 1149, + "solve_s": 0.745940800000426, + "outer_iterations": 564, + "inner_iterations": 39, + "total_iterations": 603, + "n_passive": 693, + "warm_start_errors": 504, + "index": 15 + }, + { + "warm_start_error_fraction": 0.4421235857267189, + "reconstruction_s": 1.406645699999899, + "eval_s": 1.4855167000005167, + "log_likelihood": 5874.44458244485, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7257933999999295, + "n": 1149, + "solve_s": 0.7257933999999295, + "outer_iterations": 569, + "inner_iterations": 43, + "total_iterations": 612, + "n_passive": 698, + "warm_start_errors": 508, + "index": 16 + }, + { + "warm_start_error_fraction": 0.43342036553524804, + "reconstruction_s": 1.1664559999999256, + "eval_s": 1.2535177999998268, + "log_likelihood": 5916.488069780295, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.621795099999872, + "n": 1149, + "solve_s": 0.621795099999872, + "outer_iterations": 555, + "inner_iterations": 39, + "total_iterations": 594, + "n_passive": 693, + "warm_start_errors": 498, + "index": 17 + }, + { + "warm_start_error_fraction": 0.41601392515230634, + "reconstruction_s": 1.3463624999994863, + "eval_s": 1.448651400000017, + "log_likelihood": 5942.627742848432, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7337950999999521, + "n": 1149, + "solve_s": 0.7337950999999521, + "outer_iterations": 541, + "inner_iterations": 34, + "total_iterations": 575, + "n_passive": 695, + "warm_start_errors": 478, + "index": 18 + }, + { + "warm_start_error_fraction": 0.3907745865970409, + "reconstruction_s": 1.3454429000003074, + "eval_s": 1.4300216999999975, + "log_likelihood": 6043.775381526144, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7281861999999819, + "n": 1149, + "solve_s": 0.7281861999999819, + "outer_iterations": 551, + "inner_iterations": 41, + "total_iterations": 592, + "n_passive": 716, + "warm_start_errors": 449, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.4360313315926893, + "reconstruction_s": 1.6924641000005067, + "eval_s": 1.7748816999992414, + "log_likelihood": 5903.965466512353, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.98008530000061, + "n": 1149, + "solve_s": 0.98008530000061, + "outer_iterations": 564, + "inner_iterations": 38, + "total_iterations": 602, + "n_passive": 697, + "warm_start_errors": 501, + "index": 0 + }, + { + "warm_start_error_fraction": 0.011314186248912098, + "reconstruction_s": 0.5845745000005991, + "eval_s": 0.6645078999999896, + "log_likelihood": 5873.066646242589, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04811689999951341, + "n": 1149, + "solve_s": 0.04811689999951341, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 692, + "warm_start_errors": 13, + "index": 1 + }, + { + "warm_start_error_fraction": 0.009573542210617928, + "reconstruction_s": 0.6687105999999403, + "eval_s": 0.7625335000002451, + "log_likelihood": 5835.883954470544, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.047077100000024075, + "n": 1149, + "solve_s": 0.047077100000024075, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 693, + "warm_start_errors": 11, + "index": 2 + }, + { + "warm_start_error_fraction": 0.009573542210617928, + "reconstruction_s": 0.5920022000000245, + "eval_s": 0.672594300000128, + "log_likelihood": 5882.69798398254, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.034980999999788764, + "n": 1149, + "solve_s": 0.034980999999788764, + "outer_iterations": 8, + "inner_iterations": 1, + "total_iterations": 9, + "n_passive": 698, + "warm_start_errors": 11, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0, + "reconstruction_s": 0.5084507000001395, + "eval_s": 0.617801800000052, + "log_likelihood": 5883.145650134458, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.029417700000522018, + "n": 1149, + "solve_s": 0.029417700000522018, + "outer_iterations": 0, + "inner_iterations": 0, + "total_iterations": 0, + "n_passive": 698, + "warm_start_errors": 0, + "index": 4 + }, + { + "warm_start_error_fraction": 0.012184508268059183, + "reconstruction_s": 0.7373286000001826, + "eval_s": 0.8805437000000893, + "log_likelihood": 5836.066159040409, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.050321400000029826, + "n": 1149, + "solve_s": 0.050321400000029826, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 694, + "warm_start_errors": 14, + "index": 5 + }, + { + "warm_start_error_fraction": 0.009573542210617928, + "reconstruction_s": 0.6446069000003263, + "eval_s": 0.7320697999994081, + "log_likelihood": 5893.619666630968, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04089699999985896, + "n": 1149, + "solve_s": 0.04089699999985896, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 697, + "warm_start_errors": 11, + "index": 6 + }, + { + "warm_start_error_fraction": 0.016536118363794605, + "reconstruction_s": 0.6244747999999163, + "eval_s": 0.7028310999994574, + "log_likelihood": 5917.666489099423, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04456879999997909, + "n": 1149, + "solve_s": 0.04456879999997909, + "outer_iterations": 11, + "inner_iterations": 3, + "total_iterations": 14, + "n_passive": 694, + "warm_start_errors": 19, + "index": 7 + }, + { + "warm_start_error_fraction": 0.005221932114882507, + "reconstruction_s": 0.5844984000004843, + "eval_s": 0.7027834999998959, + "log_likelihood": 5930.193628581961, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.025411499999790976, + "n": 1149, + "solve_s": 0.025411499999790976, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 696, + "warm_start_errors": 6, + "index": 8 + }, + { + "warm_start_error_fraction": 0.008703220191470844, + "reconstruction_s": 0.6983493999996426, + "eval_s": 0.8460658999993029, + "log_likelihood": 5852.094253598401, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.027613499999461055, + "n": 1149, + "solve_s": 0.027613499999461055, + "outer_iterations": 4, + "inner_iterations": 2, + "total_iterations": 6, + "n_passive": 694, + "warm_start_errors": 10, + "index": 9 + }, + { + "warm_start_error_fraction": 0.008703220191470844, + "reconstruction_s": 0.6193242000008468, + "eval_s": 0.7075537000000622, + "log_likelihood": 5833.1214488370915, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.026467300000149407, + "n": 1149, + "solve_s": 0.026467300000149407, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 694, + "warm_start_errors": 10, + "index": 10 + }, + { + "warm_start_error_fraction": 0.008703220191470844, + "reconstruction_s": 0.5463164000002507, + "eval_s": 0.6229407999999239, + "log_likelihood": 5831.220857042908, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03190919999997277, + "n": 1149, + "solve_s": 0.03190919999997277, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 692, + "warm_start_errors": 10, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0026109660574412533, + "reconstruction_s": 0.45773710000048595, + "eval_s": 0.5296349999998711, + "log_likelihood": 5822.522100957885, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.017956699999558623, + "n": 1149, + "solve_s": 0.017956699999558623, + "outer_iterations": 1, + "inner_iterations": 2, + "total_iterations": 3, + "n_passive": 691, + "warm_start_errors": 3, + "index": 12 + }, + { + "warm_start_error_fraction": 0.007832898172323759, + "reconstruction_s": 0.6307620999996288, + "eval_s": 0.7104393999998138, + "log_likelihood": 5835.50189744867, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.030975100000432576, + "n": 1149, + "solve_s": 0.030975100000432576, + "outer_iterations": 5, + "inner_iterations": 2, + "total_iterations": 7, + "n_passive": 692, + "warm_start_errors": 9, + "index": 13 + }, + { + "warm_start_error_fraction": 0.006962576153176675, + "reconstruction_s": 0.5452848000004451, + "eval_s": 0.6387636999997994, + "log_likelihood": 5811.408577960397, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.028387000000293483, + "n": 1149, + "solve_s": 0.028387000000293483, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 692, + "warm_start_errors": 8, + "index": 14 + }, + { + "warm_start_error_fraction": 0.011314186248912098, + "reconstruction_s": 0.5898686000000453, + "eval_s": 0.6688359999998283, + "log_likelihood": 5827.52540017918, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03554889999941224, + "n": 1149, + "solve_s": 0.03554889999941224, + "outer_iterations": 8, + "inner_iterations": 2, + "total_iterations": 10, + "n_passive": 693, + "warm_start_errors": 13, + "index": 15 + }, + { + "warm_start_error_fraction": 0.020887728459530026, + "reconstruction_s": 0.6715919999996913, + "eval_s": 0.7796510999996826, + "log_likelihood": 5874.444582444852, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08222099999966304, + "n": 1149, + "solve_s": 0.08222099999966304, + "outer_iterations": 16, + "inner_iterations": 4, + "total_iterations": 20, + "n_passive": 697, + "warm_start_errors": 24, + "index": 16 + }, + { + "warm_start_error_fraction": 0.024369016536118365, + "reconstruction_s": 0.7098089000000982, + "eval_s": 0.7970316999999341, + "log_likelihood": 5916.488069780289, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0972304000006261, + "n": 1149, + "solve_s": 0.0972304000006261, + "outer_iterations": 32, + "inner_iterations": 8, + "total_iterations": 40, + "n_passive": 693, + "warm_start_errors": 28, + "index": 17 + }, + { + "warm_start_error_fraction": 0.009573542210617928, + "reconstruction_s": 0.6426518999996915, + "eval_s": 0.7186912000006487, + "log_likelihood": 5942.627742848434, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03113119999943592, + "n": 1149, + "solve_s": 0.03113119999943592, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 694, + "warm_start_errors": 11, + "index": 18 + }, + { + "warm_start_error_fraction": 0.0278503046127067, + "reconstruction_s": 0.5962276999998721, + "eval_s": 0.7014917999995305, + "log_likelihood": 6043.7753815261385, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10368659999949159, + "n": 1149, + "solve_s": 0.10368659999949159, + "outer_iterations": 33, + "inner_iterations": 7, + "total_iterations": 40, + "n_passive": 716, + "warm_start_errors": 32, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.43646649260226283, + "mean_warm_start_error_fraction": 0.42188859878154916, + "p90_warm_start_error_fraction": 0.48172323759791125, + "max_warm_start_error_fraction": 0.5004351610095735, + "median_outer_iterations": 564.5, + "mean_outer_iterations": 531.7, + "p90_outer_iterations": 605.7, + "max_outer_iterations": 635.0, + "median_inner_iterations": 45.5, + "mean_inner_iterations": 45.75, + "p90_inner_iterations": 59.2, + "max_inner_iterations": 71.0, + "median_total_iterations": 620.5, + "mean_total_iterations": 577.45, + "p90_total_iterations": 657.1, + "max_total_iterations": 677.0, + "median_n_passive": 665.5, + "mean_n_passive": 669.4, + "p90_n_passive": 713.5, + "max_n_passive": 723.0, + "median_warm_start_errors": 501.5, + "mean_warm_start_errors": 484.75, + "p90_warm_start_errors": 553.5, + "max_warm_start_errors": 575.0, + "median_solve_s": 0.5576974000000519, + "mean_solve_s": 0.5991660400000001, + "p90_solve_s": 0.7186198599997627, + "max_solve_s": 1.053808300000128, + "median_reconstruction_s": 1.0667899499999294, + "mean_reconstruction_s": 1.1486675199999354, + "p90_reconstruction_s": 1.3757922599992527, + "max_reconstruction_s": 1.8110564000007798, + "median_eval_s": 1.133310050000091, + "mean_eval_s": 1.2264488300000267, + "p90_eval_s": 1.476887669999542, + "max_eval_s": 1.9071390999997675 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.13794604003481287, + "mean_warm_start_error_fraction": 0.15004351610095737, + "p90_warm_start_error_fraction": 0.1575282854656223, + "max_warm_start_error_fraction": 0.4516971279373368, + "median_outer_iterations": 131.5, + "mean_outer_iterations": 147.0, + "p90_outer_iterations": 187.20000000000005, + "max_outer_iterations": 558.0, + "median_inner_iterations": 33.0, + "mean_inner_iterations": 46.5, + "p90_inner_iterations": 76.80000000000003, + "max_inner_iterations": 136.0, + "median_total_iterations": 161.0, + "mean_total_iterations": 193.5, + "p90_total_iterations": 243.20000000000005, + "max_total_iterations": 619.0, + "median_n_passive": 665.0, + "mean_n_passive": 669.1, + "p90_n_passive": 713.5, + "max_n_passive": 723.0, + "median_warm_start_errors": 158.5, + "mean_warm_start_errors": 172.4, + "p90_warm_start_errors": 181.00000000000003, + "max_warm_start_errors": 519.0, + "median_solve_s": 0.23297185000001264, + "mean_solve_s": 0.25406614999988053, + "p90_solve_s": 0.3431758300000184, + "max_solve_s": 0.48008690000006027, + "median_reconstruction_s": 0.7540745000001152, + "mean_reconstruction_s": 0.8073641549998684, + "p90_reconstruction_s": 1.0373129099997642, + "max_reconstruction_s": 1.2319340999993074, + "median_eval_s": 0.8491830999996637, + "mean_eval_s": 0.8859470299999884, + "p90_eval_s": 1.1288937100006478, + "max_eval_s": 1.3170918000005258 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 6.692554150368649e-15, + "max_abs_reconstruction_deviation": 6.378055440563191e-10 + }, + "iteration_reduction_ratio": 3.8540372670807455, + "solve_speedup_ratio": 2.393840285854371, + "eval_speedup_ratio": 1.3345885592877906, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.4516971279373368, + "reconstruction_s": 1.358727999999246, + "eval_s": 1.4587298999995255, + "log_likelihood": 4339.419124288473, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6151958000000377, + "n": 1149, + "solve_s": 0.6151958000000377, + "outer_iterations": 558, + "inner_iterations": 61, + "total_iterations": 619, + "n_passive": 610, + "warm_start_errors": 519, + "index": 0 + }, + { + "warm_start_error_fraction": 0.44473455178416016, + "reconstruction_s": 1.252172900000005, + "eval_s": 1.3369404000004579, + "log_likelihood": 5340.949647349872, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.620875800000249, + "n": 1149, + "solve_s": 0.620875800000249, + "outer_iterations": 592, + "inner_iterations": 53, + "total_iterations": 645, + "n_passive": 670, + "warm_start_errors": 511, + "index": 1 + }, + { + "warm_start_error_fraction": 0.46823324630113144, + "reconstruction_s": 1.5293705999993108, + "eval_s": 1.6403075999996872, + "log_likelihood": 5802.68836673616, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.915429700000459, + "n": 1149, + "solve_s": 0.915429700000459, + "outer_iterations": 572, + "inner_iterations": 52, + "total_iterations": 624, + "n_passive": 696, + "warm_start_errors": 538, + "index": 2 + }, + { + "warm_start_error_fraction": 0.4142732811140122, + "reconstruction_s": 1.2835704999997688, + "eval_s": 1.3458381000000372, + "log_likelihood": 5354.284629582961, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6720247999992353, + "n": 1149, + "solve_s": 0.6720247999992353, + "outer_iterations": 563, + "inner_iterations": 59, + "total_iterations": 622, + "n_passive": 656, + "warm_start_errors": 476, + "index": 3 + }, + { + "warm_start_error_fraction": 0.4934725848563969, + "reconstruction_s": 1.2436175999991974, + "eval_s": 1.3292693000003055, + "log_likelihood": 5635.0404539055025, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.696752099999685, + "n": 1149, + "solve_s": 0.696752099999685, + "outer_iterations": 612, + "inner_iterations": 43, + "total_iterations": 655, + "n_passive": 713, + "warm_start_errors": 567, + "index": 4 + }, + { + "warm_start_error_fraction": 0.40557006092254133, + "reconstruction_s": 1.2611749000006967, + "eval_s": 1.3625574999996388, + "log_likelihood": 5571.756602168999, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6481205000000045, + "n": 1149, + "solve_s": 0.6481205000000045, + "outer_iterations": 580, + "inner_iterations": 43, + "total_iterations": 623, + "n_passive": 662, + "warm_start_errors": 466, + "index": 5 + }, + { + "warm_start_error_fraction": 0.4412532637075718, + "reconstruction_s": 1.8110564000007798, + "eval_s": 1.9071390999997675, + "log_likelihood": 5534.492855134288, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.053808300000128, + "n": 1149, + "solve_s": 1.053808300000128, + "outer_iterations": 557, + "inner_iterations": 47, + "total_iterations": 604, + "n_passive": 666, + "warm_start_errors": 507, + "index": 6 + }, + { + "warm_start_error_fraction": 0.4351610095735422, + "reconstruction_s": 1.1741511999998693, + "eval_s": 1.260276299999532, + "log_likelihood": 4990.180171731866, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5990123000001404, + "n": 1149, + "solve_s": 0.5990123000001404, + "outer_iterations": 541, + "inner_iterations": 50, + "total_iterations": 591, + "n_passive": 616, + "warm_start_errors": 500, + "index": 7 + }, + { + "warm_start_error_fraction": 0.43777197563098347, + "reconstruction_s": 1.0135738999997557, + "eval_s": 1.0832430000000386, + "log_likelihood": 5087.004415913071, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.531184999999823, + "n": 1149, + "solve_s": 0.531184999999823, + "outer_iterations": 597, + "inner_iterations": 34, + "total_iterations": 631, + "n_passive": 678, + "warm_start_errors": 503, + "index": 8 + }, + { + "warm_start_error_fraction": 0.5004351610095735, + "reconstruction_s": 1.0539430000007997, + "eval_s": 1.1211963000005198, + "log_likelihood": 4069.586330340285, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.521475399999872, + "n": 1149, + "solve_s": 0.521475399999872, + "outer_iterations": 605, + "inner_iterations": 71, + "total_iterations": 676, + "n_passive": 639, + "warm_start_errors": 575, + "index": 9 + }, + { + "warm_start_error_fraction": 0.40034812880765885, + "reconstruction_s": 0.9615143999999418, + "eval_s": 1.0281511000002865, + "log_likelihood": 5540.6586942650365, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4072161000003689, + "n": 1149, + "solve_s": 0.4072161000003689, + "outer_iterations": 477, + "inner_iterations": 36, + "total_iterations": 513, + "n_passive": 650, + "warm_start_errors": 460, + "index": 10 + }, + { + "warm_start_error_fraction": 0.4577893820713664, + "reconstruction_s": 0.9731894000005923, + "eval_s": 1.040279300000293, + "log_likelihood": 4408.075213771162, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4849402999998347, + "n": 1149, + "solve_s": 0.4849402999998347, + "outer_iterations": 592, + "inner_iterations": 55, + "total_iterations": 647, + "n_passive": 640, + "warm_start_errors": 526, + "index": 11 + }, + { + "warm_start_error_fraction": 0.4804177545691906, + "reconstruction_s": 1.0713881000001493, + "eval_s": 1.1432490999995935, + "log_likelihood": 5236.43649574859, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5556519000001572, + "n": 1149, + "solve_s": 0.5556519000001572, + "outer_iterations": 602, + "inner_iterations": 45, + "total_iterations": 647, + "n_passive": 698, + "warm_start_errors": 552, + "index": 12 + }, + { + "warm_start_error_fraction": 0.27676240208877284, + "reconstruction_s": 1.0690451999998913, + "eval_s": 1.1281417999998666, + "log_likelihood": 5789.267655641282, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6050758000001224, + "n": 1149, + "solve_s": 0.6050758000001224, + "outer_iterations": 516, + "inner_iterations": 31, + "total_iterations": 547, + "n_passive": 697, + "warm_start_errors": 318, + "index": 13 + }, + { + "warm_start_error_fraction": 0.3333333333333333, + "reconstruction_s": 1.0361179999999877, + "eval_s": 1.115964300000087, + "log_likelihood": 5786.116211641706, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5597428999999465, + "n": 1149, + "solve_s": 0.5597428999999465, + "outer_iterations": 214, + "inner_iterations": 32, + "total_iterations": 246, + "n_passive": 718, + "warm_start_errors": 383, + "index": 14 + }, + { + "warm_start_error_fraction": 0.2863359442993908, + "reconstruction_s": 0.8698270999993838, + "eval_s": 0.9299685999994836, + "log_likelihood": 5628.060316065185, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42571940000016184, + "n": 1149, + "solve_s": 0.42571940000016184, + "outer_iterations": 183, + "inner_iterations": 26, + "total_iterations": 209, + "n_passive": 712, + "warm_start_errors": 329, + "index": 15 + }, + { + "warm_start_error_fraction": 0.43777197563098347, + "reconstruction_s": 0.9995843999995486, + "eval_s": 1.0654323000007935, + "log_likelihood": 5800.423879418462, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.49191530000007333, + "n": 1149, + "solve_s": 0.49191530000007333, + "outer_iterations": 535, + "inner_iterations": 46, + "total_iterations": 581, + "n_passive": 650, + "warm_start_errors": 503, + "index": 16 + }, + { + "warm_start_error_fraction": 0.4073107049608355, + "reconstruction_s": 1.0645346999999674, + "eval_s": 1.1384783000003154, + "log_likelihood": 6083.217120649668, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5524942000001829, + "n": 1149, + "solve_s": 0.5524942000001829, + "outer_iterations": 635, + "inner_iterations": 42, + "total_iterations": 677, + "n_passive": 723, + "warm_start_errors": 468, + "index": 17 + }, + { + "warm_start_error_fraction": 0.4351610095735422, + "reconstruction_s": 0.9010846000001038, + "eval_s": 0.975254799999675, + "log_likelihood": 5322.17507818152, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.48004310000033, + "n": 1149, + "solve_s": 0.48004310000033, + "outer_iterations": 566, + "inner_iterations": 37, + "total_iterations": 603, + "n_passive": 665, + "warm_start_errors": 500, + "index": 18 + }, + { + "warm_start_error_fraction": 0.4299390774586597, + "reconstruction_s": 1.0457054999997126, + "eval_s": 1.1185595000006288, + "log_likelihood": 4947.038427136894, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5466420999991897, + "n": 1149, + "solve_s": 0.5466420999991897, + "outer_iterations": 537, + "inner_iterations": 52, + "total_iterations": 589, + "n_passive": 629, + "warm_start_errors": 494, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.4516971279373368, + "reconstruction_s": 1.0586591999999655, + "eval_s": 1.1345826999995552, + "log_likelihood": 4339.419124288473, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.48008690000006027, + "n": 1149, + "solve_s": 0.48008690000006027, + "outer_iterations": 558, + "inner_iterations": 61, + "total_iterations": 619, + "n_passive": 610, + "warm_start_errors": 519, + "index": 0 + }, + { + "warm_start_error_fraction": 0.144473455178416, + "reconstruction_s": 0.7231113999996523, + "eval_s": 0.7890454999997019, + "log_likelihood": 5340.949647349872, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2836076999992656, + "n": 1149, + "solve_s": 0.2836076999992656, + "outer_iterations": 176, + "inner_iterations": 64, + "total_iterations": 240, + "n_passive": 670, + "warm_start_errors": 166, + "index": 1 + }, + { + "warm_start_error_fraction": 0.1462140992167102, + "reconstruction_s": 0.6279076999999234, + "eval_s": 0.687702199999876, + "log_likelihood": 5802.688366736172, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.19797440000002098, + "n": 1149, + "solve_s": 0.19797440000002098, + "outer_iterations": 135, + "inner_iterations": 32, + "total_iterations": 167, + "n_passive": 696, + "warm_start_errors": 168, + "index": 2 + }, + { + "warm_start_error_fraction": 0.144473455178416, + "reconstruction_s": 0.6787138999998206, + "eval_s": 0.7606493000002956, + "log_likelihood": 5354.284629582949, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2344057000000248, + "n": 1149, + "solve_s": 0.2344057000000248, + "outer_iterations": 166, + "inner_iterations": 49, + "total_iterations": 215, + "n_passive": 656, + "warm_start_errors": 166, + "index": 3 + }, + { + "warm_start_error_fraction": 0.11053089643167972, + "reconstruction_s": 0.677199799999471, + "eval_s": 0.7469777000005706, + "log_likelihood": 5635.0404539054825, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22925630000008823, + "n": 1149, + "solve_s": 0.22925630000008823, + "outer_iterations": 138, + "inner_iterations": 20, + "total_iterations": 158, + "n_passive": 713, + "warm_start_errors": 127, + "index": 4 + }, + { + "warm_start_error_fraction": 0.1227154046997389, + "reconstruction_s": 0.6566446999995605, + "eval_s": 0.7168105000000651, + "log_likelihood": 5571.756602169036, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.15143259999967995, + "n": 1149, + "solve_s": 0.15143259999967995, + "outer_iterations": 82, + "inner_iterations": 75, + "total_iterations": 157, + "n_passive": 662, + "warm_start_errors": 141, + "index": 5 + }, + { + "warm_start_error_fraction": 0.12184508268059181, + "reconstruction_s": 0.6109016999998858, + "eval_s": 0.6789565999997649, + "log_likelihood": 5534.492855134295, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1591616999994585, + "n": 1149, + "solve_s": 0.1591616999994585, + "outer_iterations": 95, + "inner_iterations": 32, + "total_iterations": 127, + "n_passive": 666, + "warm_start_errors": 140, + "index": 6 + }, + { + "warm_start_error_fraction": 0.15056570931244562, + "reconstruction_s": 0.6364144999997734, + "eval_s": 0.7172344999999041, + "log_likelihood": 4990.180171731855, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1575091999993674, + "n": 1149, + "solve_s": 0.1575091999993674, + "outer_iterations": 88, + "inner_iterations": 46, + "total_iterations": 134, + "n_passive": 615, + "warm_start_errors": 173, + "index": 7 + }, + { + "warm_start_error_fraction": 0.14708442123585727, + "reconstruction_s": 0.8327960000005987, + "eval_s": 0.916320899999846, + "log_likelihood": 5087.0044159130675, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3307796999997663, + "n": 1149, + "solve_s": 0.3307796999997663, + "outer_iterations": 207, + "inner_iterations": 33, + "total_iterations": 240, + "n_passive": 678, + "warm_start_errors": 169, + "index": 8 + }, + { + "warm_start_error_fraction": 0.15230635335073978, + "reconstruction_s": 0.8238438999997015, + "eval_s": 0.9076755000005505, + "log_likelihood": 4069.586330340285, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24618980000013835, + "n": 1149, + "solve_s": 0.24618980000013835, + "outer_iterations": 132, + "inner_iterations": 32, + "total_iterations": 164, + "n_passive": 639, + "warm_start_errors": 175, + "index": 9 + }, + { + "warm_start_error_fraction": 0.13751087902523934, + "reconstruction_s": 1.004553499999929, + "eval_s": 1.1049862999998368, + "log_likelihood": 5540.658694265025, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3244586000000709, + "n": 1149, + "solve_s": 0.3244586000000709, + "outer_iterations": 147, + "inner_iterations": 68, + "total_iterations": 215, + "n_passive": 649, + "warm_start_errors": 158, + "index": 10 + }, + { + "warm_start_error_fraction": 0.13838120104438642, + "reconstruction_s": 0.7305429999996704, + "eval_s": 0.8142136999995273, + "log_likelihood": 4408.075213771158, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.19535679999989952, + "n": 1149, + "solve_s": 0.19535679999989952, + "outer_iterations": 126, + "inner_iterations": 22, + "total_iterations": 148, + "n_passive": 640, + "warm_start_errors": 159, + "index": 11 + }, + { + "warm_start_error_fraction": 0.1731940818102698, + "reconstruction_s": 0.8649427999998807, + "eval_s": 0.9397607000000789, + "log_likelihood": 5236.436495748587, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33738180000000284, + "n": 1149, + "solve_s": 0.33738180000000284, + "outer_iterations": 185, + "inner_iterations": 30, + "total_iterations": 215, + "n_passive": 697, + "warm_start_errors": 199, + "index": 12 + }, + { + "warm_start_error_fraction": 0.11836379460400348, + "reconstruction_s": 0.6799949000005654, + "eval_s": 0.7521910999994361, + "log_likelihood": 5789.267655641294, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2052254999998695, + "n": 1149, + "solve_s": 0.2052254999998695, + "outer_iterations": 87, + "inner_iterations": 37, + "total_iterations": 124, + "n_passive": 697, + "warm_start_errors": 136, + "index": 13 + }, + { + "warm_start_error_fraction": 0.1174934725848564, + "reconstruction_s": 0.871604799999659, + "eval_s": 0.9539177999995445, + "log_likelihood": 5786.116211641708, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23153800000000047, + "n": 1149, + "solve_s": 0.23153800000000047, + "outer_iterations": 95, + "inner_iterations": 33, + "total_iterations": 128, + "n_passive": 718, + "warm_start_errors": 135, + "index": 14 + }, + { + "warm_start_error_fraction": 0.09312445604873804, + "reconstruction_s": 0.77760600000056, + "eval_s": 0.8841524999998001, + "log_likelihood": 5628.060316065176, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2259392999994816, + "n": 1149, + "solve_s": 0.2259392999994816, + "outer_iterations": 69, + "inner_iterations": 26, + "total_iterations": 95, + "n_passive": 711, + "warm_start_errors": 107, + "index": 15 + }, + { + "warm_start_error_fraction": 0.12532637075718014, + "reconstruction_s": 1.2319340999993074, + "eval_s": 1.3170918000005258, + "log_likelihood": 5800.423879418464, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39532210000015766, + "n": 1149, + "solve_s": 0.39532210000015766, + "outer_iterations": 136, + "inner_iterations": 136, + "total_iterations": 272, + "n_passive": 649, + "warm_start_errors": 144, + "index": 16 + }, + { + "warm_start_error_fraction": 0.134029590948651, + "reconstruction_s": 1.0349410999997417, + "eval_s": 1.1282616000007692, + "log_likelihood": 6083.217120649642, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2964270000002216, + "n": 1149, + "solve_s": 0.2964270000002216, + "outer_iterations": 131, + "inner_iterations": 21, + "total_iterations": 152, + "n_passive": 723, + "warm_start_errors": 154, + "index": 17 + }, + { + "warm_start_error_fraction": 0.15578764142732812, + "reconstruction_s": 0.9422908999995343, + "eval_s": 1.009471499999563, + "log_likelihood": 5322.175078181518, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25615860000016255, + "n": 1149, + "solve_s": 0.25615860000016255, + "outer_iterations": 116, + "inner_iterations": 93, + "total_iterations": 209, + "n_passive": 664, + "warm_start_errors": 179, + "index": 18 + }, + { + "warm_start_error_fraction": 0.11575282854656223, + "reconstruction_s": 0.682679200000166, + "eval_s": 0.7589382000005571, + "log_likelihood": 4947.038427136894, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.14311129999987315, + "n": 1149, + "solve_s": 0.14311129999987315, + "outer_iterations": 71, + "inner_iterations": 20, + "total_iterations": 91, + "n_passive": 629, + "warm_start_errors": 133, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_rectangular_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_rectangular_v2026.8.17.1.png new file mode 100644 index 000000000..379c92f8c Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_rectangular_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_sersic_light_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_sersic_light_v2026.8.17.1.json new file mode 100644 index 000000000..47d7ffba5 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_sersic_light_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "sersic_light", + "configuration": { + "model": "sersic_light", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 13, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "sersic", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 2.7120639999993728, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.12310151878497202, + "mean_warm_start_error_fraction": 0.12414068745003995, + "p90_warm_start_error_fraction": 0.13029576338928858, + "max_warm_start_error_fraction": 0.13189448441247004, + "median_outer_iterations": 37.0, + "mean_outer_iterations": 36.95, + "p90_outer_iterations": 41.1, + "max_outer_iterations": 43.0, + "median_inner_iterations": 5.0, + "mean_inner_iterations": 5.25, + "p90_inner_iterations": 6.100000000000001, + "max_inner_iterations": 8.0, + "median_total_iterations": 42.5, + "mean_total_iterations": 42.2, + "p90_total_iterations": 48.0, + "max_total_iterations": 48.0, + "median_n_passive": 996.0, + "mean_n_passive": 995.3, + "p90_n_passive": 998.1, + "max_n_passive": 1000.0, + "median_warm_start_errors": 154.0, + "mean_warm_start_errors": 155.3, + "p90_warm_start_errors": 163.0, + "max_warm_start_errors": 165.0, + "median_solve_s": 0.3149608500002614, + "mean_solve_s": 0.3305632200000673, + "p90_solve_s": 0.39097595999946866, + "max_solve_s": 0.4194158000000243, + "median_reconstruction_s": 0.5603369499999644, + "mean_reconstruction_s": 0.5712399849998746, + "p90_reconstruction_s": 0.6395796199996766, + "max_reconstruction_s": 0.655620600000475, + "median_eval_s": 0.6321628500004408, + "mean_eval_s": 0.6438327599999866, + "p90_eval_s": 0.7132987799994226, + "max_eval_s": 0.7319256000000678 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.005195843325339728, + "mean_warm_start_error_fraction": 0.012190247801758593, + "p90_warm_start_error_fraction": 0.012949640287769786, + "max_warm_start_error_fraction": 0.12310151878497202, + "median_outer_iterations": 4.0, + "mean_outer_iterations": 6.1, + "p90_outer_iterations": 9.200000000000003, + "max_outer_iterations": 33.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 1.15, + "p90_inner_iterations": 1.1000000000000014, + "max_inner_iterations": 4.0, + "median_total_iterations": 5.0, + "mean_total_iterations": 7.25, + "p90_total_iterations": 10.200000000000003, + "max_total_iterations": 37.0, + "median_n_passive": 996.0, + "mean_n_passive": 995.3, + "p90_n_passive": 998.1, + "max_n_passive": 1000.0, + "median_warm_start_errors": 6.5, + "mean_warm_start_errors": 15.25, + "p90_warm_start_errors": 16.200000000000003, + "max_warm_start_errors": 154.0, + "median_solve_s": 0.05296055000007982, + "mean_solve_s": 0.06174523999993653, + "p90_solve_s": 0.07338352999986456, + "max_solve_s": 0.2844716999998127, + "median_reconstruction_s": 0.24801215000024968, + "mean_reconstruction_s": 0.28224222499998175, + "p90_reconstruction_s": 0.3761812200002169, + "max_reconstruction_s": 0.5117489999993268, + "median_eval_s": 0.33047560000022713, + "mean_eval_s": 0.3666472999999769, + "p90_eval_s": 0.4730820299992957, + "max_eval_s": 0.5850946999998996 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 3.1086064220151323e-15, + "max_abs_reconstruction_deviation": 2.731148640577885e-13 + }, + "iteration_reduction_ratio": 8.5, + "solve_speedup_ratio": 5.9470841975732265, + "eval_speedup_ratio": 1.9128881224514196, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.12310151878497202, + "reconstruction_s": 0.6133340999995198, + "eval_s": 0.6886515000005602, + "log_likelihood": 1652.5761308289184, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3176141000003554, + "n": 1251, + "solve_s": 0.3176141000003554, + "outer_iterations": 33, + "inner_iterations": 4, + "total_iterations": 37, + "n_passive": 995, + "warm_start_errors": 154, + "index": 0 + }, + { + "warm_start_error_fraction": 0.1223021582733813, + "reconstruction_s": 0.5584157000002961, + "eval_s": 0.6329523000003974, + "log_likelihood": 1701.9739872035516, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30110510000031354, + "n": 1251, + "solve_s": 0.30110510000031354, + "outer_iterations": 34, + "inner_iterations": 5, + "total_iterations": 39, + "n_passive": 997, + "warm_start_errors": 153, + "index": 1 + }, + { + "warm_start_error_fraction": 0.11990407673860912, + "reconstruction_s": 0.5463761999999406, + "eval_s": 0.6193106000000625, + "log_likelihood": 1691.809027612544, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3067485999999917, + "n": 1251, + "solve_s": 0.3067485999999917, + "outer_iterations": 36, + "inner_iterations": 5, + "total_iterations": 41, + "n_passive": 998, + "warm_start_errors": 150, + "index": 2 + }, + { + "warm_start_error_fraction": 0.1223021582733813, + "reconstruction_s": 0.5355744999997114, + "eval_s": 0.6080855000000156, + "log_likelihood": 1668.30826247179, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29634919999989506, + "n": 1251, + "solve_s": 0.29634919999989506, + "outer_iterations": 36, + "inner_iterations": 5, + "total_iterations": 41, + "n_passive": 998, + "warm_start_errors": 153, + "index": 3 + }, + { + "warm_start_error_fraction": 0.13029576338928858, + "reconstruction_s": 0.5652356000000509, + "eval_s": 0.6437353000001167, + "log_likelihood": 1626.3122397534935, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32449019999967277, + "n": 1251, + "solve_s": 0.32449019999967277, + "outer_iterations": 39, + "inner_iterations": 5, + "total_iterations": 44, + "n_passive": 990, + "warm_start_errors": 163, + "index": 4 + }, + { + "warm_start_error_fraction": 0.12949640287769784, + "reconstruction_s": 0.6390070999996169, + "eval_s": 0.7135803000001033, + "log_likelihood": 1686.8366352309404, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3885138999994524, + "n": 1251, + "solve_s": 0.3885138999994524, + "outer_iterations": 42, + "inner_iterations": 6, + "total_iterations": 48, + "n_passive": 991, + "warm_start_errors": 162, + "index": 5 + }, + { + "warm_start_error_fraction": 0.13029576338928858, + "reconstruction_s": 0.5290106000002197, + "eval_s": 0.5980372000003626, + "log_likelihood": 1660.9205535599049, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29394000000047527, + "n": 1251, + "solve_s": 0.29394000000047527, + "outer_iterations": 38, + "inner_iterations": 6, + "total_iterations": 44, + "n_passive": 992, + "warm_start_errors": 163, + "index": 6 + }, + { + "warm_start_error_fraction": 0.13189448441247004, + "reconstruction_s": 0.5625343000001521, + "eval_s": 0.6313734000004843, + "log_likelihood": 1516.2269851041783, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33881770000061806, + "n": 1251, + "solve_s": 0.33881770000061806, + "outer_iterations": 40, + "inner_iterations": 7, + "total_iterations": 47, + "n_passive": 987, + "warm_start_errors": 165, + "index": 7 + }, + { + "warm_start_error_fraction": 0.12629896083133493, + "reconstruction_s": 0.5407672000001185, + "eval_s": 0.6108864999996513, + "log_likelihood": 1591.6022046787602, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.309226000000308, + "n": 1251, + "solve_s": 0.309226000000308, + "outer_iterations": 40, + "inner_iterations": 8, + "total_iterations": 48, + "n_passive": 995, + "warm_start_errors": 158, + "index": 8 + }, + { + "warm_start_error_fraction": 0.12310151878497202, + "reconstruction_s": 0.5342347000005248, + "eval_s": 0.6068328999999721, + "log_likelihood": 1623.5528728453282, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3022676999999021, + "n": 1251, + "solve_s": 0.3022676999999021, + "outer_iterations": 36, + "inner_iterations": 6, + "total_iterations": 42, + "n_passive": 996, + "warm_start_errors": 154, + "index": 9 + }, + { + "warm_start_error_fraction": 0.12629896083133493, + "reconstruction_s": 0.606065199999648, + "eval_s": 0.6803639999998268, + "log_likelihood": 1641.3135276153607, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3559587999998257, + "n": 1251, + "solve_s": 0.3559587999998257, + "outer_iterations": 40, + "inner_iterations": 6, + "total_iterations": 46, + "n_passive": 996, + "warm_start_errors": 158, + "index": 10 + }, + { + "warm_start_error_fraction": 0.12070343725019984, + "reconstruction_s": 0.5457495999999082, + "eval_s": 0.6223276999999143, + "log_likelihood": 1672.0206961205731, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.302801100000579, + "n": 1251, + "solve_s": 0.302801100000579, + "outer_iterations": 35, + "inner_iterations": 4, + "total_iterations": 39, + "n_passive": 996, + "warm_start_errors": 151, + "index": 11 + }, + { + "warm_start_error_fraction": 0.12070343725019984, + "reconstruction_s": 0.623561399999744, + "eval_s": 0.6927658999993582, + "log_likelihood": 1674.4938575308497, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3553680000004533, + "n": 1251, + "solve_s": 0.3553680000004533, + "outer_iterations": 38, + "inner_iterations": 6, + "total_iterations": 44, + "n_passive": 999, + "warm_start_errors": 151, + "index": 12 + }, + { + "warm_start_error_fraction": 0.1254996003197442, + "reconstruction_s": 0.4975449999992634, + "eval_s": 0.5638011000000915, + "log_likelihood": 1701.1901801018976, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27158759999929316, + "n": 1251, + "solve_s": 0.27158759999929316, + "outer_iterations": 32, + "inner_iterations": 4, + "total_iterations": 36, + "n_passive": 995, + "warm_start_errors": 157, + "index": 13 + }, + { + "warm_start_error_fraction": 0.12310151878497202, + "reconstruction_s": 0.5622581999996328, + "eval_s": 0.6372253999998065, + "log_likelihood": 1689.2666380853425, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3344526999999289, + "n": 1251, + "solve_s": 0.3344526999999289, + "outer_iterations": 41, + "inner_iterations": 5, + "total_iterations": 46, + "n_passive": 997, + "warm_start_errors": 154, + "index": 14 + }, + { + "warm_start_error_fraction": 0.1254996003197442, + "reconstruction_s": 0.655620600000475, + "eval_s": 0.7319256000000678, + "log_likelihood": 1690.2049147776615, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4194158000000243, + "n": 1251, + "solve_s": 0.4194158000000243, + "outer_iterations": 43, + "inner_iterations": 5, + "total_iterations": 48, + "n_passive": 996, + "warm_start_errors": 157, + "index": 15 + }, + { + "warm_start_error_fraction": 0.11910471622701839, + "reconstruction_s": 0.6447323000002143, + "eval_s": 0.7132674999993469, + "log_likelihood": 1755.4387625243753, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4131344999996145, + "n": 1251, + "solve_s": 0.4131344999996145, + "outer_iterations": 35, + "inner_iterations": 4, + "total_iterations": 39, + "n_passive": 1000, + "warm_start_errors": 149, + "index": 16 + }, + { + "warm_start_error_fraction": 0.12150279776179057, + "reconstruction_s": 0.5142981999997573, + "eval_s": 0.5839759000000413, + "log_likelihood": 1820.9202040694381, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2872975000000224, + "n": 1251, + "solve_s": 0.2872975000000224, + "outer_iterations": 38, + "inner_iterations": 5, + "total_iterations": 43, + "n_passive": 996, + "warm_start_errors": 152, + "index": 17 + }, + { + "warm_start_error_fraction": 0.12150279776179057, + "reconstruction_s": 0.6068352999991475, + "eval_s": 0.6798982000000251, + "log_likelihood": 1775.4015601973933, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.379868300000453, + "n": 1251, + "solve_s": 0.379868300000453, + "outer_iterations": 29, + "inner_iterations": 4, + "total_iterations": 33, + "n_passive": 994, + "warm_start_errors": 152, + "index": 18 + }, + { + "warm_start_error_fraction": 0.11990407673860912, + "reconstruction_s": 0.5436438999995516, + "eval_s": 0.6176583999995273, + "log_likelihood": 1889.8864603449674, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31230760000016744, + "n": 1251, + "solve_s": 0.31230760000016744, + "outer_iterations": 34, + "inner_iterations": 5, + "total_iterations": 39, + "n_passive": 998, + "warm_start_errors": 150, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.12310151878497202, + "reconstruction_s": 0.5117489999993268, + "eval_s": 0.5850946999998996, + "log_likelihood": 1652.5761308289184, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2844716999998127, + "n": 1251, + "solve_s": 0.2844716999998127, + "outer_iterations": 33, + "inner_iterations": 4, + "total_iterations": 37, + "n_passive": 995, + "warm_start_errors": 154, + "index": 0 + }, + { + "warm_start_error_fraction": 0.007993605115907274, + "reconstruction_s": 0.25753049999912037, + "eval_s": 0.33220320000054926, + "log_likelihood": 1701.9739872035498, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06896570000026259, + "n": 1251, + "solve_s": 0.06896570000026259, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 997, + "warm_start_errors": 10, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0007993605115907274, + "reconstruction_s": 0.22852179999972577, + "eval_s": 0.2993118999993385, + "log_likelihood": 1691.809027612544, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.029854299999897194, + "n": 1251, + "solve_s": 0.029854299999897194, + "outer_iterations": 1, + "inner_iterations": 0, + "total_iterations": 1, + "n_passive": 998, + "warm_start_errors": 1, + "index": 2 + }, + { + "warm_start_error_fraction": 0.006394884092725819, + "reconstruction_s": 0.24250060000031226, + "eval_s": 0.3535953000000518, + "log_likelihood": 1668.30826247179, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05922039999950357, + "n": 1251, + "solve_s": 0.05922039999950357, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 998, + "warm_start_errors": 8, + "index": 3 + }, + { + "warm_start_error_fraction": 0.014388489208633094, + "reconstruction_s": 0.31314039999961096, + "eval_s": 0.39287460000014107, + "log_likelihood": 1626.3122397534935, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07560499999999593, + "n": 1251, + "solve_s": 0.07560499999999593, + "outer_iterations": 8, + "inner_iterations": 2, + "total_iterations": 10, + "n_passive": 990, + "warm_start_errors": 18, + "index": 4 + }, + { + "warm_start_error_fraction": 0.002398081534772182, + "reconstruction_s": 0.30539069999940693, + "eval_s": 0.4380750000000262, + "log_likelihood": 1686.8366352309386, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.056418899999698624, + "n": 1251, + "solve_s": 0.056418899999698624, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 991, + "warm_start_errors": 3, + "index": 5 + }, + { + "warm_start_error_fraction": 0.008792965627498001, + "reconstruction_s": 0.324721199999658, + "eval_s": 0.43877659999998286, + "log_likelihood": 1660.9205535599067, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.059552099999564234, + "n": 1251, + "solve_s": 0.059552099999564234, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 992, + "warm_start_errors": 11, + "index": 6 + }, + { + "warm_start_error_fraction": 0.005595523581135092, + "reconstruction_s": 0.4045836000004783, + "eval_s": 0.5120270999996137, + "log_likelihood": 1516.2269851041801, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049502200000461016, + "n": 1251, + "solve_s": 0.049502200000461016, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 987, + "warm_start_errors": 7, + "index": 7 + }, + { + "warm_start_error_fraction": 0.007993605115907274, + "reconstruction_s": 0.3365455999992264, + "eval_s": 0.42056789999969624, + "log_likelihood": 1591.6022046787602, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07313669999984995, + "n": 1251, + "solve_s": 0.07313669999984995, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 995, + "warm_start_errors": 10, + "index": 8 + }, + { + "warm_start_error_fraction": 0.002398081534772182, + "reconstruction_s": 0.24278550000053656, + "eval_s": 0.3249919999998383, + "log_likelihood": 1623.55287284533, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03736939999998867, + "n": 1251, + "solve_s": 0.03736939999998867, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 996, + "warm_start_errors": 3, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0015987210231814548, + "reconstruction_s": 0.23975380000047153, + "eval_s": 0.3127795000000333, + "log_likelihood": 1641.3135276153607, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03225159999965399, + "n": 1251, + "solve_s": 0.03225159999965399, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 996, + "warm_start_errors": 2, + "index": 10 + }, + { + "warm_start_error_fraction": 0.012789768185451638, + "reconstruction_s": 0.2532387999999628, + "eval_s": 0.328747999999905, + "log_likelihood": 1672.0206961205731, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06213409999963915, + "n": 1251, + "solve_s": 0.06213409999963915, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 996, + "warm_start_errors": 16, + "index": 11 + }, + { + "warm_start_error_fraction": 0.003996802557953637, + "reconstruction_s": 0.2240179000000353, + "eval_s": 0.2913415000002715, + "log_likelihood": 1674.4938575308515, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03517959999953746, + "n": 1251, + "solve_s": 0.03517959999953746, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 999, + "warm_start_errors": 5, + "index": 12 + }, + { + "warm_start_error_fraction": 0.004796163069544364, + "reconstruction_s": 0.20778080000036425, + "eval_s": 0.2756134999999631, + "log_likelihood": 1701.1901801018976, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.030695399999785877, + "n": 1251, + "solve_s": 0.030695399999785877, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 995, + "warm_start_errors": 6, + "index": 13 + }, + { + "warm_start_error_fraction": 0.012789768185451638, + "reconstruction_s": 0.24178280000069208, + "eval_s": 0.30955380000068544, + "log_likelihood": 1689.2666380853443, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.048568800000794, + "n": 1251, + "solve_s": 0.048568800000794, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 997, + "warm_start_errors": 16, + "index": 14 + }, + { + "warm_start_error_fraction": 0.002398081534772182, + "reconstruction_s": 0.2050954999995156, + "eval_s": 0.27431500000056985, + "log_likelihood": 1690.2049147776597, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.027305900000101246, + "n": 1251, + "solve_s": 0.027305900000101246, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 996, + "warm_start_errors": 3, + "index": 15 + }, + { + "warm_start_error_fraction": 0.011191047162270184, + "reconstruction_s": 0.24256990000048972, + "eval_s": 0.31305169999995996, + "log_likelihood": 1755.4387625243698, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.057331900000463065, + "n": 1251, + "solve_s": 0.057331900000463065, + "outer_iterations": 11, + "inner_iterations": 1, + "total_iterations": 12, + "n_passive": 1000, + "warm_start_errors": 14, + "index": 16 + }, + { + "warm_start_error_fraction": 0.004796163069544364, + "reconstruction_s": 0.21942250000029162, + "eval_s": 0.2880730999995649, + "log_likelihood": 1820.92020406944, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.02947330000006332, + "n": 1251, + "solve_s": 0.02947330000006332, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 996, + "warm_start_errors": 6, + "index": 17 + }, + { + "warm_start_error_fraction": 0.004796163069544364, + "reconstruction_s": 0.2706882000002224, + "eval_s": 0.37319680000018707, + "log_likelihood": 1775.4015601973952, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04753290000007837, + "n": 1251, + "solve_s": 0.04753290000007837, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 994, + "warm_start_errors": 6, + "index": 18 + }, + { + "warm_start_error_fraction": 0.004796163069544364, + "reconstruction_s": 0.3730254000001878, + "eval_s": 0.4687547999992603, + "log_likelihood": 1889.8864603449692, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07033489999957965, + "n": 1251, + "solve_s": 0.07033489999957965, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 998, + "warm_start_errors": 6, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.12549960031974422, + "mean_warm_start_error_fraction": 0.12537969624300566, + "p90_warm_start_error_fraction": 0.13309352517985612, + "max_warm_start_error_fraction": 0.13828936850519585, + "median_outer_iterations": 38.5, + "mean_outer_iterations": 39.65, + "p90_outer_iterations": 49.2, + "max_outer_iterations": 55.0, + "median_inner_iterations": 5.5, + "mean_inner_iterations": 5.8, + "p90_inner_iterations": 9.0, + "max_inner_iterations": 11.0, + "median_total_iterations": 43.5, + "mean_total_iterations": 45.45, + "p90_total_iterations": 54.400000000000006, + "max_total_iterations": 64.0, + "median_n_passive": 991.0, + "mean_n_passive": 990.05, + "p90_n_passive": 1001.1, + "max_n_passive": 1003.0, + "median_warm_start_errors": 157.0, + "mean_warm_start_errors": 156.85, + "p90_warm_start_errors": 166.5, + "max_warm_start_errors": 173.0, + "median_solve_s": 0.3687179000003198, + "mean_solve_s": 0.3862781100000575, + "p90_solve_s": 0.4928554599999188, + "max_solve_s": 0.6169455999997808, + "median_reconstruction_s": 0.6166522500002429, + "mean_reconstruction_s": 0.6540870200001336, + "p90_reconstruction_s": 0.8225948400003291, + "max_reconstruction_s": 0.9269345000002431, + "median_eval_s": 0.6889272999997047, + "mean_eval_s": 0.7334698950000075, + "p90_eval_s": 0.9021310599999198, + "max_eval_s": 1.0166475000005448 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.049160671462829736, + "mean_warm_start_error_fraction": 0.0521982414068745, + "p90_warm_start_error_fraction": 0.06818545163868905, + "max_warm_start_error_fraction": 0.12150279776179057, + "median_outer_iterations": 37.0, + "mean_outer_iterations": 38.55, + "p90_outer_iterations": 53.60000000000001, + "max_outer_iterations": 60.0, + "median_inner_iterations": 3.0, + "mean_inner_iterations": 3.4, + "p90_inner_iterations": 6.200000000000003, + "max_inner_iterations": 9.0, + "median_total_iterations": 41.0, + "mean_total_iterations": 41.95, + "p90_total_iterations": 58.400000000000006, + "max_total_iterations": 63.0, + "median_n_passive": 991.0, + "mean_n_passive": 990.05, + "p90_n_passive": 1001.1, + "max_n_passive": 1003.0, + "median_warm_start_errors": 61.5, + "mean_warm_start_errors": 65.3, + "p90_warm_start_errors": 85.30000000000001, + "max_warm_start_errors": 152.0, + "median_solve_s": 0.22298760000012408, + "mean_solve_s": 0.23277681000004122, + "p90_solve_s": 0.3428167800007033, + "max_solve_s": 0.4410288999997647, + "median_reconstruction_s": 0.5362130500002422, + "mean_reconstruction_s": 0.5211894100000336, + "p90_reconstruction_s": 0.7080203099996653, + "max_reconstruction_s": 0.7963267999994059, + "median_eval_s": 0.648461750000024, + "mean_eval_s": 0.6348973649999607, + "p90_eval_s": 0.8712151899996571, + "max_eval_s": 0.9387720000004265 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 4.081934162145684e-15, + "max_abs_reconstruction_deviation": 3.184119634624949e-13 + }, + "iteration_reduction_ratio": 1.0609756097560976, + "solve_speedup_ratio": 1.6535354432269536, + "eval_speedup_ratio": 1.0624023699156953, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.12150279776179057, + "reconstruction_s": 0.6058874000000287, + "eval_s": 0.680022700000336, + "log_likelihood": 2236.238572622756, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33494789999986097, + "n": 1251, + "solve_s": 0.33494789999986097, + "outer_iterations": 39, + "inner_iterations": 9, + "total_iterations": 48, + "n_passive": 997, + "warm_start_errors": 152, + "index": 0 + }, + { + "warm_start_error_fraction": 0.1366906474820144, + "reconstruction_s": 0.6578463000005286, + "eval_s": 0.7429032000000007, + "log_likelihood": 285.2360980116373, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42165860000022803, + "n": 1251, + "solve_s": 0.42165860000022803, + "outer_iterations": 43, + "inner_iterations": 11, + "total_iterations": 54, + "n_passive": 975, + "warm_start_errors": 171, + "index": 1 + }, + { + "warm_start_error_fraction": 0.13189448441247004, + "reconstruction_s": 0.9269345000002431, + "eval_s": 1.0166475000005448, + "log_likelihood": 1596.3726077201081, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6169455999997808, + "n": 1251, + "solve_s": 0.6169455999997808, + "outer_iterations": 37, + "inner_iterations": 6, + "total_iterations": 43, + "n_passive": 989, + "warm_start_errors": 165, + "index": 2 + }, + { + "warm_start_error_fraction": 0.12949640287769784, + "reconstruction_s": 0.8150240000004487, + "eval_s": 0.8924754999998186, + "log_likelihood": 1068.8436552615512, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4737519999998767, + "n": 1251, + "solve_s": 0.4737519999998767, + "outer_iterations": 48, + "inner_iterations": 4, + "total_iterations": 52, + "n_passive": 990, + "warm_start_errors": 162, + "index": 3 + }, + { + "warm_start_error_fraction": 0.11830535571542766, + "reconstruction_s": 0.7791431000005105, + "eval_s": 0.8792151000006925, + "log_likelihood": 2286.407049091591, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4918170999999347, + "n": 1251, + "solve_s": 0.4918170999999347, + "outer_iterations": 51, + "inner_iterations": 7, + "total_iterations": 58, + "n_passive": 1002, + "warm_start_errors": 148, + "index": 4 + }, + { + "warm_start_error_fraction": 0.13269384492406075, + "reconstruction_s": 0.8907323999992514, + "eval_s": 0.9890311000008296, + "log_likelihood": 1796.6040919590532, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5022006999997757, + "n": 1251, + "solve_s": 0.5022006999997757, + "outer_iterations": 49, + "inner_iterations": 5, + "total_iterations": 54, + "n_passive": 989, + "warm_start_errors": 166, + "index": 5 + }, + { + "warm_start_error_fraction": 0.13269384492406075, + "reconstruction_s": 0.6204978000005212, + "eval_s": 0.6922717999996166, + "log_likelihood": 139.22005029622778, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3367950000001656, + "n": 1251, + "solve_s": 0.3367950000001656, + "outer_iterations": 43, + "inner_iterations": 6, + "total_iterations": 49, + "n_passive": 977, + "warm_start_errors": 166, + "index": 6 + }, + { + "warm_start_error_fraction": 0.11990407673860912, + "reconstruction_s": 0.7342017000000851, + "eval_s": 0.8230269999994562, + "log_likelihood": 1737.429633983278, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.44963319999988016, + "n": 1251, + "solve_s": 0.44963319999988016, + "outer_iterations": 28, + "inner_iterations": 4, + "total_iterations": 32, + "n_passive": 994, + "warm_start_errors": 150, + "index": 7 + }, + { + "warm_start_error_fraction": 0.12709832134292565, + "reconstruction_s": 0.6839649000003192, + "eval_s": 0.7563104999999268, + "log_likelihood": 1068.8940506885701, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4128915000001143, + "n": 1251, + "solve_s": 0.4128915000001143, + "outer_iterations": 33, + "inner_iterations": 5, + "total_iterations": 38, + "n_passive": 985, + "warm_start_errors": 159, + "index": 8 + }, + { + "warm_start_error_fraction": 0.12470023980815348, + "reconstruction_s": 0.5583333000004131, + "eval_s": 0.6325190999996266, + "log_likelihood": 1447.7862755595925, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3323055999999269, + "n": 1251, + "solve_s": 0.3323055999999269, + "outer_iterations": 40, + "inner_iterations": 3, + "total_iterations": 43, + "n_passive": 997, + "warm_start_errors": 156, + "index": 9 + }, + { + "warm_start_error_fraction": 0.1278976818545164, + "reconstruction_s": 0.5942098000004989, + "eval_s": 0.6673951999991914, + "log_likelihood": 924.2316025960044, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35860150000007707, + "n": 1251, + "solve_s": 0.35860150000007707, + "outer_iterations": 55, + "inner_iterations": 9, + "total_iterations": 64, + "n_passive": 974, + "warm_start_errors": 160, + "index": 10 + }, + { + "warm_start_error_fraction": 0.11830535571542766, + "reconstruction_s": 0.4965312000003905, + "eval_s": 0.576389100000597, + "log_likelihood": 1984.5416657187025, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2686251999994056, + "n": 1251, + "solve_s": 0.2686251999994056, + "outer_iterations": 28, + "inner_iterations": 7, + "total_iterations": 35, + "n_passive": 995, + "warm_start_errors": 148, + "index": 11 + }, + { + "warm_start_error_fraction": 0.12470023980815348, + "reconstruction_s": 0.4974703000007139, + "eval_s": 0.5742153999999573, + "log_likelihood": 2455.022835894373, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26948540000012144, + "n": 1251, + "solve_s": 0.26948540000012144, + "outer_iterations": 35, + "inner_iterations": 3, + "total_iterations": 38, + "n_passive": 996, + "warm_start_errors": 156, + "index": 12 + }, + { + "warm_start_error_fraction": 0.12629896083133493, + "reconstruction_s": 0.5609006999993653, + "eval_s": 0.6352663999996366, + "log_likelihood": 1549.8176167080092, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31685210000068764, + "n": 1251, + "solve_s": 0.31685210000068764, + "outer_iterations": 37, + "inner_iterations": 5, + "total_iterations": 42, + "n_passive": 992, + "warm_start_errors": 158, + "index": 13 + }, + { + "warm_start_error_fraction": 0.13029576338928858, + "reconstruction_s": 0.5664097999997466, + "eval_s": 0.6362198000006174, + "log_likelihood": 716.7556351102248, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.316066100000171, + "n": 1251, + "solve_s": 0.316066100000171, + "outer_iterations": 44, + "inner_iterations": 7, + "total_iterations": 51, + "n_passive": 986, + "warm_start_errors": 163, + "index": 14 + }, + { + "warm_start_error_fraction": 0.12310151878497202, + "reconstruction_s": 0.6128066999999646, + "eval_s": 0.6855827999997928, + "log_likelihood": 2296.98351895181, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3788343000005625, + "n": 1251, + "solve_s": 0.3788343000005625, + "outer_iterations": 38, + "inner_iterations": 6, + "total_iterations": 44, + "n_passive": 1003, + "warm_start_errors": 154, + "index": 15 + }, + { + "warm_start_error_fraction": 0.10871302957633892, + "reconstruction_s": 0.600066599999991, + "eval_s": 0.6757920999998532, + "log_likelihood": 2279.074679919747, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3545808999997462, + "n": 1251, + "solve_s": 0.3545808999997462, + "outer_iterations": 31, + "inner_iterations": 3, + "total_iterations": 34, + "n_passive": 1001, + "warm_start_errors": 136, + "index": 16 + }, + { + "warm_start_error_fraction": 0.1223021582733813, + "reconstruction_s": 0.6539686999994956, + "eval_s": 0.7345793999993475, + "log_likelihood": 1638.3558680333863, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38953400000082183, + "n": 1251, + "solve_s": 0.38953400000082183, + "outer_iterations": 37, + "inner_iterations": 3, + "total_iterations": 40, + "n_passive": 996, + "warm_start_errors": 153, + "index": 17 + }, + { + "warm_start_error_fraction": 0.13828936850519585, + "reconstruction_s": 0.6397850000003018, + "eval_s": 0.7204030000002604, + "log_likelihood": 445.6194860795349, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3795853999999963, + "n": 1251, + "solve_s": 0.3795853999999963, + "outer_iterations": 35, + "inner_iterations": 4, + "total_iterations": 39, + "n_passive": 976, + "warm_start_errors": 173, + "index": 18 + }, + { + "warm_start_error_fraction": 0.11270983213429256, + "reconstruction_s": 0.5870261999998547, + "eval_s": 0.6591312000000471, + "log_likelihood": 1470.7554900904652, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3204501000000164, + "n": 1251, + "solve_s": 0.3204501000000164, + "outer_iterations": 42, + "inner_iterations": 9, + "total_iterations": 51, + "n_passive": 987, + "warm_start_errors": 141, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.12150279776179057, + "reconstruction_s": 0.5640638999993826, + "eval_s": 0.6374088000002303, + "log_likelihood": 2236.238572622756, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33271420000073704, + "n": 1251, + "solve_s": 0.33271420000073704, + "outer_iterations": 39, + "inner_iterations": 9, + "total_iterations": 48, + "n_passive": 997, + "warm_start_errors": 152, + "index": 0 + }, + { + "warm_start_error_fraction": 0.07034372501998401, + "reconstruction_s": 0.38853199999994104, + "eval_s": 0.4587509000002683, + "log_likelihood": 285.2360980116373, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20164790000035282, + "n": 1251, + "solve_s": 0.20164790000035282, + "outer_iterations": 47, + "inner_iterations": 8, + "total_iterations": 55, + "n_passive": 975, + "warm_start_errors": 88, + "index": 1 + }, + { + "warm_start_error_fraction": 0.04476418864908074, + "reconstruction_s": 0.29603659999975207, + "eval_s": 0.3642409999993106, + "log_likelihood": 1596.3726077201081, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11235209999995277, + "n": 1251, + "solve_s": 0.11235209999995277, + "outer_iterations": 37, + "inner_iterations": 2, + "total_iterations": 39, + "n_passive": 989, + "warm_start_errors": 56, + "index": 2 + }, + { + "warm_start_error_fraction": 0.053557154276578735, + "reconstruction_s": 0.2969762000002447, + "eval_s": 0.3679859000003489, + "log_likelihood": 1068.843655261553, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.12171440000020084, + "n": 1251, + "solve_s": 0.12171440000020084, + "outer_iterations": 43, + "inner_iterations": 3, + "total_iterations": 46, + "n_passive": 990, + "warm_start_errors": 67, + "index": 3 + }, + { + "warm_start_error_fraction": 0.023980815347721823, + "reconstruction_s": 0.282517700000426, + "eval_s": 0.35502700000051846, + "log_likelihood": 2286.407049091591, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08350890000019717, + "n": 1251, + "solve_s": 0.08350890000019717, + "outer_iterations": 22, + "inner_iterations": 3, + "total_iterations": 25, + "n_passive": 1002, + "warm_start_errors": 30, + "index": 4 + }, + { + "warm_start_error_fraction": 0.058353317346123104, + "reconstruction_s": 0.6270964000004824, + "eval_s": 0.752921499999502, + "log_likelihood": 1796.604091959055, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43374000000039814, + "n": 1251, + "solve_s": 0.43374000000039814, + "outer_iterations": 45, + "inner_iterations": 4, + "total_iterations": 49, + "n_passive": 989, + "warm_start_errors": 73, + "index": 5 + }, + { + "warm_start_error_fraction": 0.04316546762589928, + "reconstruction_s": 0.3899664999999004, + "eval_s": 0.4729639999995925, + "log_likelihood": 139.22005029622778, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.13066629999957513, + "n": 1251, + "solve_s": 0.13066629999957513, + "outer_iterations": 27, + "inner_iterations": 2, + "total_iterations": 29, + "n_passive": 977, + "warm_start_errors": 54, + "index": 6 + }, + { + "warm_start_error_fraction": 0.05515587529976019, + "reconstruction_s": 0.36995779999961087, + "eval_s": 0.4490107999999964, + "log_likelihood": 1737.429633983278, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.16292969999994966, + "n": 1251, + "solve_s": 0.16292969999994966, + "outer_iterations": 49, + "inner_iterations": 3, + "total_iterations": 52, + "n_passive": 994, + "warm_start_errors": 69, + "index": 7 + }, + { + "warm_start_error_fraction": 0.05675459632294165, + "reconstruction_s": 0.47270819999994274, + "eval_s": 0.6047834999999395, + "log_likelihood": 1068.8940506885701, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2707885999998325, + "n": 1251, + "solve_s": 0.2707885999998325, + "outer_iterations": 42, + "inner_iterations": 1, + "total_iterations": 43, + "n_passive": 985, + "warm_start_errors": 71, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0319744204636291, + "reconstruction_s": 0.7503618000000642, + "eval_s": 0.9060685000004014, + "log_likelihood": 1447.7862755595943, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30189810000047146, + "n": 1251, + "solve_s": 0.30189810000047146, + "outer_iterations": 31, + "inner_iterations": 1, + "total_iterations": 32, + "n_passive": 997, + "warm_start_errors": 40, + "index": 9 + }, + { + "warm_start_error_fraction": 0.05995203836930456, + "reconstruction_s": 0.6979865000002974, + "eval_s": 0.8466979000004358, + "log_likelihood": 924.2316025960063, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28196680000019114, + "n": 1251, + "solve_s": 0.28196680000019114, + "outer_iterations": 37, + "inner_iterations": 6, + "total_iterations": 43, + "n_passive": 974, + "warm_start_errors": 75, + "index": 10 + }, + { + "warm_start_error_fraction": 0.06634692246203037, + "reconstruction_s": 0.5908419000006688, + "eval_s": 0.6894056000001001, + "log_likelihood": 1984.5416657187025, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2569220999994286, + "n": 1251, + "solve_s": 0.2569220999994286, + "outer_iterations": 60, + "inner_iterations": 2, + "total_iterations": 62, + "n_passive": 995, + "warm_start_errors": 83, + "index": 11 + }, + { + "warm_start_error_fraction": 0.03597122302158273, + "reconstruction_s": 0.3791174999996656, + "eval_s": 0.5181928999991214, + "log_likelihood": 2455.0228358943714, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1548821999995198, + "n": 1251, + "solve_s": 0.1548821999995198, + "outer_iterations": 31, + "inner_iterations": 3, + "total_iterations": 34, + "n_passive": 996, + "warm_start_errors": 45, + "index": 12 + }, + { + "warm_start_error_fraction": 0.04156674660271783, + "reconstruction_s": 0.7033156999996208, + "eval_s": 0.8673425999995743, + "log_likelihood": 1549.8176167080073, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2481078000000707, + "n": 1251, + "solve_s": 0.2481078000000707, + "outer_iterations": 32, + "inner_iterations": 2, + "total_iterations": 34, + "n_passive": 992, + "warm_start_errors": 52, + "index": 13 + }, + { + "warm_start_error_fraction": 0.03996802557953637, + "reconstruction_s": 0.6280768999995416, + "eval_s": 0.7521997000003466, + "log_likelihood": 716.7556351102266, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21909529999993538, + "n": 1251, + "solve_s": 0.21909529999993538, + "outer_iterations": 34, + "inner_iterations": 3, + "total_iterations": 37, + "n_passive": 986, + "warm_start_errors": 50, + "index": 14 + }, + { + "warm_start_error_fraction": 0.037569944044764186, + "reconstruction_s": 0.4848624000005657, + "eval_s": 0.6187363999997615, + "log_likelihood": 2296.9835189518135, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22687990000031277, + "n": 1251, + "solve_s": 0.22687990000031277, + "outer_iterations": 37, + "inner_iterations": 2, + "total_iterations": 39, + "n_passive": 1003, + "warm_start_errors": 47, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06554756195043965, + "reconstruction_s": 0.7963267999994059, + "eval_s": 0.9387720000004265, + "log_likelihood": 2279.074679919747, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4410288999997647, + "n": 1251, + "solve_s": 0.4410288999997647, + "outer_iterations": 53, + "inner_iterations": 5, + "total_iterations": 58, + "n_passive": 1001, + "warm_start_errors": 82, + "index": 16 + }, + { + "warm_start_error_fraction": 0.037569944044764186, + "reconstruction_s": 0.5363575999999739, + "eval_s": 0.6595146999998178, + "log_likelihood": 1638.3558680333863, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20035459999962768, + "n": 1251, + "solve_s": 0.20035459999962768, + "outer_iterations": 31, + "inner_iterations": 4, + "total_iterations": 35, + "n_passive": 996, + "warm_start_errors": 47, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0319744204636291, + "reconstruction_s": 0.5360685000005105, + "eval_s": 0.6885747999995147, + "log_likelihood": 445.6194860795331, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.17259279999962018, + "n": 1251, + "solve_s": 0.17259279999962018, + "outer_iterations": 15, + "inner_iterations": 1, + "total_iterations": 16, + "n_passive": 976, + "warm_start_errors": 40, + "index": 18 + }, + { + "warm_start_error_fraction": 0.06794564348521183, + "reconstruction_s": 0.6326173000006747, + "eval_s": 0.749348800000007, + "log_likelihood": 1470.7554900904634, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30174560000068595, + "n": 1251, + "solve_s": 0.30174560000068595, + "outer_iterations": 59, + "inner_iterations": 4, + "total_iterations": 63, + "n_passive": 987, + "warm_start_errors": 85, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_sersic_light_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_sersic_light_v2026.8.17.1.png new file mode 100644 index 000000000..12ffd5eec Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_sersic_light_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_subhalo_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_subhalo_v2026.8.17.1.json new file mode 100644 index 000000000..ca4e8ebec --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_subhalo_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "subhalo", + "configuration": { + "model": "subhalo", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 14, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 2.7374765000004118, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.06183206106870229, + "mean_warm_start_error_fraction": 0.06259541984732826, + "p90_warm_start_error_fraction": 0.06740458015267176, + "max_warm_start_error_fraction": 0.0717557251908397, + "median_outer_iterations": 62.0, + "mean_outer_iterations": 62.05, + "p90_outer_iterations": 70.0, + "max_outer_iterations": 71.0, + "median_inner_iterations": 14.0, + "mean_inner_iterations": 14.05, + "p90_inner_iterations": 17.0, + "max_inner_iterations": 18.0, + "median_total_iterations": 76.0, + "mean_total_iterations": 76.1, + "p90_total_iterations": 84.0, + "max_total_iterations": 88.0, + "median_n_passive": 1188.0, + "mean_n_passive": 1188.4, + "p90_n_passive": 1192.0, + "max_n_passive": 1193.0, + "median_warm_start_errors": 81.0, + "mean_warm_start_errors": 82.0, + "p90_warm_start_errors": 88.30000000000001, + "max_warm_start_errors": 94.0, + "median_solve_s": 0.4036032500002875, + "mean_solve_s": 0.4604451700001846, + "p90_solve_s": 0.6455924500000038, + "max_solve_s": 0.7732499000003372, + "median_reconstruction_s": 0.9840459500001089, + "mean_reconstruction_s": 1.0865259650001007, + "p90_reconstruction_s": 1.434440619999714, + "max_reconstruction_s": 1.4876538000007713, + "median_eval_s": 1.0526281999996172, + "mean_eval_s": 1.16322361499997, + "p90_eval_s": 1.5402030500004913, + "max_eval_s": 1.602410499999678 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.007633587786259542, + "mean_warm_start_error_fraction": 0.010839694656488551, + "p90_warm_start_error_fraction": 0.016106870229007635, + "max_warm_start_error_fraction": 0.055725190839694655, + "median_outer_iterations": 5.5, + "mean_outer_iterations": 8.95, + "p90_outer_iterations": 13.200000000000003, + "max_outer_iterations": 59.0, + "median_inner_iterations": 2.0, + "mean_inner_iterations": 2.6, + "p90_inner_iterations": 5.0, + "max_inner_iterations": 14.0, + "median_total_iterations": 6.5, + "mean_total_iterations": 11.55, + "p90_total_iterations": 16.1, + "max_total_iterations": 73.0, + "median_n_passive": 1188.0, + "mean_n_passive": 1188.0, + "p90_n_passive": 1192.0, + "max_n_passive": 1193.0, + "median_warm_start_errors": 10.0, + "mean_warm_start_errors": 14.2, + "p90_warm_start_errors": 21.1, + "max_warm_start_errors": 73.0, + "median_solve_s": 0.07093154999984108, + "mean_solve_s": 0.0915008499999658, + "p90_solve_s": 0.10920226999987798, + "max_solve_s": 0.3825311000000511, + "median_reconstruction_s": 0.6542952000004334, + "mean_reconstruction_s": 0.6882919649998712, + "p90_reconstruction_s": 0.8228681499997947, + "max_reconstruction_s": 0.956489999999576, + "median_eval_s": 0.730627750000167, + "mean_eval_s": 0.7714512450000257, + "p90_eval_s": 0.9106982099998278, + "max_eval_s": 1.0636429999995016 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.4510589374701657e-14, + "max_abs_reconstruction_deviation": 7.97813548558679e-09 + }, + "iteration_reduction_ratio": 11.692307692307692, + "solve_speedup_ratio": 5.690038494875577, + "eval_speedup_ratio": 1.4407175199674207, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 1.120085999999901, + "eval_s": 1.2114232999992964, + "log_likelihood": 7200.585539058202, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5212651000001642, + "n": 1310, + "solve_s": 0.5212651000001642, + "outer_iterations": 59, + "inner_iterations": 14, + "total_iterations": 73, + "n_passive": 1192, + "warm_start_errors": 73, + "index": 0 + }, + { + "warm_start_error_fraction": 0.06259541984732825, + "reconstruction_s": 0.9883221000000049, + "eval_s": 1.0602523999996265, + "log_likelihood": 7094.239448673183, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4006864000002679, + "n": 1310, + "solve_s": 0.4006864000002679, + "outer_iterations": 63, + "inner_iterations": 13, + "total_iterations": 76, + "n_passive": 1188, + "warm_start_errors": 82, + "index": 1 + }, + { + "warm_start_error_fraction": 0.06183206106870229, + "reconstruction_s": 0.9415749000008873, + "eval_s": 1.013500699999895, + "log_likelihood": 7078.344806804799, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38785180000013497, + "n": 1310, + "solve_s": 0.38785180000013497, + "outer_iterations": 68, + "inner_iterations": 16, + "total_iterations": 84, + "n_passive": 1189, + "warm_start_errors": 81, + "index": 2 + }, + { + "warm_start_error_fraction": 0.05801526717557252, + "reconstruction_s": 0.9797698000002129, + "eval_s": 1.045003999999608, + "log_likelihood": 7048.388123173122, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3785869000003004, + "n": 1310, + "solve_s": 0.3785869000003004, + "outer_iterations": 55, + "inner_iterations": 10, + "total_iterations": 65, + "n_passive": 1191, + "warm_start_errors": 76, + "index": 3 + }, + { + "warm_start_error_fraction": 0.06183206106870229, + "reconstruction_s": 0.8571173000000272, + "eval_s": 0.9264259000001402, + "log_likelihood": 7018.1194730618745, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3513233000003311, + "n": 1310, + "solve_s": 0.3513233000003311, + "outer_iterations": 59, + "inner_iterations": 14, + "total_iterations": 73, + "n_passive": 1188, + "warm_start_errors": 81, + "index": 4 + }, + { + "warm_start_error_fraction": 0.059541984732824425, + "reconstruction_s": 0.9199729000001753, + "eval_s": 0.9815377999993871, + "log_likelihood": 7026.906465862045, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3942389000003459, + "n": 1310, + "solve_s": 0.3942389000003459, + "outer_iterations": 71, + "inner_iterations": 17, + "total_iterations": 88, + "n_passive": 1193, + "warm_start_errors": 78, + "index": 5 + }, + { + "warm_start_error_fraction": 0.059541984732824425, + "reconstruction_s": 0.9563115999999354, + "eval_s": 1.023275399999875, + "log_likelihood": 6978.947006271124, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39085640000030253, + "n": 1310, + "solve_s": 0.39085640000030253, + "outer_iterations": 69, + "inner_iterations": 15, + "total_iterations": 84, + "n_passive": 1187, + "warm_start_errors": 78, + "index": 6 + }, + { + "warm_start_error_fraction": 0.0648854961832061, + "reconstruction_s": 0.9552389000000403, + "eval_s": 1.0251558999998451, + "log_likelihood": 6957.412576711931, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39122860000043147, + "n": 1310, + "solve_s": 0.39122860000043147, + "outer_iterations": 70, + "inner_iterations": 14, + "total_iterations": 84, + "n_passive": 1187, + "warm_start_errors": 85, + "index": 7 + }, + { + "warm_start_error_fraction": 0.06564885496183206, + "reconstruction_s": 0.897840499999802, + "eval_s": 0.9585696000003736, + "log_likelihood": 6984.318471225175, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.40652010000030714, + "n": 1310, + "solve_s": 0.40652010000030714, + "outer_iterations": 69, + "inner_iterations": 12, + "total_iterations": 81, + "n_passive": 1188, + "warm_start_errors": 86, + "index": 8 + }, + { + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 0.9251117000003433, + "eval_s": 0.9890305999997508, + "log_likelihood": 6932.66845955451, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37565360000007786, + "n": 1310, + "solve_s": 0.37565360000007786, + "outer_iterations": 68, + "inner_iterations": 12, + "total_iterations": 80, + "n_passive": 1186, + "warm_start_errors": 87, + "index": 9 + }, + { + "warm_start_error_fraction": 0.06946564885496183, + "reconstruction_s": 0.9621906000002127, + "eval_s": 1.0238749000000098, + "log_likelihood": 6908.895182297787, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38093449999996665, + "n": 1310, + "solve_s": 0.38093449999996665, + "outer_iterations": 64, + "inner_iterations": 13, + "total_iterations": 77, + "n_passive": 1181, + "warm_start_errors": 91, + "index": 10 + }, + { + "warm_start_error_fraction": 0.0717557251908397, + "reconstruction_s": 0.9365238000000318, + "eval_s": 1.010593300000437, + "log_likelihood": 6829.589790684989, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4580090999997992, + "n": 1310, + "solve_s": 0.4580090999997992, + "outer_iterations": 70, + "inner_iterations": 12, + "total_iterations": 82, + "n_passive": 1186, + "warm_start_errors": 94, + "index": 11 + }, + { + "warm_start_error_fraction": 0.06564885496183206, + "reconstruction_s": 1.0449853000000076, + "eval_s": 1.1185258000004978, + "log_likelihood": 6956.7001115295425, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3903520000003482, + "n": 1310, + "solve_s": 0.3903520000003482, + "outer_iterations": 56, + "inner_iterations": 12, + "total_iterations": 68, + "n_passive": 1188, + "warm_start_errors": 86, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 1.4303511999996772, + "eval_s": 1.5387513000005129, + "log_likelihood": 6957.25785424003, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7732499000003372, + "n": 1310, + "solve_s": 0.7732499000003372, + "outer_iterations": 62, + "inner_iterations": 16, + "total_iterations": 78, + "n_passive": 1188, + "warm_start_errors": 77, + "index": 13 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 1.4876538000007713, + "eval_s": 1.602410499999678, + "log_likelihood": 6941.3561431213375, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6431695999999647, + "n": 1310, + "solve_s": 0.6431695999999647, + "outer_iterations": 50, + "inner_iterations": 10, + "total_iterations": 60, + "n_passive": 1192, + "warm_start_errors": 75, + "index": 14 + }, + { + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 1.4712454000000434, + "eval_s": 1.5532688000002963, + "log_likelihood": 6829.010053928672, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5461961000000883, + "n": 1310, + "solve_s": 0.5461961000000883, + "outer_iterations": 58, + "inner_iterations": 16, + "total_iterations": 74, + "n_passive": 1184, + "warm_start_errors": 88, + "index": 15 + }, + { + "warm_start_error_fraction": 0.0648854961832061, + "reconstruction_s": 1.4109815000001618, + "eval_s": 1.5205525999999736, + "log_likelihood": 6930.197831915088, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.667398100000355, + "n": 1310, + "solve_s": 0.667398100000355, + "outer_iterations": 62, + "inner_iterations": 14, + "total_iterations": 76, + "n_passive": 1191, + "warm_start_errors": 85, + "index": 16 + }, + { + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 1.2396355999999287, + "eval_s": 1.3162653999997929, + "log_likelihood": 6926.551165134853, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.47043829999984155, + "n": 1310, + "solve_s": 0.47043829999984155, + "outer_iterations": 57, + "inner_iterations": 17, + "total_iterations": 74, + "n_passive": 1190, + "warm_start_errors": 80, + "index": 17 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 1.0557533999999578, + "eval_s": 1.1238110000003871, + "log_likelihood": 6914.750455418389, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.455846300000303, + "n": 1310, + "solve_s": 0.455846300000303, + "outer_iterations": 56, + "inner_iterations": 18, + "total_iterations": 74, + "n_passive": 1189, + "warm_start_errors": 79, + "index": 18 + }, + { + "warm_start_error_fraction": 0.059541984732824425, + "reconstruction_s": 1.1498529999998937, + "eval_s": 1.2222431000000142, + "log_likelihood": 6828.720082250775, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4250984000000244, + "n": 1310, + "solve_s": 0.4250984000000244, + "outer_iterations": 55, + "inner_iterations": 16, + "total_iterations": 71, + "n_passive": 1190, + "warm_start_errors": 78, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.8826357999996617, + "eval_s": 0.9524448000001939, + "log_likelihood": 7200.585539058202, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3825311000000511, + "n": 1310, + "solve_s": 0.3825311000000511, + "outer_iterations": 59, + "inner_iterations": 14, + "total_iterations": 73, + "n_passive": 1192, + "warm_start_errors": 73, + "index": 0 + }, + { + "warm_start_error_fraction": 0.011450381679389313, + "reconstruction_s": 0.6790793000000122, + "eval_s": 0.7823502999999619, + "log_likelihood": 7094.239448673157, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.15773449999960576, + "n": 1310, + "solve_s": 0.15773449999960576, + "outer_iterations": 11, + "inner_iterations": 5, + "total_iterations": 16, + "n_passive": 1187, + "warm_start_errors": 15, + "index": 1 + }, + { + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 0.6942541999997047, + "eval_s": 0.8003638000000137, + "log_likelihood": 7078.344806804865, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06159939999997732, + "n": 1310, + "solve_s": 0.06159939999997732, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1188, + "warm_start_errors": 3, + "index": 2 + }, + { + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 0.956489999999576, + "eval_s": 1.0636429999995016, + "log_likelihood": 7048.388123173132, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08624370000052295, + "n": 1310, + "solve_s": 0.08624370000052295, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1190, + "warm_start_errors": 4, + "index": 3 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.7932170999993104, + "eval_s": 0.889638599999671, + "log_likelihood": 7018.119473061857, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07347029999982624, + "n": 1310, + "solve_s": 0.07347029999982624, + "outer_iterations": 4, + "inner_iterations": 2, + "total_iterations": 6, + "n_passive": 1188, + "warm_start_errors": 10, + "index": 4 + }, + { + "warm_start_error_fraction": 0.01603053435114504, + "reconstruction_s": 0.7167679000003773, + "eval_s": 0.7929741999996622, + "log_likelihood": 7026.906465862083, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10380979999990814, + "n": 1310, + "solve_s": 0.10380979999990814, + "outer_iterations": 15, + "inner_iterations": 2, + "total_iterations": 17, + "n_passive": 1193, + "warm_start_errors": 21, + "index": 5 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.661419100000785, + "eval_s": 0.7379933000001984, + "log_likelihood": 6978.9470062711025, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05699519999961922, + "n": 1310, + "solve_s": 0.05699519999961922, + "outer_iterations": 2, + "inner_iterations": 3, + "total_iterations": 5, + "n_passive": 1187, + "warm_start_errors": 10, + "index": 6 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.7015879000000496, + "eval_s": 0.7839407999999821, + "log_likelihood": 6957.412576711922, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06980980000025738, + "n": 1310, + "solve_s": 0.06980980000025738, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1186, + "warm_start_errors": 9, + "index": 7 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.6028011999997034, + "eval_s": 0.68153059999986, + "log_likelihood": 6984.318471225165, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06860560000041005, + "n": 1310, + "solve_s": 0.06860560000041005, + "outer_iterations": 6, + "inner_iterations": 2, + "total_iterations": 8, + "n_passive": 1188, + "warm_start_errors": 8, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.6364264999992884, + "eval_s": 0.7096971000000849, + "log_likelihood": 6932.668459554512, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06311159999950178, + "n": 1310, + "solve_s": 0.06311159999950178, + "outer_iterations": 3, + "inner_iterations": 3, + "total_iterations": 6, + "n_passive": 1186, + "warm_start_errors": 8, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0053435114503816794, + "reconstruction_s": 0.6038060000000769, + "eval_s": 0.6818452999996225, + "log_likelihood": 6908.895182297797, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.050800700000763754, + "n": 1310, + "solve_s": 0.050800700000763754, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1181, + "warm_start_errors": 7, + "index": 10 + }, + { + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 0.6250347999994119, + "eval_s": 0.7003921000004993, + "log_likelihood": 6829.589790684988, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07924099999945611, + "n": 1310, + "solve_s": 0.07924099999945611, + "outer_iterations": 9, + "inner_iterations": 2, + "total_iterations": 11, + "n_passive": 1185, + "warm_start_errors": 14, + "index": 11 + }, + { + "warm_start_error_fraction": 0.01603053435114504, + "reconstruction_s": 0.6471713000000818, + "eval_s": 0.7232622000001356, + "log_likelihood": 6956.700111529559, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10130829999980051, + "n": 1310, + "solve_s": 0.10130829999980051, + "outer_iterations": 13, + "inner_iterations": 1, + "total_iterations": 14, + "n_passive": 1188, + "warm_start_errors": 21, + "index": 12 + }, + { + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 0.6329365999999936, + "eval_s": 0.7158107000004748, + "log_likelihood": 6957.257854240131, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07209919999968406, + "n": 1310, + "solve_s": 0.07209919999968406, + "outer_iterations": 9, + "inner_iterations": 3, + "total_iterations": 12, + "n_passive": 1188, + "warm_start_errors": 14, + "index": 13 + }, + { + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 0.6129863999994996, + "eval_s": 0.6840825999997833, + "log_likelihood": 6941.35614312134, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06815989999995509, + "n": 1310, + "solve_s": 0.06815989999995509, + "outer_iterations": 11, + "inner_iterations": 1, + "total_iterations": 12, + "n_passive": 1192, + "warm_start_errors": 14, + "index": 14 + }, + { + "warm_start_error_fraction": 0.016793893129770993, + "reconstruction_s": 0.6095086999994237, + "eval_s": 0.6947101000005205, + "log_likelihood": 6829.010053928681, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08887179999965156, + "n": 1310, + "solve_s": 0.08887179999965156, + "outer_iterations": 8, + "inner_iterations": 5, + "total_iterations": 13, + "n_passive": 1184, + "warm_start_errors": 22, + "index": 15 + }, + { + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 0.6311353000000963, + "eval_s": 0.6984824000001026, + "log_likelihood": 6930.197831915118, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049840100000437815, + "n": 1310, + "solve_s": 0.049840100000437815, + "outer_iterations": 6, + "inner_iterations": 0, + "total_iterations": 6, + "n_passive": 1190, + "warm_start_errors": 6, + "index": 16 + }, + { + "warm_start_error_fraction": 0.008396946564885497, + "reconstruction_s": 0.5912797999999384, + "eval_s": 0.6781480999998166, + "log_likelihood": 6926.551165134855, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07205329999942478, + "n": 1310, + "solve_s": 0.07205329999942478, + "outer_iterations": 5, + "inner_iterations": 2, + "total_iterations": 7, + "n_passive": 1189, + "warm_start_errors": 11, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 0.6710741000006237, + "eval_s": 0.7516552000006413, + "log_likelihood": 6914.7504554183815, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.058379900000545604, + "n": 1310, + "solve_s": 0.058379900000545604, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 1189, + "warm_start_errors": 4, + "index": 18 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.8162272999998095, + "eval_s": 0.906059699999787, + "log_likelihood": 6828.720082250775, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0653517999999167, + "n": 1310, + "solve_s": 0.0653517999999167, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 1189, + "warm_start_errors": 10, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.07442748091603053, + "mean_warm_start_error_fraction": 0.07431297709923664, + "p90_warm_start_error_fraction": 0.0900763358778626, + "max_warm_start_error_fraction": 0.09770992366412214, + "median_outer_iterations": 75.0, + "mean_outer_iterations": 74.5, + "p90_outer_iterations": 94.70000000000002, + "max_outer_iterations": 102.0, + "median_inner_iterations": 15.0, + "mean_inner_iterations": 15.1, + "p90_inner_iterations": 20.1, + "max_inner_iterations": 22.0, + "median_total_iterations": 87.5, + "mean_total_iterations": 89.6, + "p90_total_iterations": 114.30000000000001, + "max_total_iterations": 123.0, + "median_n_passive": 1175.0, + "mean_n_passive": 1172.9, + "p90_n_passive": 1192.1, + "max_n_passive": 1195.0, + "median_warm_start_errors": 97.5, + "mean_warm_start_errors": 97.35, + "p90_warm_start_errors": 118.0, + "max_warm_start_errors": 128.0, + "median_solve_s": 0.5802484000000732, + "mean_solve_s": 0.5724114799998461, + "p90_solve_s": 0.7186797699999261, + "max_solve_s": 0.7976369999996678, + "median_reconstruction_s": 1.3058789500000785, + "mean_reconstruction_s": 1.2956204000000526, + "p90_reconstruction_s": 1.5525849600003312, + "max_reconstruction_s": 1.8097600999999486, + "median_eval_s": 1.4082094999998844, + "mean_eval_s": 1.385332185000061, + "p90_eval_s": 1.6611958899997261, + "max_eval_s": 1.9175733000001856 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.07251908396946566, + "mean_warm_start_error_fraction": 0.07423664122137404, + "p90_warm_start_error_fraction": 0.09564885496183206, + "max_warm_start_error_fraction": 0.09770992366412214, + "median_outer_iterations": 78.5, + "mean_outer_iterations": 79.9, + "p90_outer_iterations": 102.60000000000001, + "max_outer_iterations": 121.0, + "median_inner_iterations": 17.0, + "mean_inner_iterations": 18.25, + "p90_inner_iterations": 28.0, + "max_inner_iterations": 32.0, + "median_total_iterations": 94.0, + "mean_total_iterations": 98.15, + "p90_total_iterations": 124.60000000000002, + "max_total_iterations": 140.0, + "median_n_passive": 1174.5, + "mean_n_passive": 1172.7, + "p90_n_passive": 1192.1, + "max_n_passive": 1195.0, + "median_warm_start_errors": 95.0, + "mean_warm_start_errors": 97.25, + "p90_warm_start_errors": 125.30000000000001, + "max_warm_start_errors": 128.0, + "median_solve_s": 0.3703699999996388, + "mean_solve_s": 0.37541384500004826, + "p90_solve_s": 0.51596780999962, + "max_solve_s": 0.6096664000006058, + "median_reconstruction_s": 0.9098219499996958, + "mean_reconstruction_s": 0.9473985600000561, + "p90_reconstruction_s": 1.173711980000644, + "max_reconstruction_s": 1.2459149999995134, + "median_eval_s": 0.9822666499999286, + "mean_eval_s": 1.0266241600001194, + "p90_eval_s": 1.2610422500001732, + "max_eval_s": 1.4035745999999563 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 6.411290636331018e-15, + "max_abs_reconstruction_deviation": 2.2606099214783626e-09 + }, + "iteration_reduction_ratio": 0.9308510638297872, + "solve_speedup_ratio": 1.566672246673972, + "eval_speedup_ratio": 1.43363260882367, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.09770992366412214, + "reconstruction_s": 1.383552299999792, + "eval_s": 1.4565745999998398, + "log_likelihood": 5464.283580084498, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6396687999995265, + "n": 1310, + "solve_s": 0.6396687999995265, + "outer_iterations": 102, + "inner_iterations": 21, + "total_iterations": 123, + "n_passive": 1146, + "warm_start_errors": 128, + "index": 0 + }, + { + "warm_start_error_fraction": 0.08396946564885496, + "reconstruction_s": 1.1759739999997691, + "eval_s": 1.277781099999629, + "log_likelihood": 6079.625151806911, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5535839000003762, + "n": 1310, + "solve_s": 0.5535839000003762, + "outer_iterations": 77, + "inner_iterations": 12, + "total_iterations": 89, + "n_passive": 1155, + "warm_start_errors": 110, + "index": 1 + }, + { + "warm_start_error_fraction": 0.06793893129770992, + "reconstruction_s": 1.4351219999998648, + "eval_s": 1.5319650000001275, + "log_likelihood": 6809.197735899768, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6045561999999336, + "n": 1310, + "solve_s": 0.6045561999999336, + "outer_iterations": 94, + "inner_iterations": 15, + "total_iterations": 109, + "n_passive": 1189, + "warm_start_errors": 89, + "index": 2 + }, + { + "warm_start_error_fraction": 0.08931297709923663, + "reconstruction_s": 1.2390766000007716, + "eval_s": 1.3382970000002388, + "log_likelihood": 6686.451585403479, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5858106000005137, + "n": 1310, + "solve_s": 0.5858106000005137, + "outer_iterations": 86, + "inner_iterations": 16, + "total_iterations": 102, + "n_passive": 1171, + "warm_start_errors": 117, + "index": 3 + }, + { + "warm_start_error_fraction": 0.07862595419847328, + "reconstruction_s": 1.4508334000001923, + "eval_s": 1.556764099999782, + "log_likelihood": 6801.730263960315, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7544337999997879, + "n": 1310, + "solve_s": 0.7544337999997879, + "outer_iterations": 84, + "inner_iterations": 19, + "total_iterations": 103, + "n_passive": 1178, + "warm_start_errors": 103, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 1.3877235999998447, + "eval_s": 1.4895874999992884, + "log_likelihood": 7417.421943071614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5746861999996327, + "n": 1310, + "solve_s": 0.5746861999996327, + "outer_iterations": 50, + "inner_iterations": 12, + "total_iterations": 62, + "n_passive": 1192, + "warm_start_errors": 72, + "index": 5 + }, + { + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 1.2493322999998782, + "eval_s": 1.3231916000004276, + "log_likelihood": 6601.8881017864815, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4961712999993324, + "n": 1310, + "solve_s": 0.4961712999993324, + "outer_iterations": 61, + "inner_iterations": 13, + "total_iterations": 74, + "n_passive": 1174, + "warm_start_errors": 87, + "index": 6 + }, + { + "warm_start_error_fraction": 0.06793893129770992, + "reconstruction_s": 0.9720938999998907, + "eval_s": 1.0408696000004056, + "log_likelihood": 6637.590793199195, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42069590000028256, + "n": 1310, + "solve_s": 0.42069590000028256, + "outer_iterations": 58, + "inner_iterations": 14, + "total_iterations": 72, + "n_passive": 1176, + "warm_start_errors": 89, + "index": 7 + }, + { + "warm_start_error_fraction": 0.08549618320610687, + "reconstruction_s": 0.9533145000004879, + "eval_s": 1.0208246000001964, + "log_likelihood": 6269.399936359759, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41473649999988993, + "n": 1310, + "solve_s": 0.41473649999988993, + "outer_iterations": 84, + "inner_iterations": 10, + "total_iterations": 94, + "n_passive": 1164, + "warm_start_errors": 112, + "index": 8 + }, + { + "warm_start_error_fraction": 0.07022900763358779, + "reconstruction_s": 1.0159938999995575, + "eval_s": 1.0959807999997793, + "log_likelihood": 6274.172025124486, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43577749999985826, + "n": 1310, + "solve_s": 0.43577749999985826, + "outer_iterations": 67, + "inner_iterations": 15, + "total_iterations": 82, + "n_passive": 1183, + "warm_start_errors": 92, + "index": 9 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 1.3131342999995468, + "eval_s": 1.445433700000649, + "log_likelihood": 7401.9321941412645, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.602846299999328, + "n": 1310, + "solve_s": 0.602846299999328, + "outer_iterations": 48, + "inner_iterations": 16, + "total_iterations": 64, + "n_passive": 1190, + "warm_start_errors": 65, + "index": 10 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 1.8097600999999486, + "eval_s": 1.9175733000001856, + "log_likelihood": 7192.52358346913, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7147070999999414, + "n": 1310, + "solve_s": 0.7147070999999414, + "outer_iterations": 77, + "inner_iterations": 20, + "total_iterations": 97, + "n_passive": 1195, + "warm_start_errors": 75, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0900763358778626, + "reconstruction_s": 1.5715215000000171, + "eval_s": 1.6693371999999727, + "log_likelihood": 6082.3179780646315, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6835519000005661, + "n": 1310, + "solve_s": 0.6835519000005661, + "outer_iterations": 89, + "inner_iterations": 16, + "total_iterations": 105, + "n_passive": 1167, + "warm_start_errors": 118, + "index": 12 + }, + { + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 1.3488279000002876, + "eval_s": 1.4432025999994949, + "log_likelihood": 5756.225977266235, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5916909999996278, + "n": 1310, + "solve_s": 0.5916909999996278, + "outer_iterations": 73, + "inner_iterations": 13, + "total_iterations": 86, + "n_passive": 1148, + "warm_start_errors": 115, + "index": 13 + }, + { + "warm_start_error_fraction": 0.0900763358778626, + "reconstruction_s": 1.550480900000366, + "eval_s": 1.6602912999996988, + "log_likelihood": 5666.107090340166, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7976369999996678, + "n": 1310, + "solve_s": 0.7976369999996678, + "outer_iterations": 92, + "inner_iterations": 22, + "total_iterations": 114, + "n_passive": 1149, + "warm_start_errors": 118, + "index": 14 + }, + { + "warm_start_error_fraction": 0.083206106870229, + "reconstruction_s": 1.4947331000003032, + "eval_s": 1.5776098000005732, + "log_likelihood": 5804.330690551785, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7085767999997188, + "n": 1310, + "solve_s": 0.7085767999997188, + "outer_iterations": 101, + "inner_iterations": 16, + "total_iterations": 117, + "n_passive": 1165, + "warm_start_errors": 109, + "index": 15 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 1.2986236000006102, + "eval_s": 1.373216400000274, + "log_likelihood": 7243.68310542631, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5179354999991119, + "n": 1310, + "solve_s": 0.5179354999991119, + "outer_iterations": 50, + "inner_iterations": 13, + "total_iterations": 63, + "n_passive": 1193, + "warm_start_errors": 75, + "index": 16 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 1.2346865999998045, + "eval_s": 1.3040269000002809, + "log_likelihood": 6957.9566782621705, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5138147999996363, + "n": 1310, + "solve_s": 0.5138147999996363, + "outer_iterations": 62, + "inner_iterations": 18, + "total_iterations": 80, + "n_passive": 1185, + "warm_start_errors": 79, + "index": 17 + }, + { + "warm_start_error_fraction": 0.07938931297709924, + "reconstruction_s": 1.0107723999999507, + "eval_s": 1.0918872000002011, + "log_likelihood": 6616.329964407656, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4715230000001611, + "n": 1310, + "solve_s": 0.4715230000001611, + "outer_iterations": 72, + "inner_iterations": 12, + "total_iterations": 84, + "n_passive": 1155, + "warm_start_errors": 104, + "index": 18 + }, + { + "warm_start_error_fraction": 0.06870229007633588, + "reconstruction_s": 1.0168511000001672, + "eval_s": 1.0922294000001784, + "log_likelihood": 7182.922429006099, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3658255000000281, + "n": 1310, + "solve_s": 0.3658255000000281, + "outer_iterations": 63, + "inner_iterations": 9, + "total_iterations": 72, + "n_passive": 1183, + "warm_start_errors": 90, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.09770992366412214, + "reconstruction_s": 1.2459149999995134, + "eval_s": 1.3181540000005043, + "log_likelihood": 5464.283580084498, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5848071000000346, + "n": 1310, + "solve_s": 0.5848071000000346, + "outer_iterations": 102, + "inner_iterations": 21, + "total_iterations": 123, + "n_passive": 1146, + "warm_start_errors": 128, + "index": 0 + }, + { + "warm_start_error_fraction": 0.09541984732824428, + "reconstruction_s": 1.1101077000002988, + "eval_s": 1.1820676000006642, + "log_likelihood": 6079.62515180693, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5044551999999385, + "n": 1310, + "solve_s": 0.5044551999999385, + "outer_iterations": 96, + "inner_iterations": 23, + "total_iterations": 119, + "n_passive": 1155, + "warm_start_errors": 125, + "index": 1 + }, + { + "warm_start_error_fraction": 0.07633587786259542, + "reconstruction_s": 0.8873739000000569, + "eval_s": 0.9508050000004005, + "log_likelihood": 6809.197735899812, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.36768299999948795, + "n": 1310, + "solve_s": 0.36768299999948795, + "outer_iterations": 98, + "inner_iterations": 20, + "total_iterations": 118, + "n_passive": 1189, + "warm_start_errors": 100, + "index": 2 + }, + { + "warm_start_error_fraction": 0.06564885496183206, + "reconstruction_s": 0.7897607000004427, + "eval_s": 0.8669089000004533, + "log_likelihood": 6686.451585403482, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28373999999985244, + "n": 1310, + "solve_s": 0.28373999999985244, + "outer_iterations": 54, + "inner_iterations": 11, + "total_iterations": 65, + "n_passive": 1171, + "warm_start_errors": 86, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0633587786259542, + "reconstruction_s": 0.8467154999998456, + "eval_s": 0.9162888999999268, + "log_likelihood": 6801.73026396028, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30918950000068435, + "n": 1310, + "solve_s": 0.30918950000068435, + "outer_iterations": 66, + "inner_iterations": 20, + "total_iterations": 86, + "n_passive": 1178, + "warm_start_errors": 83, + "index": 4 + }, + { + "warm_start_error_fraction": 0.059541984732824425, + "reconstruction_s": 0.9121875999999247, + "eval_s": 0.9841989999995349, + "log_likelihood": 7417.421943071615, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3039153999998234, + "n": 1310, + "solve_s": 0.3039153999998234, + "outer_iterations": 61, + "inner_iterations": 14, + "total_iterations": 75, + "n_passive": 1192, + "warm_start_errors": 78, + "index": 5 + }, + { + "warm_start_error_fraction": 0.06793893129770992, + "reconstruction_s": 0.8555087000004278, + "eval_s": 0.9205872000002273, + "log_likelihood": 6601.888101786487, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35046130000046105, + "n": 1310, + "solve_s": 0.35046130000046105, + "outer_iterations": 86, + "inner_iterations": 12, + "total_iterations": 98, + "n_passive": 1173, + "warm_start_errors": 89, + "index": 6 + }, + { + "warm_start_error_fraction": 0.08625954198473282, + "reconstruction_s": 0.8865445000001273, + "eval_s": 0.9604382999996233, + "log_likelihood": 6637.590793199215, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3756381000002875, + "n": 1310, + "solve_s": 0.3756381000002875, + "outer_iterations": 94, + "inner_iterations": 28, + "total_iterations": 122, + "n_passive": 1176, + "warm_start_errors": 113, + "index": 7 + }, + { + "warm_start_error_fraction": 0.07938931297709924, + "reconstruction_s": 0.9853872000003321, + "eval_s": 1.068279999999504, + "log_likelihood": 6269.399936359754, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3730569999997897, + "n": 1310, + "solve_s": 0.3730569999997897, + "outer_iterations": 71, + "inner_iterations": 14, + "total_iterations": 85, + "n_passive": 1164, + "warm_start_errors": 104, + "index": 8 + }, + { + "warm_start_error_fraction": 0.08854961832061069, + "reconstruction_s": 0.9403664999999819, + "eval_s": 1.0042338000002928, + "log_likelihood": 6274.1720251244815, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4287880000001678, + "n": 1310, + "solve_s": 0.4287880000001678, + "outer_iterations": 121, + "inner_iterations": 18, + "total_iterations": 139, + "n_passive": 1182, + "warm_start_errors": 116, + "index": 9 + }, + { + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 0.9074562999994669, + "eval_s": 0.9803343000003224, + "log_likelihood": 7401.9321941412645, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37588919999961945, + "n": 1310, + "solve_s": 0.37588919999961945, + "outer_iterations": 84, + "inner_iterations": 28, + "total_iterations": 112, + "n_passive": 1190, + "warm_start_errors": 88, + "index": 10 + }, + { + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.8343383999999787, + "eval_s": 0.9087564999999813, + "log_likelihood": 7192.523583469125, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25176030000056926, + "n": 1310, + "solve_s": 0.25176030000056926, + "outer_iterations": 53, + "inner_iterations": 10, + "total_iterations": 63, + "n_passive": 1195, + "warm_start_errors": 73, + "index": 11 + }, + { + "warm_start_error_fraction": 0.08244274809160305, + "reconstruction_s": 0.9369692999998733, + "eval_s": 1.0161749999997483, + "log_likelihood": 6082.3179780646315, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3790281000001414, + "n": 1310, + "solve_s": 0.3790281000001414, + "outer_iterations": 73, + "inner_iterations": 17, + "total_iterations": 90, + "n_passive": 1167, + "warm_start_errors": 108, + "index": 12 + }, + { + "warm_start_error_fraction": 0.09770992366412214, + "reconstruction_s": 0.8746639000000869, + "eval_s": 0.9347250999999233, + "log_likelihood": 5756.225977266254, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35407740000027843, + "n": 1310, + "solve_s": 0.35407740000027843, + "outer_iterations": 108, + "inner_iterations": 32, + "total_iterations": 140, + "n_passive": 1147, + "warm_start_errors": 128, + "index": 13 + }, + { + "warm_start_error_fraction": 0.07862595419847328, + "reconstruction_s": 0.7797380000001795, + "eval_s": 0.8391621000000669, + "log_likelihood": 5666.107090340168, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21701169999960257, + "n": 1310, + "solve_s": 0.21701169999960257, + "outer_iterations": 67, + "inner_iterations": 12, + "total_iterations": 79, + "n_passive": 1148, + "warm_start_errors": 103, + "index": 14 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 0.729218899999978, + "eval_s": 0.7992953000002672, + "log_likelihood": 5804.330690551776, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2450171999998929, + "n": 1310, + "solve_s": 0.2450171999998929, + "outer_iterations": 56, + "inner_iterations": 8, + "total_iterations": 64, + "n_passive": 1165, + "warm_start_errors": 79, + "index": 15 + }, + { + "warm_start_error_fraction": 0.07633587786259542, + "reconstruction_s": 1.1386271999999735, + "eval_s": 1.2293355000001611, + "log_likelihood": 7243.68310542634, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5083189999995739, + "n": 1310, + "solve_s": 0.5083189999995739, + "outer_iterations": 90, + "inner_iterations": 28, + "total_iterations": 118, + "n_passive": 1193, + "warm_start_errors": 100, + "index": 16 + }, + { + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 0.9153648999999859, + "eval_s": 0.9944656000006944, + "log_likelihood": 6957.95667826217, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26760860000013054, + "n": 1310, + "solve_s": 0.26760860000013054, + "outer_iterations": 56, + "inner_iterations": 17, + "total_iterations": 73, + "n_passive": 1185, + "warm_start_errors": 88, + "index": 17 + }, + { + "warm_start_error_fraction": 0.06870229007633588, + "reconstruction_s": 1.2010528999999224, + "eval_s": 1.4035745999999563, + "log_likelihood": 6616.32996440764, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6096664000006058, + "n": 1310, + "solve_s": 0.6096664000006058, + "outer_iterations": 93, + "inner_iterations": 16, + "total_iterations": 109, + "n_passive": 1155, + "warm_start_errors": 90, + "index": 18 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 1.1706741000007241, + "eval_s": 1.2546965000001364, + "log_likelihood": 7182.922429006094, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4181644000000233, + "n": 1310, + "solve_s": 0.4181644000000233, + "outer_iterations": 69, + "inner_iterations": 16, + "total_iterations": 85, + "n_passive": 1183, + "warm_start_errors": 66, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_subhalo_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_subhalo_v2026.8.17.1.png new file mode 100644 index 000000000..1913bc55a Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_subhalo_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_v2026.8.17.1.json new file mode 100644 index 000000000..b41788b33 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_v2026.8.17.1.json @@ -0,0 +1,1767 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "euclid", + "model": "fiducial", + "configuration": { + "model": "fiducial", + "dataset": "euclid", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.1, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 3841, + "over_sampled_pixels": 4468, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 2.0711403000004793, + "total_memo_seed_retries": 0, + "total_warm_start_fallbacks": 0, + "nnls_warm_start_error_tolerance": 1.5, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.05648854961832061, + "mean_warm_start_error_fraction": 0.056832061068702276, + "p90_warm_start_error_fraction": 0.06038167938931298, + "max_warm_start_error_fraction": 0.06412213740458016, + "median_outer_iterations": 58.5, + "mean_outer_iterations": 63.75, + "p90_outer_iterations": 69.0, + "max_outer_iterations": 177.0, + "median_inner_iterations": 11.5, + "mean_inner_iterations": 12.5, + "p90_inner_iterations": 15.200000000000003, + "max_inner_iterations": 29.0, + "median_total_iterations": 70.0, + "mean_total_iterations": 76.25, + "p90_total_iterations": 82.2, + "max_total_iterations": 206.0, + "median_n_passive": 1189.0, + "mean_n_passive": 1190.75, + "p90_n_passive": 1197.3, + "max_n_passive": 1200.0, + "median_warm_start_errors": 74.0, + "mean_warm_start_errors": 74.45, + "p90_warm_start_errors": 79.1, + "max_warm_start_errors": 84.0, + "median_solve_s": 0.2568941500003348, + "mean_solve_s": 0.29206237500020504, + "p90_solve_s": 0.3575439199996254, + "max_solve_s": 0.5988543000003119, + "median_reconstruction_s": 0.6729664499998762, + "mean_reconstruction_s": 0.7014127149999695, + "p90_reconstruction_s": 0.7879641499997888, + "max_reconstruction_s": 0.9778188000000227, + "median_eval_s": 0.7272570999994059, + "mean_eval_s": 0.7551419100000203, + "p90_eval_s": 0.8464794999992592, + "max_eval_s": 1.0335336999996798 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.006870229007633588, + "mean_warm_start_error_fraction": 0.009160305343511452, + "p90_warm_start_error_fraction": 0.010763358778625956, + "max_warm_start_error_fraction": 0.05877862595419847, + "median_outer_iterations": 4.5, + "mean_outer_iterations": 9.6, + "p90_outer_iterations": 16.900000000000013, + "max_outer_iterations": 69.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 2.35, + "p90_inner_iterations": 3.1000000000000014, + "max_inner_iterations": 15.0, + "median_total_iterations": 7.0, + "mean_total_iterations": 11.95, + "p90_total_iterations": 19.100000000000016, + "max_total_iterations": 84.0, + "median_n_passive": 1189.0, + "mean_n_passive": 1190.5, + "p90_n_passive": 1196.4, + "max_n_passive": 1200.0, + "median_warm_start_errors": 9.0, + "mean_warm_start_errors": 12.0, + "p90_warm_start_errors": 14.100000000000001, + "max_warm_start_errors": 77.0, + "median_solve_s": 0.05424290000064502, + "mean_solve_s": 0.07458023000008325, + "p90_solve_s": 0.10705721000049385, + "max_solve_s": 0.35296980000020994, + "median_reconstruction_s": 0.4475397000005614, + "mean_reconstruction_s": 0.4683875850001641, + "p90_reconstruction_s": 0.5187077900000077, + "max_reconstruction_s": 0.7739710000005289, + "median_eval_s": 0.5069869500011919, + "mean_eval_s": 0.5275833249999777, + "p90_eval_s": 0.5756955300006666, + "max_eval_s": 0.8280987999987701 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.1137321561171996e-14, + "max_abs_reconstruction_deviation": 4.839854513688513e-09 + }, + "iteration_reduction_ratio": 10.0, + "solve_speedup_ratio": 4.735995863002899, + "eval_speedup_ratio": 1.434469072621487, + "rows": { + "off": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 0.7291141000005155, + "eval_s": 0.7798621000001731, + "log_likelihood": 7206.2076831241275, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32469289999971807, + "n": 1310, + "solve_s": 0.32469289999971807, + "outer_iterations": 69, + "inner_iterations": 15, + "total_iterations": 84, + "n_passive": 1195, + "warm_start_errors": 77, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 0.660032300000239, + "eval_s": 0.7080792999986443, + "log_likelihood": 7180.432957367257, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25509840000086115, + "n": 1310, + "solve_s": 0.25509840000086115, + "outer_iterations": 61, + "inner_iterations": 11, + "total_iterations": 72, + "n_passive": 1187, + "warm_start_errors": 77, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05801526717557252, + "reconstruction_s": 0.8220997999997053, + "eval_s": 0.8877876999995351, + "log_likelihood": 7143.994889035799, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4047895999992761, + "n": 1310, + "solve_s": 0.4047895999992761, + "outer_iterations": 59, + "inner_iterations": 12, + "total_iterations": 71, + "n_passive": 1188, + "warm_start_errors": 76, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 0.7841712999997981, + "eval_s": 0.8418896999992285, + "log_likelihood": 7182.878722012542, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35229439999966417, + "n": 1310, + "solve_s": 0.35229439999966417, + "outer_iterations": 68, + "inner_iterations": 12, + "total_iterations": 80, + "n_passive": 1190, + "warm_start_errors": 79, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.6560490000010759, + "eval_s": 0.7085649999989982, + "log_likelihood": 7186.1758717889115, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24026620000040566, + "n": 1310, + "solve_s": 0.24026620000040566, + "outer_iterations": 56, + "inner_iterations": 9, + "total_iterations": 65, + "n_passive": 1189, + "warm_start_errors": 72, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.7201297000010527, + "eval_s": 0.7704969000005804, + "log_likelihood": 7145.024669580045, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3372854000008374, + "n": 1310, + "solve_s": 0.3372854000008374, + "outer_iterations": 63, + "inner_iterations": 15, + "total_iterations": 78, + "n_passive": 1189, + "warm_start_errors": 75, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.6468849999982922, + "eval_s": 0.6958778000007442, + "log_likelihood": 7202.126666486725, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2586898999998084, + "n": 1310, + "solve_s": 0.2586898999998084, + "outer_iterations": 58, + "inner_iterations": 11, + "total_iterations": 69, + "n_passive": 1191, + "warm_start_errors": 74, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.7236192999989726, + "eval_s": 0.7736064000000624, + "log_likelihood": 7222.820219978847, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27783260000069276, + "n": 1310, + "solve_s": 0.27783260000069276, + "outer_iterations": 62, + "inner_iterations": 13, + "total_iterations": 75, + "n_passive": 1194, + "warm_start_errors": 73, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.6628464000004897, + "eval_s": 0.7118669999999838, + "log_likelihood": 7245.156725065453, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2534591999992699, + "n": 1310, + "solve_s": 0.2534591999992699, + "outer_iterations": 49, + "inner_iterations": 17, + "total_iterations": 66, + "n_passive": 1186, + "warm_start_errors": 74, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.6224462000009225, + "eval_s": 0.6727979000006599, + "log_likelihood": 7163.555514905283, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20819880000090052, + "n": 1310, + "solve_s": 0.20819880000090052, + "outer_iterations": 43, + "inner_iterations": 12, + "total_iterations": 55, + "n_passive": 1186, + "warm_start_errors": 69, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.6451372999999876, + "eval_s": 0.7091694000009738, + "log_likelihood": 7149.741003169525, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24615319999975327, + "n": 1310, + "solve_s": 0.24615319999975327, + "outer_iterations": 49, + "inner_iterations": 9, + "total_iterations": 58, + "n_passive": 1186, + "warm_start_errors": 74, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05648854961832061, + "reconstruction_s": 0.6641243000012764, + "eval_s": 0.718319400000837, + "log_likelihood": 7135.510856830086, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23362079999969865, + "n": 1310, + "solve_s": 0.23362079999969865, + "outer_iterations": 52, + "inner_iterations": 8, + "total_iterations": 60, + "n_passive": 1187, + "warm_start_errors": 74, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 0.612508499998512, + "eval_s": 0.6623699000010674, + "log_likelihood": 7124.427599165532, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23399419999987003, + "n": 1310, + "solve_s": 0.23399419999987003, + "outer_iterations": 54, + "inner_iterations": 8, + "total_iterations": 62, + "n_passive": 1187, + "warm_start_errors": 77, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.6747557999988203, + "eval_s": 0.7359755999987101, + "log_likelihood": 7138.787138439643, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22379220000038913, + "n": 1310, + "solve_s": 0.22379220000038913, + "outer_iterations": 51, + "inner_iterations": 10, + "total_iterations": 61, + "n_passive": 1189, + "warm_start_errors": 69, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.06412213740458016, + "reconstruction_s": 0.9778188000000227, + "eval_s": 1.0335336999996798, + "log_likelihood": 7112.5359410066085, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5988543000003119, + "n": 1310, + "solve_s": 0.5988543000003119, + "outer_iterations": 177, + "inner_iterations": 29, + "total_iterations": 206, + "n_passive": 1189, + "warm_start_errors": 84, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 0.6809902999993938, + "eval_s": 0.7303474000000278, + "log_likelihood": 7123.2058733607955, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29394549999960873, + "n": 1310, + "solve_s": 0.29394549999960873, + "outer_iterations": 69, + "inner_iterations": 11, + "total_iterations": 80, + "n_passive": 1200, + "warm_start_errors": 80, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05801526717557252, + "reconstruction_s": 0.721406399999978, + "eval_s": 0.7763215000013588, + "log_likelihood": 7175.679342224557, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3250086000007286, + "n": 1310, + "solve_s": 0.3250086000007286, + "outer_iterations": 68, + "inner_iterations": 14, + "total_iterations": 82, + "n_passive": 1197, + "warm_start_errors": 76, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 0.7106050000002142, + "eval_s": 0.7638234999994893, + "log_likelihood": 7224.232752132497, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3037913000007393, + "n": 1310, + "solve_s": 0.3037913000007393, + "outer_iterations": 60, + "inner_iterations": 11, + "total_iterations": 71, + "n_passive": 1192, + "warm_start_errors": 72, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 0.6711771000009321, + "eval_s": 0.724166799998784, + "log_likelihood": 7250.149755309723, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24293580000085058, + "n": 1310, + "solve_s": 0.24293580000085058, + "outer_iterations": 53, + "inner_iterations": 11, + "total_iterations": 64, + "n_passive": 1193, + "warm_start_errors": 69, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.051908396946564885, + "reconstruction_s": 0.6423376999991888, + "eval_s": 0.6979812000008678, + "log_likelihood": 7349.56988625817, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22654420000071696, + "n": 1310, + "solve_s": 0.22654420000071696, + "outer_iterations": 54, + "inner_iterations": 12, + "total_iterations": 66, + "n_passive": 1200, + "warm_start_errors": 68, + "index": 19 + } + ], + "on": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05877862595419847, + "reconstruction_s": 0.7739710000005289, + "eval_s": 0.8280987999987701, + "log_likelihood": 7206.2076831241275, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35296980000020994, + "n": 1310, + "solve_s": 0.35296980000020994, + "outer_iterations": 69, + "inner_iterations": 15, + "total_iterations": 84, + "n_passive": 1195, + "warm_start_errors": 77, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 0.39692029999969236, + "eval_s": 0.44677620000038587, + "log_likelihood": 7180.432957367257, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04616890000033891, + "n": 1310, + "solve_s": 0.04616890000033891, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1187, + "warm_start_errors": 8, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0015267175572519084, + "reconstruction_s": 0.39134529999864753, + "eval_s": 0.443003200000021, + "log_likelihood": 7143.9948890357955, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.029132400000889902, + "n": 1310, + "solve_s": 0.029132400000889902, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1187, + "warm_start_errors": 2, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.008396946564885497, + "reconstruction_s": 0.45465839999997115, + "eval_s": 0.5267707999992126, + "log_likelihood": 7182.878722012543, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05978509999840753, + "n": 1310, + "solve_s": 0.05978509999840753, + "outer_iterations": 9, + "inner_iterations": 3, + "total_iterations": 12, + "n_passive": 1190, + "warm_start_errors": 11, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.5071994999998424, + "eval_s": 0.5749274000008882, + "log_likelihood": 7186.1758717889115, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07923050000135845, + "n": 1310, + "solve_s": 0.07923050000135845, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 1189, + "warm_start_errors": 9, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.4667812999996386, + "eval_s": 0.5388029999994615, + "log_likelihood": 7145.024669580045, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07362110000030953, + "n": 1310, + "solve_s": 0.07362110000030953, + "outer_iterations": 4, + "inner_iterations": 3, + "total_iterations": 7, + "n_passive": 1188, + "warm_start_errors": 9, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0053435114503816794, + "reconstruction_s": 0.4774087999994663, + "eval_s": 0.54250520000096, + "log_likelihood": 7202.126666486729, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05328230000122858, + "n": 1310, + "solve_s": 0.05328230000122858, + "outer_iterations": 5, + "inner_iterations": 2, + "total_iterations": 7, + "n_passive": 1191, + "warm_start_errors": 7, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 0.4435164000005898, + "eval_s": 0.4930784999996831, + "log_likelihood": 7222.820219978841, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05963319999864325, + "n": 1310, + "solve_s": 0.05963319999864325, + "outer_iterations": 10, + "inner_iterations": 2, + "total_iterations": 12, + "n_passive": 1194, + "warm_start_errors": 13, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.4041503999997076, + "eval_s": 0.4571149000003061, + "log_likelihood": 7245.156725065451, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.043212999999013846, + "n": 1310, + "solve_s": 0.043212999999013846, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1186, + "warm_start_errors": 10, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0015267175572519084, + "reconstruction_s": 0.42375150000043504, + "eval_s": 0.5000037999998312, + "log_likelihood": 7163.55551490529, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.042228799999065814, + "n": 1310, + "solve_s": 0.042228799999065814, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1186, + "warm_start_errors": 2, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 0.4429546000010305, + "eval_s": 0.4949154999994789, + "log_likelihood": 7149.7410031695235, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04304790000060166, + "n": 1310, + "solve_s": 0.04304790000060166, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1186, + "warm_start_errors": 6, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 0.41365589999986696, + "eval_s": 0.46555499999885797, + "log_likelihood": 7135.510856830093, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049503800000820775, + "n": 1310, + "solve_s": 0.049503800000820775, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1187, + "warm_start_errors": 9, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 0.42201700000077835, + "eval_s": 0.47937309999906574, + "log_likelihood": 7124.427599165536, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.037420799999381416, + "n": 1310, + "solve_s": 0.037420799999381416, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1187, + "warm_start_errors": 4, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 0.5042181000007986, + "eval_s": 0.5713537000010547, + "log_likelihood": 7138.787138439637, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05520350000006147, + "n": 1310, + "solve_s": 0.05520350000006147, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1188, + "warm_start_errors": 3, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 0.4481747000008909, + "eval_s": 0.5059522000010475, + "log_likelihood": 7112.535941006625, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05862460000025749, + "n": 1310, + "solve_s": 0.05862460000025749, + "outer_iterations": 8, + "inner_iterations": 3, + "total_iterations": 11, + "n_passive": 1189, + "warm_start_errors": 13, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.011450381679389313, + "reconstruction_s": 0.522390500000256, + "eval_s": 0.5826086999986728, + "log_likelihood": 7123.205873360791, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10302740000042832, + "n": 1310, + "solve_s": 0.10302740000042832, + "outer_iterations": 16, + "inner_iterations": 2, + "total_iterations": 18, + "n_passive": 1200, + "warm_start_errors": 15, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 0.44690470000023197, + "eval_s": 0.5080217000013363, + "log_likelihood": 7175.679342224557, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.042369000000689994, + "n": 1310, + "solve_s": 0.042369000000689994, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 1196, + "warm_start_errors": 10, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 0.5182985999999801, + "eval_s": 0.5696698000010656, + "log_likelihood": 7224.2327521324605, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.14332550000108313, + "n": 1310, + "solve_s": 0.14332550000108313, + "outer_iterations": 25, + "inner_iterations": 4, + "total_iterations": 29, + "n_passive": 1192, + "warm_start_errors": 14, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 0.43841450000036275, + "eval_s": 0.4872765000000072, + "log_likelihood": 7250.14975530976, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03789309999956458, + "n": 1310, + "solve_s": 0.03789309999956458, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1192, + "warm_start_errors": 6, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.00916030534351145, + "reconstruction_s": 0.47102020000056655, + "eval_s": 0.5358584999994491, + "log_likelihood": 7349.569886258088, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08192389999931038, + "n": 1310, + "solve_s": 0.08192389999931038, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 1200, + "warm_start_errors": 12, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.0732824427480916, + "mean_warm_start_error_fraction": 0.07412213740458015, + "p90_warm_start_error_fraction": 0.09618320610687024, + "max_warm_start_error_fraction": 0.10610687022900764, + "median_outer_iterations": 70.0, + "mean_outer_iterations": 70.3, + "p90_outer_iterations": 92.30000000000001, + "max_outer_iterations": 106.0, + "median_inner_iterations": 17.0, + "mean_inner_iterations": 15.65, + "p90_inner_iterations": 20.200000000000003, + "max_inner_iterations": 28.0, + "median_total_iterations": 86.0, + "mean_total_iterations": 85.95, + "p90_total_iterations": 110.50000000000003, + "max_total_iterations": 134.0, + "median_n_passive": 1177.0, + "mean_n_passive": 1176.0, + "p90_n_passive": 1194.1, + "max_n_passive": 1196.0, + "median_warm_start_errors": 96.0, + "mean_warm_start_errors": 97.1, + "p90_warm_start_errors": 126.00000000000001, + "max_warm_start_errors": 139.0, + "median_solve_s": 0.3922603499995603, + "mean_solve_s": 0.38337681499970133, + "p90_solve_s": 0.4474004999996396, + "max_solve_s": 0.5113624999994499, + "median_reconstruction_s": 0.9306532499995228, + "mean_reconstruction_s": 0.9257930599998871, + "p90_reconstruction_s": 1.0876904899992952, + "max_reconstruction_s": 1.1924402999993617, + "median_eval_s": 0.989971300000434, + "mean_eval_s": 0.9909155099998316, + "p90_eval_s": 1.1549794599995948, + "max_eval_s": 1.2674207999989449 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.07328244274809159, + "mean_warm_start_error_fraction": 0.07561068702290076, + "p90_warm_start_error_fraction": 0.09809160305343512, + "max_warm_start_error_fraction": 0.10610687022900764, + "median_outer_iterations": 80.5, + "mean_outer_iterations": 77.0, + "p90_outer_iterations": 100.60000000000001, + "max_outer_iterations": 122.0, + "median_inner_iterations": 14.5, + "mean_inner_iterations": 15.45, + "p90_inner_iterations": 20.400000000000006, + "max_inner_iterations": 28.0, + "median_total_iterations": 94.0, + "mean_total_iterations": 92.45, + "p90_total_iterations": 120.50000000000003, + "max_total_iterations": 146.0, + "median_n_passive": 1177.0, + "mean_n_passive": 1175.7, + "p90_n_passive": 1194.1, + "max_n_passive": 1195.0, + "median_warm_start_errors": 96.0, + "mean_warm_start_errors": 99.05, + "p90_warm_start_errors": 128.5, + "max_warm_start_errors": 139.0, + "median_solve_s": 0.26892039999984263, + "mean_solve_s": 0.2747805399999379, + "p90_solve_s": 0.3892266299993935, + "max_solve_s": 0.49499420000029204, + "median_reconstruction_s": 0.7255309999991368, + "mean_reconstruction_s": 0.7270278550000512, + "p90_reconstruction_s": 0.8674844900002427, + "max_reconstruction_s": 1.0336895000000368, + "median_eval_s": 0.783289999999397, + "mean_eval_s": 0.787830545000088, + "p90_eval_s": 0.9305499699999928, + "max_eval_s": 1.0934159999997064 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 8.576436264303573e-15, + "max_abs_reconstruction_deviation": 1.2760601464378851e-09 + }, + "iteration_reduction_ratio": 0.9148936170212766, + "solve_speedup_ratio": 1.4586485443268336, + "eval_speedup_ratio": 1.2638630647668119, + "rows": { + "off": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.10610687022900764, + "reconstruction_s": 0.9280454999989161, + "eval_s": 0.9842375000007451, + "log_likelihood": 5471.7973737752245, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.45048299999871233, + "n": 1310, + "solve_s": 0.45048299999871233, + "outer_iterations": 106, + "inner_iterations": 28, + "total_iterations": 134, + "n_passive": 1140, + "warm_start_errors": 139, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.07862595419847328, + "reconstruction_s": 0.7660015999990719, + "eval_s": 0.8194540000004054, + "log_likelihood": 6654.532108619627, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3375080000005255, + "n": 1310, + "solve_s": 0.3375080000005255, + "outer_iterations": 71, + "inner_iterations": 18, + "total_iterations": 89, + "n_passive": 1167, + "warm_start_errors": 103, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0633587786259542, + "reconstruction_s": 0.7110603999990417, + "eval_s": 0.768450199999279, + "log_likelihood": 7226.971211085075, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.279094299999997, + "n": 1310, + "solve_s": 0.279094299999997, + "outer_iterations": 56, + "inner_iterations": 11, + "total_iterations": 67, + "n_passive": 1195, + "warm_start_errors": 83, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0900763358778626, + "reconstruction_s": 0.9341862000001129, + "eval_s": 0.9957051000001229, + "log_likelihood": 6607.394360361324, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41435939999973925, + "n": 1310, + "solve_s": 0.41435939999973925, + "outer_iterations": 75, + "inner_iterations": 12, + "total_iterations": 87, + "n_passive": 1160, + "warm_start_errors": 118, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.06564885496183206, + "reconstruction_s": 0.850544599999921, + "eval_s": 0.9199214000000211, + "log_likelihood": 7059.915357499807, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.347889299999224, + "n": 1310, + "solve_s": 0.347889299999224, + "outer_iterations": 66, + "inner_iterations": 11, + "total_iterations": 77, + "n_passive": 1183, + "warm_start_errors": 86, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 1.0701867999996466, + "eval_s": 1.145360200000141, + "log_likelihood": 7022.075373233347, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4381278000000748, + "n": 1310, + "solve_s": 0.4381278000000748, + "outer_iterations": 71, + "inner_iterations": 15, + "total_iterations": 86, + "n_passive": 1187, + "warm_start_errors": 87, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.07633587786259542, + "reconstruction_s": 1.0042253000010533, + "eval_s": 1.0681240000012622, + "log_likelihood": 6851.779927925216, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42368249999890395, + "n": 1310, + "solve_s": 0.42368249999890395, + "outer_iterations": 69, + "inner_iterations": 17, + "total_iterations": 86, + "n_passive": 1178, + "warm_start_errors": 100, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.07480916030534351, + "reconstruction_s": 0.7969069999999192, + "eval_s": 0.8606085999999777, + "log_likelihood": 6489.95669202, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3327458999992814, + "n": 1310, + "solve_s": 0.3327458999992814, + "outer_iterations": 65, + "inner_iterations": 19, + "total_iterations": 84, + "n_passive": 1174, + "warm_start_errors": 98, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.07480916030534351, + "reconstruction_s": 1.0098929000014323, + "eval_s": 1.094081199998982, + "log_likelihood": 6595.61511794514, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3961743999989267, + "n": 1310, + "solve_s": 0.3961743999989267, + "outer_iterations": 58, + "inner_iterations": 17, + "total_iterations": 75, + "n_passive": 1173, + "warm_start_errors": 98, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.10305343511450382, + "reconstruction_s": 1.1013010999995458, + "eval_s": 1.160488899999109, + "log_likelihood": 5179.313174140179, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4470579999997426, + "n": 1310, + "solve_s": 0.4470579999997426, + "outer_iterations": 91, + "inner_iterations": 18, + "total_iterations": 109, + "n_passive": 1154, + "warm_start_errors": 135, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0717557251908397, + "reconstruction_s": 1.0861781999992672, + "eval_s": 1.1543672999996488, + "log_likelihood": 6904.463092361186, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4292840999987675, + "n": 1310, + "solve_s": 0.4292840999987675, + "outer_iterations": 74, + "inner_iterations": 12, + "total_iterations": 86, + "n_passive": 1189, + "warm_start_errors": 94, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0816793893129771, + "reconstruction_s": 0.9332610000001296, + "eval_s": 0.9984859999985929, + "log_likelihood": 5670.821967236286, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3883463000001939, + "n": 1310, + "solve_s": 0.3883463000001939, + "outer_iterations": 61, + "inner_iterations": 17, + "total_iterations": 78, + "n_passive": 1153, + "warm_start_errors": 107, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.07022900763358779, + "reconstruction_s": 0.9374697000002925, + "eval_s": 0.9997781000001851, + "log_likelihood": 6645.924461980561, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37514220000048226, + "n": 1310, + "solve_s": 0.37514220000048226, + "outer_iterations": 64, + "inner_iterations": 22, + "total_iterations": 86, + "n_passive": 1158, + "warm_start_errors": 92, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.7743806000016775, + "eval_s": 0.8361005999995541, + "log_likelihood": 7151.479245469516, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2949178999988362, + "n": 1310, + "solve_s": 0.2949178999988362, + "outer_iterations": 39, + "inner_iterations": 11, + "total_iterations": 50, + "n_passive": 1176, + "warm_start_errors": 73, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.051908396946564885, + "reconstruction_s": 0.9001101000012568, + "eval_s": 0.960570799999914, + "log_likelihood": 7191.732883022228, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3808439999993425, + "n": 1310, + "solve_s": 0.3808439999993425, + "outer_iterations": 74, + "inner_iterations": 17, + "total_iterations": 91, + "n_passive": 1193, + "warm_start_errors": 68, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.055725190839694655, + "reconstruction_s": 0.878289099999165, + "eval_s": 0.9536510999987513, + "log_likelihood": 7160.20116161501, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4205039000007673, + "n": 1310, + "solve_s": 0.4205039000007673, + "outer_iterations": 80, + "inner_iterations": 19, + "total_iterations": 99, + "n_passive": 1196, + "warm_start_errors": 73, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.06870229007633588, + "reconstruction_s": 0.8947962000002008, + "eval_s": 0.956658700000844, + "log_likelihood": 7076.978400817326, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2964915000011388, + "n": 1310, + "solve_s": 0.2964915000011388, + "outer_iterations": 51, + "inner_iterations": 7, + "total_iterations": 58, + "n_passive": 1192, + "warm_start_errors": 90, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05343511450381679, + "reconstruction_s": 0.7500361999991583, + "eval_s": 0.8232007000005979, + "log_likelihood": 7435.3671852353855, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2718583999994735, + "n": 1310, + "solve_s": 0.2718583999994735, + "outer_iterations": 43, + "inner_iterations": 7, + "total_iterations": 50, + "n_passive": 1194, + "warm_start_errors": 70, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.07862595419847328, + "reconstruction_s": 1.1924402999993617, + "eval_s": 1.2674207999989449, + "log_likelihood": 6574.8371202406925, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5113624999994499, + "n": 1310, + "solve_s": 0.5113624999994499, + "outer_iterations": 88, + "inner_iterations": 15, + "total_iterations": 103, + "n_passive": 1188, + "warm_start_errors": 103, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.09541984732824428, + "reconstruction_s": 0.9965483999985736, + "eval_s": 1.051644999999553, + "log_likelihood": 6171.152613875354, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4316629000004468, + "n": 1310, + "solve_s": 0.4316629000004468, + "outer_iterations": 104, + "inner_iterations": 20, + "total_iterations": 124, + "n_passive": 1170, + "warm_start_errors": 125, + "index": 19 + } + ], + "on": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.10610687022900764, + "reconstruction_s": 1.0336895000000368, + "eval_s": 1.0934159999997064, + "log_likelihood": 5471.7973737752245, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.49499420000029204, + "n": 1310, + "solve_s": 0.49499420000029204, + "outer_iterations": 106, + "inner_iterations": 28, + "total_iterations": 134, + "n_passive": 1140, + "warm_start_errors": 139, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 0.8617240000003221, + "eval_s": 0.9238500999999815, + "log_likelihood": 6654.5321086196445, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39517049999994924, + "n": 1310, + "solve_s": 0.39517049999994924, + "outer_iterations": 100, + "inner_iterations": 19, + "total_iterations": 119, + "n_passive": 1167, + "warm_start_errors": 115, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.08244274809160305, + "reconstruction_s": 0.7383224999994127, + "eval_s": 0.7944993999990402, + "log_likelihood": 7226.971211085047, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.316097400000217, + "n": 1310, + "solve_s": 0.316097400000217, + "outer_iterations": 91, + "inner_iterations": 20, + "total_iterations": 111, + "n_passive": 1195, + "warm_start_errors": 108, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.08778625954198473, + "reconstruction_s": 0.8376602999996976, + "eval_s": 0.9004343999986304, + "log_likelihood": 6607.394360361362, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38856619999933173, + "n": 1310, + "solve_s": 0.38856619999933173, + "outer_iterations": 97, + "inner_iterations": 19, + "total_iterations": 116, + "n_passive": 1160, + "warm_start_errors": 115, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06946564885496183, + "reconstruction_s": 0.7509394999997312, + "eval_s": 0.8201950999991823, + "log_likelihood": 7059.915357499809, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2943816000006336, + "n": 1310, + "solve_s": 0.2943816000006336, + "outer_iterations": 80, + "inner_iterations": 14, + "total_iterations": 94, + "n_passive": 1183, + "warm_start_errors": 91, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06259541984732825, + "reconstruction_s": 0.9193288999995275, + "eval_s": 0.9908488000000943, + "log_likelihood": 7022.075373233368, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3028870999987703, + "n": 1310, + "solve_s": 0.3028870999987703, + "outer_iterations": 81, + "inner_iterations": 13, + "total_iterations": 94, + "n_passive": 1187, + "warm_start_errors": 82, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.07557251908396946, + "reconstruction_s": 0.8297017000004416, + "eval_s": 0.8936468000010791, + "log_likelihood": 6851.7799279252395, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22800099999949452, + "n": 1310, + "solve_s": 0.22800099999949452, + "outer_iterations": 55, + "inner_iterations": 16, + "total_iterations": 71, + "n_passive": 1178, + "warm_start_errors": 99, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.10152671755725191, + "reconstruction_s": 0.8243256000005204, + "eval_s": 0.8858139000003575, + "log_likelihood": 6489.956692020017, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34455479999996896, + "n": 1310, + "solve_s": 0.34455479999996896, + "outer_iterations": 97, + "inner_iterations": 18, + "total_iterations": 115, + "n_passive": 1173, + "warm_start_errors": 133, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.08091603053435115, + "reconstruction_s": 0.6498172000010527, + "eval_s": 0.7066890999994939, + "log_likelihood": 6595.615117945145, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24345919999905163, + "n": 1310, + "solve_s": 0.24345919999905163, + "outer_iterations": 82, + "inner_iterations": 17, + "total_iterations": 99, + "n_passive": 1173, + "warm_start_errors": 106, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.083206106870229, + "reconstruction_s": 0.6101783000012801, + "eval_s": 0.6645657000008214, + "log_likelihood": 5179.313174140187, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1845432999998593, + "n": 1310, + "solve_s": 0.1845432999998593, + "outer_iterations": 59, + "inner_iterations": 12, + "total_iterations": 71, + "n_passive": 1154, + "warm_start_errors": 109, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.07938931297709924, + "reconstruction_s": 0.7483554000009462, + "eval_s": 0.8250086000007286, + "log_likelihood": 6904.463092361187, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31627470000057656, + "n": 1310, + "solve_s": 0.31627470000057656, + "outer_iterations": 87, + "inner_iterations": 16, + "total_iterations": 103, + "n_passive": 1188, + "warm_start_errors": 104, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.07099236641221374, + "reconstruction_s": 0.6168073999997432, + "eval_s": 0.675082100000509, + "log_likelihood": 5670.821967236286, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.17969540000012785, + "n": 1310, + "solve_s": 0.17969540000012785, + "outer_iterations": 57, + "inner_iterations": 12, + "total_iterations": 69, + "n_passive": 1153, + "warm_start_errors": 93, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.09770992366412214, + "reconstruction_s": 0.7748444999997446, + "eval_s": 0.833579500000269, + "log_likelihood": 6645.9244619805595, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35118619999957446, + "n": 1310, + "solve_s": 0.35118619999957446, + "outer_iterations": 122, + "inner_iterations": 24, + "total_iterations": 146, + "n_passive": 1157, + "warm_start_errors": 128, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 0.5851015000007465, + "eval_s": 0.6410281000007672, + "log_likelihood": 7151.479245469577, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.194641899999624, + "n": 1310, + "solve_s": 0.194641899999624, + "outer_iterations": 56, + "inner_iterations": 8, + "total_iterations": 64, + "n_passive": 1176, + "warm_start_errors": 75, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06793893129770992, + "reconstruction_s": 0.6388877999997931, + "eval_s": 0.6956086000009236, + "log_likelihood": 7191.732883022249, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22034979999989446, + "n": 1310, + "solve_s": 0.22034979999989446, + "outer_iterations": 71, + "inner_iterations": 11, + "total_iterations": 82, + "n_passive": 1193, + "warm_start_errors": 89, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 0.5379042999993544, + "eval_s": 0.5934401000013168, + "log_likelihood": 7160.201161614967, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1657854999994015, + "n": 1310, + "solve_s": 0.1657854999994015, + "outer_iterations": 46, + "inner_iterations": 12, + "total_iterations": 58, + "n_passive": 1195, + "warm_start_errors": 66, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06412213740458016, + "reconstruction_s": 0.6373148000002402, + "eval_s": 0.6939734000006865, + "log_likelihood": 7076.978400817332, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22371650000059162, + "n": 1310, + "solve_s": 0.22371650000059162, + "outer_iterations": 70, + "inner_iterations": 15, + "total_iterations": 85, + "n_passive": 1191, + "warm_start_errors": 84, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 0.5361632000003738, + "eval_s": 0.5903491999997641, + "log_likelihood": 7435.367185235396, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.15365940000083356, + "n": 1310, + "solve_s": 0.15365940000083356, + "outer_iterations": 44, + "inner_iterations": 13, + "total_iterations": 57, + "n_passive": 1194, + "warm_start_errors": 67, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06946564885496183, + "reconstruction_s": 0.712739499998861, + "eval_s": 0.7720805999997538, + "log_likelihood": 6574.837120240636, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29987009999967995, + "n": 1310, + "solve_s": 0.29987009999967995, + "outer_iterations": 88, + "inner_iterations": 13, + "total_iterations": 101, + "n_passive": 1187, + "warm_start_errors": 91, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 0.6967511999991984, + "eval_s": 0.762501399998655, + "log_likelihood": 6171.1526138753525, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1977760000008857, + "n": 1310, + "solve_s": 0.1977760000008857, + "outer_iterations": 51, + "inner_iterations": 9, + "total_iterations": 60, + "n_passive": 1170, + "warm_start_errors": 87, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_v2026.8.17.1.png new file mode 100644 index 000000000..4b32c1fdb Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_euclid_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_mesh_600_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_mesh_600_v2026.8.17.1.json new file mode 100644 index 000000000..022a614cf --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_mesh_600_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "hst", + "model": "mesh_600", + "configuration": { + "model": "mesh_600", + "dataset": "hst", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.05, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 15361, + "over_sampled_pixels": 17980, + "source_pixels": 600, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_600", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 5.901554100000794, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.031818181818181815, + "mean_warm_start_error_fraction": 0.03272727272727273, + "p90_warm_start_error_fraction": 0.03636363636363636, + "max_warm_start_error_fraction": 0.03787878787878788, + "median_outer_iterations": 15.0, + "mean_outer_iterations": 15.55, + "p90_outer_iterations": 17.1, + "max_outer_iterations": 20.0, + "median_inner_iterations": 9.0, + "mean_inner_iterations": 8.35, + "p90_inner_iterations": 9.0, + "max_inner_iterations": 10.0, + "median_total_iterations": 24.0, + "mean_total_iterations": 23.9, + "p90_total_iterations": 26.1, + "max_total_iterations": 29.0, + "median_n_passive": 626.5, + "mean_n_passive": 625.9, + "p90_n_passive": 629.1, + "max_n_passive": 630.0, + "median_warm_start_errors": 21.0, + "mean_warm_start_errors": 21.6, + "p90_warm_start_errors": 24.0, + "max_warm_start_errors": 25.0, + "median_solve_s": 0.05885119999993549, + "mean_solve_s": 0.06370723000009093, + "p90_solve_s": 0.07876615999975912, + "max_solve_s": 0.10647980000067037, + "median_reconstruction_s": 1.8247200000000703, + "mean_reconstruction_s": 1.883189850000008, + "p90_reconstruction_s": 2.1586241399995743, + "max_reconstruction_s": 2.3933619999997973, + "median_eval_s": 1.8414591499999915, + "mean_eval_s": 1.9011740149999241, + "p90_eval_s": 2.174590090000038, + "max_eval_s": 2.40998910000053 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.0053030303030303025, + "mean_warm_start_error_fraction": 0.008181818181818182, + "p90_warm_start_error_fraction": 0.01681818181818182, + "max_warm_start_error_fraction": 0.030303030303030304, + "median_outer_iterations": 2.0, + "mean_outer_iterations": 3.1, + "p90_outer_iterations": 9.100000000000001, + "max_outer_iterations": 14.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 1.4, + "p90_inner_iterations": 2.200000000000003, + "max_inner_iterations": 7.0, + "median_total_iterations": 3.0, + "mean_total_iterations": 4.5, + "p90_total_iterations": 10.100000000000001, + "max_total_iterations": 21.0, + "median_n_passive": 626.5, + "mean_n_passive": 625.65, + "p90_n_passive": 629.0, + "max_n_passive": 630.0, + "median_warm_start_errors": 3.5, + "mean_warm_start_errors": 5.4, + "p90_warm_start_errors": 11.100000000000001, + "max_warm_start_errors": 20.0, + "median_solve_s": 0.012259250000170141, + "mean_solve_s": 0.015033554999945409, + "p90_solve_s": 0.020394459999897675, + "max_solve_s": 0.04814139999962208, + "median_reconstruction_s": 1.8910689499998625, + "mean_reconstruction_s": 1.943360649999977, + "p90_reconstruction_s": 2.158086189999449, + "max_reconstruction_s": 2.8915098000006765, + "median_eval_s": 1.9097309499998119, + "mean_eval_s": 1.9632644950000668, + "p90_eval_s": 2.176243689999956, + "max_eval_s": 2.914845800000876 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.0158493086010438e-15, + "max_abs_reconstruction_deviation": 1.586546005682976e-09 + }, + "iteration_reduction_ratio": 8.0, + "solve_speedup_ratio": 4.8005546831265145, + "eval_speedup_ratio": 0.9642505662905934, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.030303030303030304, + "reconstruction_s": 1.6545872999995481, + "eval_s": 1.6724142000002757, + "log_likelihood": 29187.796477370277, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05593399999997928, + "n": 660, + "solve_s": 0.05593399999997928, + "outer_iterations": 14, + "inner_iterations": 7, + "total_iterations": 21, + "n_passive": 628, + "warm_start_errors": 20, + "index": 0 + }, + { + "warm_start_error_fraction": 0.030303030303030304, + "reconstruction_s": 1.8246318000001338, + "eval_s": 1.8419365999998263, + "log_likelihood": 29016.413029348027, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07019679999939399, + "n": 660, + "solve_s": 0.07019679999939399, + "outer_iterations": 15, + "inner_iterations": 10, + "total_iterations": 25, + "n_passive": 627, + "warm_start_errors": 20, + "index": 1 + }, + { + "warm_start_error_fraction": 0.030303030303030304, + "reconstruction_s": 1.5956459999997605, + "eval_s": 1.612196999999469, + "log_likelihood": 28839.256792704808, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05097170000044571, + "n": 660, + "solve_s": 0.05097170000044571, + "outer_iterations": 15, + "inner_iterations": 9, + "total_iterations": 24, + "n_passive": 628, + "warm_start_errors": 20, + "index": 2 + }, + { + "warm_start_error_fraction": 0.031818181818181815, + "reconstruction_s": 1.7864417000000685, + "eval_s": 1.8037667999997211, + "log_likelihood": 29046.10608618408, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05381840000063676, + "n": 660, + "solve_s": 0.05381840000063676, + "outer_iterations": 16, + "inner_iterations": 9, + "total_iterations": 25, + "n_passive": 628, + "warm_start_errors": 21, + "index": 3 + }, + { + "warm_start_error_fraction": 0.031818181818181815, + "reconstruction_s": 2.1552132999995592, + "eval_s": 2.1712502000000313, + "log_likelihood": 29055.111150890763, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06871649999993679, + "n": 660, + "solve_s": 0.06871649999993679, + "outer_iterations": 16, + "inner_iterations": 9, + "total_iterations": 25, + "n_passive": 629, + "warm_start_errors": 21, + "index": 4 + }, + { + "warm_start_error_fraction": 0.03333333333333333, + "reconstruction_s": 2.0852279000000635, + "eval_s": 2.099683099999311, + "log_likelihood": 28848.2936445414, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07788559999971767, + "n": 660, + "solve_s": 0.07788559999971767, + "outer_iterations": 13, + "inner_iterations": 7, + "total_iterations": 20, + "n_passive": 625, + "warm_start_errors": 22, + "index": 5 + }, + { + "warm_start_error_fraction": 0.02878787878787879, + "reconstruction_s": 1.9465776000006372, + "eval_s": 1.9632752999996228, + "log_likelihood": 29162.50458337732, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05970249999973021, + "n": 660, + "solve_s": 0.05970249999973021, + "outer_iterations": 15, + "inner_iterations": 8, + "total_iterations": 23, + "n_passive": 629, + "warm_start_errors": 19, + "index": 6 + }, + { + "warm_start_error_fraction": 0.03333333333333333, + "reconstruction_s": 1.9727957000004608, + "eval_s": 1.9947462000000087, + "log_likelihood": 29251.734622168755, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07300070000019332, + "n": 660, + "solve_s": 0.07300070000019332, + "outer_iterations": 15, + "inner_iterations": 8, + "total_iterations": 23, + "n_passive": 623, + "warm_start_errors": 22, + "index": 7 + }, + { + "warm_start_error_fraction": 0.031818181818181815, + "reconstruction_s": 2.0962509000000864, + "eval_s": 2.113644199999726, + "log_likelihood": 29353.366162905244, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0605199999999968, + "n": 660, + "solve_s": 0.0605199999999968, + "outer_iterations": 18, + "inner_iterations": 9, + "total_iterations": 27, + "n_passive": 626, + "warm_start_errors": 21, + "index": 8 + }, + { + "warm_start_error_fraction": 0.03484848484848485, + "reconstruction_s": 2.1893216999997094, + "eval_s": 2.204649100000097, + "log_likelihood": 28913.98371855758, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05799990000014077, + "n": 660, + "solve_s": 0.05799990000014077, + "outer_iterations": 16, + "inner_iterations": 9, + "total_iterations": 25, + "n_passive": 624, + "warm_start_errors": 23, + "index": 9 + }, + { + "warm_start_error_fraction": 0.031818181818181815, + "reconstruction_s": 1.7079084999995757, + "eval_s": 1.7286404999995284, + "log_likelihood": 28828.888810354918, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.052484499999991385, + "n": 660, + "solve_s": 0.052484499999991385, + "outer_iterations": 17, + "inner_iterations": 9, + "total_iterations": 26, + "n_passive": 627, + "warm_start_errors": 21, + "index": 10 + }, + { + "warm_start_error_fraction": 0.03636363636363636, + "reconstruction_s": 1.8248082000000068, + "eval_s": 1.8409817000001567, + "log_likelihood": 28735.729420256117, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05141559999992751, + "n": 660, + "solve_s": 0.05141559999992751, + "outer_iterations": 17, + "inner_iterations": 9, + "total_iterations": 26, + "n_passive": 622, + "warm_start_errors": 24, + "index": 11 + }, + { + "warm_start_error_fraction": 0.03636363636363636, + "reconstruction_s": 1.7543218999999226, + "eval_s": 1.7726744000001418, + "log_likelihood": 28682.713576995862, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.051668099999915285, + "n": 660, + "solve_s": 0.051668099999915285, + "outer_iterations": 15, + "inner_iterations": 8, + "total_iterations": 23, + "n_passive": 624, + "warm_start_errors": 24, + "index": 12 + }, + { + "warm_start_error_fraction": 0.03636363636363636, + "reconstruction_s": 1.7218374000003678, + "eval_s": 1.747208199999477, + "log_likelihood": 28752.80136832575, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06603629999972327, + "n": 660, + "solve_s": 0.06603629999972327, + "outer_iterations": 14, + "inner_iterations": 7, + "total_iterations": 21, + "n_passive": 623, + "warm_start_errors": 24, + "index": 13 + }, + { + "warm_start_error_fraction": 0.03636363636363636, + "reconstruction_s": 1.8018752000007225, + "eval_s": 1.8193653000007544, + "log_likelihood": 28649.751700686247, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07068309999976918, + "n": 660, + "solve_s": 0.07068309999976918, + "outer_iterations": 14, + "inner_iterations": 9, + "total_iterations": 23, + "n_passive": 622, + "warm_start_errors": 24, + "index": 14 + }, + { + "warm_start_error_fraction": 0.03636363636363636, + "reconstruction_s": 1.9689331999998103, + "eval_s": 1.998399099999915, + "log_likelihood": 28709.32579975877, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10647980000067037, + "n": 660, + "solve_s": 0.10647980000067037, + "outer_iterations": 16, + "inner_iterations": 9, + "total_iterations": 25, + "n_passive": 623, + "warm_start_errors": 24, + "index": 15 + }, + { + "warm_start_error_fraction": 0.03787878787878788, + "reconstruction_s": 2.3933619999997973, + "eval_s": 2.40998910000053, + "log_likelihood": 28897.133359014708, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08669120000013208, + "n": 660, + "solve_s": 0.08669120000013208, + "outer_iterations": 16, + "inner_iterations": 8, + "total_iterations": 24, + "n_passive": 622, + "warm_start_errors": 25, + "index": 16 + }, + { + "warm_start_error_fraction": 0.02878787878787879, + "reconstruction_s": 1.6578521000001274, + "eval_s": 1.6740442000000257, + "log_likelihood": 29257.862473028435, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.052129900000181806, + "n": 660, + "solve_s": 0.052129900000181806, + "outer_iterations": 15, + "inner_iterations": 8, + "total_iterations": 23, + "n_passive": 630, + "warm_start_errors": 19, + "index": 17 + }, + { + "warm_start_error_fraction": 0.031818181818181815, + "reconstruction_s": 1.8982760999997481, + "eval_s": 1.9127377999993769, + "log_likelihood": 29420.307318413943, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.056743800000731426, + "n": 660, + "solve_s": 0.056743800000731426, + "outer_iterations": 20, + "inner_iterations": 9, + "total_iterations": 29, + "n_passive": 630, + "warm_start_errors": 21, + "index": 18 + }, + { + "warm_start_error_fraction": 0.025757575757575757, + "reconstruction_s": 1.6279285000000527, + "eval_s": 1.6418773000004876, + "log_likelihood": 29937.157497533954, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05106620000060502, + "n": 660, + "solve_s": 0.05106620000060502, + "outer_iterations": 14, + "inner_iterations": 6, + "total_iterations": 20, + "n_passive": 628, + "warm_start_errors": 17, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.030303030303030304, + "reconstruction_s": 1.741932299999462, + "eval_s": 1.758082000000286, + "log_likelihood": 29187.796477370277, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04814139999962208, + "n": 660, + "solve_s": 0.04814139999962208, + "outer_iterations": 14, + "inner_iterations": 7, + "total_iterations": 21, + "n_passive": 628, + "warm_start_errors": 20, + "index": 0 + }, + { + "warm_start_error_fraction": 0.007575757575757576, + "reconstruction_s": 1.6793829999996888, + "eval_s": 1.6965172000000166, + "log_likelihood": 29016.413029348038, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.012379300000247895, + "n": 660, + "solve_s": 0.012379300000247895, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 627, + "warm_start_errors": 5, + "index": 1 + }, + { + "warm_start_error_fraction": 0.004545454545454545, + "reconstruction_s": 1.628159500000038, + "eval_s": 1.6440279999997074, + "log_likelihood": 28839.256792704808, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.009035000000039872, + "n": 660, + "solve_s": 0.009035000000039872, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 628, + "warm_start_errors": 3, + "index": 2 + }, + { + "warm_start_error_fraction": 0.006060606060606061, + "reconstruction_s": 1.6567586999999548, + "eval_s": 1.6725415000000794, + "log_likelihood": 29046.106086184078, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.00985569999920699, + "n": 660, + "solve_s": 0.00985569999920699, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 628, + "warm_start_errors": 4, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0, + "reconstruction_s": 1.7967382999995607, + "eval_s": 1.8143731999998636, + "log_likelihood": 29055.111150890763, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.008006800000657677, + "n": 660, + "solve_s": 0.008006800000657677, + "outer_iterations": 0, + "inner_iterations": 0, + "total_iterations": 0, + "n_passive": 628, + "warm_start_errors": 0, + "index": 4 + }, + { + "warm_start_error_fraction": 0.015151515151515152, + "reconstruction_s": 1.9907055999992735, + "eval_s": 2.009308000000601, + "log_likelihood": 28848.293644541405, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.016933299999436713, + "n": 660, + "solve_s": 0.016933299999436713, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 624, + "warm_start_errors": 10, + "index": 5 + }, + { + "warm_start_error_fraction": 0.016666666666666666, + "reconstruction_s": 1.895968300000277, + "eval_s": 1.915357199999562, + "log_likelihood": 29162.504583377315, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.01944809999986319, + "n": 660, + "solve_s": 0.01944809999986319, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 629, + "warm_start_errors": 11, + "index": 6 + }, + { + "warm_start_error_fraction": 0.015151515151515152, + "reconstruction_s": 1.6561450000008335, + "eval_s": 1.6709754000003159, + "log_likelihood": 29251.734622168755, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.012139200000092387, + "n": 660, + "solve_s": 0.012139200000092387, + "outer_iterations": 2, + "inner_iterations": 4, + "total_iterations": 6, + "n_passive": 623, + "warm_start_errors": 10, + "index": 7 + }, + { + "warm_start_error_fraction": 0.004545454545454545, + "reconstruction_s": 1.7977061000001413, + "eval_s": 1.8134035000002768, + "log_likelihood": 29353.36616290525, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.011982500000158325, + "n": 660, + "solve_s": 0.011982500000158325, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 626, + "warm_start_errors": 3, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0030303030303030303, + "reconstruction_s": 1.9720305999999255, + "eval_s": 1.9896765000003143, + "log_likelihood": 28913.983718557578, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.008563299999877927, + "n": 660, + "solve_s": 0.008563299999877927, + "outer_iterations": 0, + "inner_iterations": 1, + "total_iterations": 1, + "n_passive": 624, + "warm_start_errors": 2, + "index": 9 + }, + { + "warm_start_error_fraction": 0.004545454545454545, + "reconstruction_s": 1.886169599999448, + "eval_s": 1.9041047000000617, + "log_likelihood": 28828.888810354925, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.015753800000311458, + "n": 660, + "solve_s": 0.015753800000311458, + "outer_iterations": 3, + "inner_iterations": 0, + "total_iterations": 3, + "n_passive": 627, + "warm_start_errors": 3, + "index": 10 + }, + { + "warm_start_error_fraction": 0.007575757575757576, + "reconstruction_s": 2.092911300000196, + "eval_s": 2.114150399999744, + "log_likelihood": 28735.729420256088, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.009863100000075065, + "n": 660, + "solve_s": 0.009863100000075065, + "outer_iterations": 0, + "inner_iterations": 1, + "total_iterations": 1, + "n_passive": 622, + "warm_start_errors": 5, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0015151515151515152, + "reconstruction_s": 2.136373599999388, + "eval_s": 2.153252699999939, + "log_likelihood": 28682.713576995862, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.008325499999955355, + "n": 660, + "solve_s": 0.008325499999955355, + "outer_iterations": 1, + "inner_iterations": 0, + "total_iterations": 1, + "n_passive": 623, + "warm_start_errors": 1, + "index": 12 + }, + { + "warm_start_error_fraction": 0.007575757575757576, + "reconstruction_s": 2.0184784000002765, + "eval_s": 2.0396852999992916, + "log_likelihood": 28752.801368325763, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.011659599999802595, + "n": 660, + "solve_s": 0.011659599999802595, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 622, + "warm_start_errors": 5, + "index": 13 + }, + { + "warm_start_error_fraction": 0.0030303030303030303, + "reconstruction_s": 2.353499499999998, + "eval_s": 2.383162600000105, + "log_likelihood": 28649.751700686218, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.01869529999930819, + "n": 660, + "solve_s": 0.01869529999930819, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 622, + "warm_start_errors": 2, + "index": 14 + }, + { + "warm_start_error_fraction": 0.004545454545454545, + "reconstruction_s": 1.8436474999998609, + "eval_s": 1.865426600000319, + "log_likelihood": 28709.32579975878, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.014101699999628181, + "n": 660, + "solve_s": 0.014101699999628181, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 623, + "warm_start_errors": 3, + "index": 15 + }, + { + "warm_start_error_fraction": 0.007575757575757576, + "reconstruction_s": 1.8019890000005034, + "eval_s": 1.819347900000139, + "log_likelihood": 28897.133359014704, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.010914000000411761, + "n": 660, + "solve_s": 0.010914000000411761, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 622, + "warm_start_errors": 5, + "index": 16 + }, + { + "warm_start_error_fraction": 0.01818181818181818, + "reconstruction_s": 1.9086262000000715, + "eval_s": 1.9390874000000622, + "log_likelihood": 29257.86247302843, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.02891170000020793, + "n": 660, + "solve_s": 0.02891170000020793, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 630, + "warm_start_errors": 12, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0015151515151515152, + "reconstruction_s": 2.8915098000006765, + "eval_s": 2.914845800000876, + "log_likelihood": 29420.307318413947, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.013430899999548274, + "n": 660, + "solve_s": 0.013430899999548274, + "outer_iterations": 0, + "inner_iterations": 1, + "total_iterations": 1, + "n_passive": 629, + "warm_start_errors": 1, + "index": 18 + }, + { + "warm_start_error_fraction": 0.004545454545454545, + "reconstruction_s": 2.1184806999999637, + "eval_s": 2.1479639999997744, + "log_likelihood": 29937.15749753396, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.012530900000456313, + "n": 660, + "solve_s": 0.012530900000456313, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 628, + "warm_start_errors": 3, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.051515151515151514, + "mean_warm_start_error_fraction": 0.051515151515151514, + "p90_warm_start_error_fraction": 0.06227272727272727, + "max_warm_start_error_fraction": 0.06515151515151515, + "median_outer_iterations": 20.0, + "mean_outer_iterations": 19.65, + "p90_outer_iterations": 28.1, + "max_outer_iterations": 30.0, + "median_inner_iterations": 11.0, + "mean_inner_iterations": 10.95, + "p90_inner_iterations": 14.0, + "max_inner_iterations": 22.0, + "median_total_iterations": 30.5, + "mean_total_iterations": 30.6, + "p90_total_iterations": 42.0, + "max_total_iterations": 51.0, + "median_n_passive": 613.0, + "mean_n_passive": 613.1, + "p90_n_passive": 621.1, + "max_n_passive": 623.0, + "median_warm_start_errors": 34.0, + "mean_warm_start_errors": 34.0, + "p90_warm_start_errors": 41.1, + "max_warm_start_errors": 43.0, + "median_solve_s": 0.06889384999976755, + "mean_solve_s": 0.08115661000001637, + "p90_solve_s": 0.10093506000011988, + "max_solve_s": 0.17890970000007655, + "median_reconstruction_s": 1.861924949999775, + "mean_reconstruction_s": 1.9197677099999964, + "p90_reconstruction_s": 2.429773230000046, + "max_reconstruction_s": 2.491513599999962, + "median_eval_s": 1.8806060500000967, + "mean_eval_s": 1.939316719999897, + "p90_eval_s": 2.4498171600002934, + "max_eval_s": 2.5175430999997843 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.04772727272727273, + "mean_warm_start_error_fraction": 0.04840909090909091, + "p90_warm_start_error_fraction": 0.06696969696969697, + "max_warm_start_error_fraction": 0.07121212121212121, + "median_outer_iterations": 27.5, + "mean_outer_iterations": 26.85, + "p90_outer_iterations": 37.70000000000001, + "max_outer_iterations": 44.0, + "median_inner_iterations": 7.0, + "mean_inner_iterations": 6.7, + "p90_inner_iterations": 10.0, + "max_inner_iterations": 11.0, + "median_total_iterations": 36.0, + "mean_total_iterations": 33.55, + "p90_total_iterations": 47.60000000000001, + "max_total_iterations": 54.0, + "median_n_passive": 613.0, + "mean_n_passive": 612.75, + "p90_n_passive": 621.1, + "max_n_passive": 623.0, + "median_warm_start_errors": 31.5, + "mean_warm_start_errors": 31.95, + "p90_warm_start_errors": 44.2, + "max_warm_start_errors": 47.0, + "median_solve_s": 0.040678899999875284, + "mean_solve_s": 0.04159805499984941, + "p90_solve_s": 0.061568650000390336, + "max_solve_s": 0.06396859999949811, + "median_reconstruction_s": 1.7758342500001163, + "mean_reconstruction_s": 1.7686940599999617, + "p90_reconstruction_s": 1.9616980599996168, + "max_reconstruction_s": 2.0328591999996206, + "median_eval_s": 1.7919427500000893, + "mean_eval_s": 1.7863379650001208, + "p90_eval_s": 1.9804949199997282, + "max_eval_s": 2.058980099999644 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 6.444340100706452e-15, + "max_abs_reconstruction_deviation": 1.0576335363055023e-09 + }, + "iteration_reduction_ratio": 0.8472222222222222, + "solve_speedup_ratio": 1.6936015968961493, + "eval_speedup_ratio": 1.0494788686747962, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.06515151515151515, + "reconstruction_s": 1.9921449000003122, + "eval_s": 2.0083403999997245, + "log_likelihood": 19526.535596202903, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0649639000002935, + "n": 660, + "solve_s": 0.0649639000002935, + "outer_iterations": 25, + "inner_iterations": 9, + "total_iterations": 34, + "n_passive": 615, + "warm_start_errors": 43, + "index": 0 + }, + { + "warm_start_error_fraction": 0.06363636363636363, + "reconstruction_s": 1.78406780000023, + "eval_s": 1.799011399999472, + "log_likelihood": 25185.4735301975, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06569309999940742, + "n": 660, + "solve_s": 0.06569309999940742, + "outer_iterations": 27, + "inner_iterations": 14, + "total_iterations": 41, + "n_passive": 616, + "warm_start_errors": 42, + "index": 1 + }, + { + "warm_start_error_fraction": 0.03636363636363636, + "reconstruction_s": 1.6245291000004727, + "eval_s": 1.6459555000001274, + "log_likelihood": 28654.382000591315, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05757160000030126, + "n": 660, + "solve_s": 0.05757160000030126, + "outer_iterations": 8, + "inner_iterations": 7, + "total_iterations": 15, + "n_passive": 621, + "warm_start_errors": 24, + "index": 2 + }, + { + "warm_start_error_fraction": 0.051515151515151514, + "reconstruction_s": 1.855851299999813, + "eval_s": 1.8790575000002718, + "log_likelihood": 25668.81051195281, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06570960000044579, + "n": 660, + "solve_s": 0.06570960000044579, + "outer_iterations": 20, + "inner_iterations": 11, + "total_iterations": 31, + "n_passive": 618, + "warm_start_errors": 34, + "index": 3 + }, + { + "warm_start_error_fraction": 0.045454545454545456, + "reconstruction_s": 1.867998599999737, + "eval_s": 1.8821545999999216, + "log_likelihood": 27947.140164089666, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06546700000035344, + "n": 660, + "solve_s": 0.06546700000035344, + "outer_iterations": 15, + "inner_iterations": 8, + "total_iterations": 23, + "n_passive": 622, + "warm_start_errors": 30, + "index": 4 + }, + { + "warm_start_error_fraction": 0.05303030303030303, + "reconstruction_s": 1.604360799999995, + "eval_s": 1.619336900000235, + "log_likelihood": 27806.960101290595, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07804779999969469, + "n": 660, + "solve_s": 0.07804779999969469, + "outer_iterations": 22, + "inner_iterations": 10, + "total_iterations": 32, + "n_passive": 613, + "warm_start_errors": 35, + "index": 5 + }, + { + "warm_start_error_fraction": 0.05909090909090909, + "reconstruction_s": 1.7005328000004738, + "eval_s": 1.7164220999993631, + "log_likelihood": 26457.65754502258, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09853560000010475, + "n": 660, + "solve_s": 0.09853560000010475, + "outer_iterations": 20, + "inner_iterations": 13, + "total_iterations": 33, + "n_passive": 605, + "warm_start_errors": 39, + "index": 6 + }, + { + "warm_start_error_fraction": 0.051515151515151514, + "reconstruction_s": 1.8735877999997683, + "eval_s": 1.8924349999997503, + "log_likelihood": 24507.466310976957, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07734440000058385, + "n": 660, + "solve_s": 0.07734440000058385, + "outer_iterations": 19, + "inner_iterations": 11, + "total_iterations": 30, + "n_passive": 603, + "warm_start_errors": 34, + "index": 7 + }, + { + "warm_start_error_fraction": 0.04393939393939394, + "reconstruction_s": 1.734262499999204, + "eval_s": 1.7578663000003871, + "log_likelihood": 24911.875301704178, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08902579999994487, + "n": 660, + "solve_s": 0.08902579999994487, + "outer_iterations": 16, + "inner_iterations": 11, + "total_iterations": 27, + "n_passive": 611, + "warm_start_errors": 29, + "index": 8 + }, + { + "warm_start_error_fraction": 0.05454545454545454, + "reconstruction_s": 2.0622355000004973, + "eval_s": 2.103362100000595, + "log_likelihood": 17471.600454752745, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.12253020000025572, + "n": 660, + "solve_s": 0.12253020000025572, + "outer_iterations": 28, + "inner_iterations": 14, + "total_iterations": 42, + "n_passive": 614, + "warm_start_errors": 36, + "index": 9 + }, + { + "warm_start_error_fraction": 0.045454545454545456, + "reconstruction_s": 2.4241578000001027, + "eval_s": 2.444024000000354, + "log_likelihood": 27063.05376212334, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0647289999997156, + "n": 660, + "solve_s": 0.0647289999997156, + "outer_iterations": 13, + "inner_iterations": 8, + "total_iterations": 21, + "n_passive": 610, + "warm_start_errors": 30, + "index": 10 + }, + { + "warm_start_error_fraction": 0.051515151515151514, + "reconstruction_s": 2.0139682999997603, + "eval_s": 2.0319433999993635, + "log_likelihood": 20513.104421310196, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06780950000029407, + "n": 660, + "solve_s": 0.06780950000029407, + "outer_iterations": 14, + "inner_iterations": 7, + "total_iterations": 21, + "n_passive": 614, + "warm_start_errors": 34, + "index": 11 + }, + { + "warm_start_error_fraction": 0.06212121212121212, + "reconstruction_s": 2.480312099999537, + "eval_s": 2.5019555999997465, + "log_likelihood": 24420.312542276853, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.17890970000007655, + "n": 660, + "solve_s": 0.17890970000007655, + "outer_iterations": 30, + "inner_iterations": 12, + "total_iterations": 42, + "n_passive": 603, + "warm_start_errors": 41, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05454545454545454, + "reconstruction_s": 2.491513599999962, + "eval_s": 2.5175430999997843, + "log_likelihood": 27962.673329269674, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06946239999979298, + "n": 660, + "solve_s": 0.06946239999979298, + "outer_iterations": 16, + "inner_iterations": 11, + "total_iterations": 27, + "n_passive": 610, + "warm_start_errors": 36, + "index": 13 + }, + { + "warm_start_error_fraction": 0.04696969696969697, + "reconstruction_s": 1.7915369000002102, + "eval_s": 1.8077518000000055, + "log_likelihood": 28423.22644263006, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049907900000107475, + "n": 660, + "solve_s": 0.049907900000107475, + "outer_iterations": 13, + "inner_iterations": 9, + "total_iterations": 22, + "n_passive": 610, + "warm_start_errors": 31, + "index": 14 + }, + { + "warm_start_error_fraction": 0.04090909090909091, + "reconstruction_s": 1.6875198999996428, + "eval_s": 1.7020137999998042, + "log_likelihood": 27661.63156534093, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06538079999972979, + "n": 660, + "solve_s": 0.06538079999972979, + "outer_iterations": 29, + "inner_iterations": 22, + "total_iterations": 51, + "n_passive": 613, + "warm_start_errors": 27, + "index": 15 + }, + { + "warm_start_error_fraction": 0.05, + "reconstruction_s": 1.6664144000005763, + "eval_s": 1.687202399999478, + "log_likelihood": 28235.207124831424, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06832529999974213, + "n": 660, + "solve_s": 0.06832529999974213, + "outer_iterations": 21, + "inner_iterations": 13, + "total_iterations": 34, + "n_passive": 620, + "warm_start_errors": 33, + "index": 16 + }, + { + "warm_start_error_fraction": 0.03939393939393939, + "reconstruction_s": 1.8848908999998457, + "eval_s": 1.9016422999993665, + "log_likelihood": 30201.185721482638, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08468639999955485, + "n": 660, + "solve_s": 0.08468639999955485, + "outer_iterations": 14, + "inner_iterations": 9, + "total_iterations": 23, + "n_passive": 623, + "warm_start_errors": 26, + "index": 17 + }, + { + "warm_start_error_fraction": 0.05454545454545454, + "reconstruction_s": 2.2994134000000486, + "eval_s": 2.3157756999999037, + "log_likelihood": 25258.438496056275, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09677389999978914, + "n": 660, + "solve_s": 0.09677389999978914, + "outer_iterations": 21, + "inner_iterations": 9, + "total_iterations": 30, + "n_passive": 613, + "warm_start_errors": 36, + "index": 18 + }, + { + "warm_start_error_fraction": 0.06060606060606061, + "reconstruction_s": 1.55605579999974, + "eval_s": 1.5725405000002866, + "log_likelihood": 23282.663001283567, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09225830000013957, + "n": 660, + "solve_s": 0.09225830000013957, + "outer_iterations": 22, + "inner_iterations": 11, + "total_iterations": 33, + "n_passive": 608, + "warm_start_errors": 40, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.06515151515151515, + "reconstruction_s": 1.916570700000193, + "eval_s": 1.9327598999998372, + "log_likelihood": 19526.535596202903, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06396859999949811, + "n": 660, + "solve_s": 0.06396859999949811, + "outer_iterations": 25, + "inner_iterations": 9, + "total_iterations": 34, + "n_passive": 615, + "warm_start_errors": 43, + "index": 0 + }, + { + "warm_start_error_fraction": 0.048484848484848485, + "reconstruction_s": 1.68157599999995, + "eval_s": 1.7071030999995855, + "log_likelihood": 25185.473530197567, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04846059999999852, + "n": 660, + "solve_s": 0.04846059999999852, + "outer_iterations": 26, + "inner_iterations": 6, + "total_iterations": 32, + "n_passive": 615, + "warm_start_errors": 32, + "index": 1 + }, + { + "warm_start_error_fraction": 0.03939393939393939, + "reconstruction_s": 1.8040454000001773, + "eval_s": 1.8273275999999896, + "log_likelihood": 28654.382000591308, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06268240000008518, + "n": 660, + "solve_s": 0.06268240000008518, + "outer_iterations": 44, + "inner_iterations": 9, + "total_iterations": 53, + "n_passive": 621, + "warm_start_errors": 26, + "index": 2 + }, + { + "warm_start_error_fraction": 0.031818181818181815, + "reconstruction_s": 2.0328591999996206, + "eval_s": 2.058980099999644, + "log_likelihood": 25668.810511952834, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049280900000667316, + "n": 660, + "solve_s": 0.049280900000667316, + "outer_iterations": 32, + "inner_iterations": 8, + "total_iterations": 40, + "n_passive": 618, + "warm_start_errors": 21, + "index": 3 + }, + { + "warm_start_error_fraction": 0.02727272727272727, + "reconstruction_s": 1.9017063000001144, + "eval_s": 1.9213358000006338, + "log_likelihood": 27947.14016408979, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.024164200000086566, + "n": 660, + "solve_s": 0.024164200000086566, + "outer_iterations": 14, + "inner_iterations": 2, + "total_iterations": 16, + "n_passive": 622, + "warm_start_errors": 18, + "index": 4 + }, + { + "warm_start_error_fraction": 0.04696969696969697, + "reconstruction_s": 1.9966222000002745, + "eval_s": 2.0126998000005187, + "log_likelihood": 27806.960101290555, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04386819999945146, + "n": 660, + "solve_s": 0.04386819999945146, + "outer_iterations": 34, + "inner_iterations": 10, + "total_iterations": 44, + "n_passive": 613, + "warm_start_errors": 31, + "index": 5 + }, + { + "warm_start_error_fraction": 0.051515151515151514, + "reconstruction_s": 1.800567499999488, + "eval_s": 1.818730100000721, + "log_likelihood": 26457.657545022594, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03599519999988843, + "n": 660, + "solve_s": 0.03599519999988843, + "outer_iterations": 18, + "inner_iterations": 5, + "total_iterations": 23, + "n_passive": 605, + "warm_start_errors": 34, + "index": 6 + }, + { + "warm_start_error_fraction": 0.06212121212121212, + "reconstruction_s": 1.7824079000001802, + "eval_s": 1.7966777000001457, + "log_likelihood": 24507.466310976808, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.056889099999352766, + "n": 660, + "solve_s": 0.056889099999352766, + "outer_iterations": 37, + "inner_iterations": 10, + "total_iterations": 47, + "n_passive": 602, + "warm_start_errors": 41, + "index": 7 + }, + { + "warm_start_error_fraction": 0.05757575757575758, + "reconstruction_s": 1.6593392000004314, + "eval_s": 1.6769885000003342, + "log_likelihood": 24911.875301704298, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.047877799999696435, + "n": 660, + "solve_s": 0.047877799999696435, + "outer_iterations": 29, + "inner_iterations": 7, + "total_iterations": 36, + "n_passive": 610, + "warm_start_errors": 38, + "index": 8 + }, + { + "warm_start_error_fraction": 0.04242424242424243, + "reconstruction_s": 1.6195429999997941, + "eval_s": 1.6366973000003782, + "log_likelihood": 17471.600454752745, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03730649999943125, + "n": 660, + "solve_s": 0.03730649999943125, + "outer_iterations": 28, + "inner_iterations": 8, + "total_iterations": 36, + "n_passive": 614, + "warm_start_errors": 28, + "index": 9 + }, + { + "warm_start_error_fraction": 0.03333333333333333, + "reconstruction_s": 1.8590266000001066, + "eval_s": 1.875670700000228, + "log_likelihood": 27063.053762123352, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.018878099999710685, + "n": 660, + "solve_s": 0.018878099999710685, + "outer_iterations": 10, + "inner_iterations": 4, + "total_iterations": 14, + "n_passive": 610, + "warm_start_errors": 22, + "index": 10 + }, + { + "warm_start_error_fraction": 0.04090909090909091, + "reconstruction_s": 1.5782626999998683, + "eval_s": 1.5938697999999931, + "log_likelihood": 20513.10442131013, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.027847600000313832, + "n": 660, + "solve_s": 0.027847600000313832, + "outer_iterations": 21, + "inner_iterations": 4, + "total_iterations": 25, + "n_passive": 613, + "warm_start_errors": 27, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0696969696969697, + "reconstruction_s": 1.893771199999719, + "eval_s": 1.9081292999999278, + "log_likelihood": 24420.312542276923, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06144490000042424, + "n": 660, + "solve_s": 0.06144490000042424, + "outer_iterations": 32, + "inner_iterations": 7, + "total_iterations": 39, + "n_passive": 603, + "warm_start_errors": 46, + "index": 12 + }, + { + "warm_start_error_fraction": 0.06060606060606061, + "reconstruction_s": 1.5564406000003146, + "eval_s": 1.57132140000067, + "log_likelihood": 27962.673329269634, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03979369999979099, + "n": 660, + "solve_s": 0.03979369999979099, + "outer_iterations": 34, + "inner_iterations": 5, + "total_iterations": 39, + "n_passive": 609, + "warm_start_errors": 40, + "index": 13 + }, + { + "warm_start_error_fraction": 0.07121212121212121, + "reconstruction_s": 1.6290518999994674, + "eval_s": 1.6433065999999599, + "log_likelihood": 28423.22644263013, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04156409999995958, + "n": 660, + "solve_s": 0.04156409999995958, + "outer_iterations": 29, + "inner_iterations": 8, + "total_iterations": 37, + "n_passive": 610, + "warm_start_errors": 47, + "index": 14 + }, + { + "warm_start_error_fraction": 0.05454545454545454, + "reconstruction_s": 1.538239199999225, + "eval_s": 1.5528260000000955, + "log_likelihood": 27661.631565341107, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03204169999935402, + "n": 660, + "solve_s": 0.03204169999935402, + "outer_iterations": 26, + "inner_iterations": 11, + "total_iterations": 37, + "n_passive": 612, + "warm_start_errors": 36, + "index": 15 + }, + { + "warm_start_error_fraction": 0.06666666666666667, + "reconstruction_s": 1.6381596000001082, + "eval_s": 1.653112200000578, + "log_likelihood": 28235.2071248314, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.046439799999461684, + "n": 660, + "solve_s": 0.046439799999461684, + "outer_iterations": 44, + "inner_iterations": 10, + "total_iterations": 54, + "n_passive": 620, + "warm_start_errors": 44, + "index": 16 + }, + { + "warm_start_error_fraction": 0.025757575757575757, + "reconstruction_s": 1.9578175999995437, + "eval_s": 1.9769165999996403, + "log_likelihood": 30201.185721482645, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03476819999923464, + "n": 660, + "solve_s": 0.03476819999923464, + "outer_iterations": 15, + "inner_iterations": 3, + "total_iterations": 18, + "n_passive": 623, + "warm_start_errors": 17, + "index": 17 + }, + { + "warm_start_error_fraction": 0.03333333333333333, + "reconstruction_s": 1.758613800000603, + "eval_s": 1.7750989999995, + "log_likelihood": 25258.438496056253, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03340410000055272, + "n": 660, + "solve_s": 0.03340410000055272, + "outer_iterations": 27, + "inner_iterations": 4, + "total_iterations": 31, + "n_passive": 613, + "warm_start_errors": 22, + "index": 18 + }, + { + "warm_start_error_fraction": 0.03939393939393939, + "reconstruction_s": 1.7692606000000524, + "eval_s": 1.7872078000000329, + "log_likelihood": 23282.663001283538, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.025285400000029767, + "n": 660, + "solve_s": 0.025285400000029767, + "outer_iterations": 12, + "inner_iterations": 4, + "total_iterations": 16, + "n_passive": 607, + "warm_start_errors": 26, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_mesh_600_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_mesh_600_v2026.8.17.1.png new file mode 100644 index 000000000..adad0009e Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_mesh_600_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_no_edge_zeroing_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_no_edge_zeroing_v2026.8.17.1.json new file mode 100644 index 000000000..4c627ef03 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_no_edge_zeroing_v2026.8.17.1.json @@ -0,0 +1,1602 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "hst", + "model": "no_edge_zeroing", + "configuration": { + "model": "no_edge_zeroing", + "dataset": "hst", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true, + "use_edge_zeroed_pixels": false + }, + "pixel_scale_arcsec": 0.05, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 15361, + "over_sampled_pixels": 17980, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 5.970449800000097, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.02366412213740458, + "mean_warm_start_error_fraction": 0.023702290076335876, + "p90_warm_start_error_fraction": 0.02748091603053435, + "max_warm_start_error_fraction": 0.028244274809160305, + "median_outer_iterations": 24.5, + "mean_outer_iterations": 23.45, + "p90_outer_iterations": 27.0, + "max_outer_iterations": 29.0, + "median_inner_iterations": 8.0, + "mean_inner_iterations": 8.45, + "p90_inner_iterations": 11.0, + "max_inner_iterations": 12.0, + "median_total_iterations": 32.0, + "mean_total_iterations": 31.9, + "p90_total_iterations": 38.0, + "max_total_iterations": 41.0, + "median_n_passive": 1257.0, + "mean_n_passive": 1258.05, + "p90_n_passive": 1263.0, + "max_n_passive": 1263.0, + "median_warm_start_errors": 31.0, + "mean_warm_start_errors": 31.05, + "p90_warm_start_errors": 36.0, + "max_warm_start_errors": 37.0, + "median_solve_s": 0.28433029999996506, + "mean_solve_s": 0.2953639950000252, + "p90_solve_s": 0.3635436599995956, + "max_solve_s": 0.4491017999998803, + "median_reconstruction_s": 2.1473253499993916, + "mean_reconstruction_s": 2.1517514999999094, + "p90_reconstruction_s": 2.255084269999679, + "max_reconstruction_s": 2.6460017999997945, + "median_eval_s": 2.243750149999869, + "mean_eval_s": 2.2398862449999344, + "p90_eval_s": 2.343723410000257, + "max_eval_s": 2.7554336000002877 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.007251908396946565, + "mean_warm_start_error_fraction": 0.008625954198473281, + "p90_warm_start_error_fraction": 0.013816793893129772, + "max_warm_start_error_fraction": 0.02366412213740458, + "median_outer_iterations": 5.5, + "mean_outer_iterations": 8.15, + "p90_outer_iterations": 17.400000000000006, + "max_outer_iterations": 29.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 2.0, + "p90_inner_iterations": 2.200000000000003, + "max_inner_iterations": 12.0, + "median_total_iterations": 7.5, + "mean_total_iterations": 10.15, + "p90_total_iterations": 19.400000000000006, + "max_total_iterations": 41.0, + "median_n_passive": 1256.5, + "mean_n_passive": 1257.75, + "p90_n_passive": 1262.1, + "max_n_passive": 1263.0, + "median_warm_start_errors": 9.5, + "mean_warm_start_errors": 11.3, + "p90_warm_start_errors": 18.1, + "max_warm_start_errors": 31.0, + "median_solve_s": 0.084063050000168, + "mean_solve_s": 0.09567235500007883, + "p90_solve_s": 0.15441879000054554, + "max_solve_s": 0.26967879999938305, + "median_reconstruction_s": 1.7193857000002026, + "mean_reconstruction_s": 1.7460026350000135, + "p90_reconstruction_s": 1.928048250000211, + "max_reconstruction_s": 2.0826659999993353, + "median_eval_s": 1.8082381999997779, + "mean_eval_s": 1.8267093999998907, + "p90_eval_s": 2.007785029999559, + "max_eval_s": 2.1647752999997465 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.2731920227827472e-15, + "max_abs_reconstruction_deviation": 2.462922310542126e-09 + }, + "iteration_reduction_ratio": 4.266666666666667, + "solve_speedup_ratio": 3.3823457511879096, + "eval_speedup_ratio": 1.2408487720258008, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 2.089164900000469, + "eval_s": 2.1834391000002142, + "log_likelihood": 29034.99992341116, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27923269999973854, + "n": 1310, + "solve_s": 0.27923269999973854, + "outer_iterations": 29, + "inner_iterations": 12, + "total_iterations": 41, + "n_passive": 1262, + "warm_start_errors": 31, + "index": 0 + }, + { + "warm_start_error_fraction": 0.024427480916030534, + "reconstruction_s": 2.1413651999992, + "eval_s": 2.230876999999964, + "log_likelihood": 28849.582048876684, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39268259999971633, + "n": 1310, + "solve_s": 0.39268259999971633, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1260, + "warm_start_errors": 32, + "index": 1 + }, + { + "warm_start_error_fraction": 0.024427480916030534, + "reconstruction_s": 2.6460017999997945, + "eval_s": 2.7554336000002877, + "log_likelihood": 28670.06086112797, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33447569999952975, + "n": 1310, + "solve_s": 0.33447569999952975, + "outer_iterations": 25, + "inner_iterations": 11, + "total_iterations": 36, + "n_passive": 1257, + "warm_start_errors": 32, + "index": 2 + }, + { + "warm_start_error_fraction": 0.021374045801526718, + "reconstruction_s": 2.293066700000054, + "eval_s": 2.39432510000006, + "log_likelihood": 28883.348745978776, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28492129999995086, + "n": 1310, + "solve_s": 0.28492129999995086, + "outer_iterations": 27, + "inner_iterations": 7, + "total_iterations": 34, + "n_passive": 1261, + "warm_start_errors": 28, + "index": 3 + }, + { + "warm_start_error_fraction": 0.019083969465648856, + "reconstruction_s": 2.1839912000004915, + "eval_s": 2.2566232999997737, + "log_likelihood": 28888.219309590837, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32471800000075746, + "n": 1310, + "solve_s": 0.32471800000075746, + "outer_iterations": 22, + "inner_iterations": 7, + "total_iterations": 29, + "n_passive": 1262, + "warm_start_errors": 25, + "index": 4 + }, + { + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 2.099209800000608, + "eval_s": 2.176061899999695, + "log_likelihood": 28682.766464090804, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2965254000000641, + "n": 1310, + "solve_s": 0.2965254000000641, + "outer_iterations": 24, + "inner_iterations": 7, + "total_iterations": 31, + "n_passive": 1255, + "warm_start_errors": 36, + "index": 5 + }, + { + "warm_start_error_fraction": 0.022900763358778626, + "reconstruction_s": 2.239533599999959, + "eval_s": 2.3371516999995947, + "log_likelihood": 29003.39603966668, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23001850000036939, + "n": 1310, + "solve_s": 0.23001850000036939, + "outer_iterations": 25, + "inner_iterations": 10, + "total_iterations": 35, + "n_passive": 1262, + "warm_start_errors": 30, + "index": 6 + }, + { + "warm_start_error_fraction": 0.021374045801526718, + "reconstruction_s": 1.892291999999543, + "eval_s": 1.9720000999996046, + "log_likelihood": 29098.354207998218, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20296499999949447, + "n": 1310, + "solve_s": 0.20296499999949447, + "outer_iterations": 13, + "inner_iterations": 8, + "total_iterations": 21, + "n_passive": 1254, + "warm_start_errors": 28, + "index": 7 + }, + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 2.1992413999996643, + "eval_s": 2.277060200000051, + "log_likelihood": 29188.767818380577, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24046700000053534, + "n": 1310, + "solve_s": 0.24046700000053534, + "outer_iterations": 20, + "inner_iterations": 9, + "total_iterations": 29, + "n_passive": 1256, + "warm_start_errors": 31, + "index": 8 + }, + { + "warm_start_error_fraction": 0.022137404580152672, + "reconstruction_s": 2.2508639999996376, + "eval_s": 2.338101000000279, + "log_likelihood": 28740.24100740815, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27296279999973194, + "n": 1310, + "solve_s": 0.27296279999973194, + "outer_iterations": 16, + "inner_iterations": 6, + "total_iterations": 22, + "n_passive": 1254, + "warm_start_errors": 29, + "index": 9 + }, + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 2.1532854999995834, + "eval_s": 2.2566289999995206, + "log_likelihood": 28659.773960522565, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26041500000064843, + "n": 1310, + "solve_s": 0.26041500000064843, + "outer_iterations": 23, + "inner_iterations": 8, + "total_iterations": 31, + "n_passive": 1255, + "warm_start_errors": 31, + "index": 10 + }, + { + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 2.159088999999767, + "eval_s": 2.2600286000006236, + "log_likelihood": 28569.897575338688, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3316534000005049, + "n": 1310, + "solve_s": 0.3316534000005049, + "outer_iterations": 26, + "inner_iterations": 6, + "total_iterations": 32, + "n_passive": 1257, + "warm_start_errors": 36, + "index": 11 + }, + { + "warm_start_error_fraction": 0.025190839694656488, + "reconstruction_s": 1.9415662999999768, + "eval_s": 2.0213905999999042, + "log_likelihood": 28516.18652537265, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2521954000003461, + "n": 1310, + "solve_s": 0.2521954000003461, + "outer_iterations": 26, + "inner_iterations": 7, + "total_iterations": 33, + "n_passive": 1257, + "warm_start_errors": 33, + "index": 12 + }, + { + "warm_start_error_fraction": 0.022900763358778626, + "reconstruction_s": 2.0295082000002367, + "eval_s": 2.1068679999998494, + "log_likelihood": 28593.06738215032, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2559160000000702, + "n": 1310, + "solve_s": 0.2559160000000702, + "outer_iterations": 23, + "inner_iterations": 7, + "total_iterations": 30, + "n_passive": 1257, + "warm_start_errors": 30, + "index": 13 + }, + { + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 2.0253241000000344, + "eval_s": 2.1020054000000528, + "log_likelihood": 28494.22512230963, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2896248999995805, + "n": 1310, + "solve_s": 0.2896248999995805, + "outer_iterations": 24, + "inner_iterations": 8, + "total_iterations": 32, + "n_passive": 1256, + "warm_start_errors": 36, + "index": 14 + }, + { + "warm_start_error_fraction": 0.028244274809160305, + "reconstruction_s": 2.040524099999857, + "eval_s": 2.120097099999839, + "log_likelihood": 28573.68521003123, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2624004999997851, + "n": 1310, + "solve_s": 0.2624004999997851, + "outer_iterations": 26, + "inner_iterations": 6, + "total_iterations": 32, + "n_passive": 1254, + "warm_start_errors": 37, + "index": 15 + }, + { + "warm_start_error_fraction": 0.025190839694656488, + "reconstruction_s": 2.236867199999324, + "eval_s": 2.328301499999725, + "log_likelihood": 28754.169652775512, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4491017999998803, + "n": 1310, + "solve_s": 0.4491017999998803, + "outer_iterations": 23, + "inner_iterations": 10, + "total_iterations": 33, + "n_passive": 1253, + "warm_start_errors": 33, + "index": 16 + }, + { + "warm_start_error_fraction": 0.020610687022900764, + "reconstruction_s": 2.2199688999999125, + "eval_s": 2.3188498999998046, + "log_likelihood": 29100.48957849467, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28373929999997927, + "n": 1310, + "solve_s": 0.28373929999997927, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1263, + "warm_start_errors": 27, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0183206106870229, + "reconstruction_s": 2.118691700000454, + "eval_s": 2.2066192000002047, + "log_likelihood": 29261.487227838043, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3029586000002382, + "n": 1310, + "solve_s": 0.3029586000002382, + "outer_iterations": 16, + "inner_iterations": 7, + "total_iterations": 23, + "n_passive": 1263, + "warm_start_errors": 24, + "index": 18 + }, + { + "warm_start_error_fraction": 0.024427480916030534, + "reconstruction_s": 2.075474399999621, + "eval_s": 2.1558625999996366, + "log_likelihood": 29790.892417880026, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3603059999995821, + "n": 1310, + "solve_s": 0.3603059999995821, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1263, + "warm_start_errors": 32, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.858231700000033, + "eval_s": 1.9240453000002162, + "log_likelihood": 29034.99992341116, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26967879999938305, + "n": 1310, + "solve_s": 0.26967879999938305, + "outer_iterations": 29, + "inner_iterations": 12, + "total_iterations": 41, + "n_passive": 1262, + "warm_start_errors": 31, + "index": 0 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 1.6025470999993559, + "eval_s": 1.683962000000065, + "log_likelihood": 28849.582048876684, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07291760000043723, + "n": 1310, + "solve_s": 0.07291760000043723, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 1260, + "warm_start_errors": 10, + "index": 1 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 1.5354112000004534, + "eval_s": 1.630279199999677, + "log_likelihood": 28670.060861127982, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05691449999994802, + "n": 1310, + "solve_s": 0.05691449999994802, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1257, + "warm_start_errors": 9, + "index": 2 + }, + { + "warm_start_error_fraction": 0.00916030534351145, + "reconstruction_s": 1.5660809000000881, + "eval_s": 1.6370908999997482, + "log_likelihood": 28883.348745978772, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10773710000012215, + "n": 1310, + "solve_s": 0.10773710000012215, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 1261, + "warm_start_errors": 12, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 1.6355100000000675, + "eval_s": 1.7037869999994655, + "log_likelihood": 28888.21930959083, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.044437399999878835, + "n": 1310, + "solve_s": 0.044437399999878835, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1261, + "warm_start_errors": 4, + "index": 4 + }, + { + "warm_start_error_fraction": 0.012977099236641221, + "reconstruction_s": 1.9981226999998398, + "eval_s": 2.0678729000001113, + "log_likelihood": 28682.76646409081, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08566690000043309, + "n": 1310, + "solve_s": 0.08566690000043309, + "outer_iterations": 8, + "inner_iterations": 2, + "total_iterations": 10, + "n_passive": 1254, + "warm_start_errors": 17, + "index": 5 + }, + { + "warm_start_error_fraction": 0.012213740458015267, + "reconstruction_s": 1.6924182000002475, + "eval_s": 1.7633821999997963, + "log_likelihood": 29003.396039666695, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08915830000023561, + "n": 1310, + "solve_s": 0.08915830000023561, + "outer_iterations": 15, + "inner_iterations": 1, + "total_iterations": 16, + "n_passive": 1262, + "warm_start_errors": 16, + "index": 6 + }, + { + "warm_start_error_fraction": 0.013740458015267175, + "reconstruction_s": 1.4576883999998245, + "eval_s": 1.5482270000002245, + "log_likelihood": 29098.3542079982, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08399559999998019, + "n": 1310, + "solve_s": 0.08399559999998019, + "outer_iterations": 5, + "inner_iterations": 4, + "total_iterations": 9, + "n_passive": 1254, + "warm_start_errors": 18, + "index": 7 + }, + { + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 1.7477804999998625, + "eval_s": 1.818304800000078, + "log_likelihood": 29188.767818380573, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08528089999981603, + "n": 1310, + "solve_s": 0.08528089999981603, + "outer_iterations": 10, + "inner_iterations": 2, + "total_iterations": 12, + "n_passive": 1256, + "warm_start_errors": 14, + "index": 8 + }, + { + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 1.7719044000004942, + "eval_s": 1.8428727999998955, + "log_likelihood": 28740.241007408153, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05580340000051365, + "n": 1310, + "solve_s": 0.05580340000051365, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1254, + "warm_start_errors": 6, + "index": 9 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 1.5774464000005537, + "eval_s": 1.6543869999995877, + "log_likelihood": 28659.77396052256, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1262839000000895, + "n": 1310, + "solve_s": 0.1262839000000895, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1255, + "warm_start_errors": 9, + "index": 10 + }, + { + "warm_start_error_fraction": 0.012213740458015267, + "reconstruction_s": 1.690384799999265, + "eval_s": 1.7704106000001048, + "log_likelihood": 28569.89757533868, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08636810000007245, + "n": 1310, + "solve_s": 0.08636810000007245, + "outer_iterations": 10, + "inner_iterations": 2, + "total_iterations": 12, + "n_passive": 1257, + "warm_start_errors": 16, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 1.8736406999996689, + "eval_s": 1.960771900000509, + "log_likelihood": 28516.18652537265, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.056644900000719645, + "n": 1310, + "solve_s": 0.056644900000719645, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1256, + "warm_start_errors": 3, + "index": 12 + }, + { + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 1.6842249000001175, + "eval_s": 1.77453770000011, + "log_likelihood": 28593.06738215032, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0491662999993423, + "n": 1310, + "solve_s": 0.0491662999993423, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1256, + "warm_start_errors": 4, + "index": 13 + }, + { + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 1.8969379999998637, + "eval_s": 2.0011085999994975, + "log_likelihood": 28494.225122309614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08413050000035582, + "n": 1310, + "solve_s": 0.08413050000035582, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1256, + "warm_start_errors": 6, + "index": 14 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 2.0826659999993353, + "eval_s": 2.1647752999997465, + "log_likelihood": 28573.685210031195, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22337580000021262, + "n": 1310, + "solve_s": 0.22337580000021262, + "outer_iterations": 21, + "inner_iterations": 2, + "total_iterations": 23, + "n_passive": 1254, + "warm_start_errors": 8, + "index": 15 + }, + { + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 1.7228076000001238, + "eval_s": 1.8035745999995925, + "log_likelihood": 28754.16965277553, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06549180000001797, + "n": 1310, + "solve_s": 0.06549180000001797, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 1252, + "warm_start_errors": 8, + "index": 16 + }, + { + "warm_start_error_fraction": 0.01450381679389313, + "reconstruction_s": 1.7159638000002815, + "eval_s": 1.8129017999999633, + "log_likelihood": 29100.48957849465, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1467569000005824, + "n": 1310, + "solve_s": 0.1467569000005824, + "outer_iterations": 17, + "inner_iterations": 2, + "total_iterations": 19, + "n_passive": 1263, + "warm_start_errors": 19, + "index": 17 + }, + { + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 1.920262200000252, + "eval_s": 1.9983296999998856, + "log_likelihood": 29261.487227838028, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049273799999355106, + "n": 1310, + "solve_s": 0.049273799999355106, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1262, + "warm_start_errors": 3, + "index": 18 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 1.8900232000005417, + "eval_s": 1.9735666999995374, + "log_likelihood": 29790.892417880037, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07436460000008083, + "n": 1310, + "solve_s": 0.07436460000008083, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 1263, + "warm_start_errors": 13, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.0366412213740458, + "mean_warm_start_error_fraction": 0.037442748091603056, + "p90_warm_start_error_fraction": 0.04770992366412214, + "max_warm_start_error_fraction": 0.05267175572519084, + "median_outer_iterations": 31.0, + "mean_outer_iterations": 33.65, + "p90_outer_iterations": 51.300000000000004, + "max_outer_iterations": 55.0, + "median_inner_iterations": 12.5, + "mean_inner_iterations": 13.05, + "p90_inner_iterations": 17.0, + "max_inner_iterations": 17.0, + "median_total_iterations": 43.0, + "mean_total_iterations": 46.7, + "p90_total_iterations": 66.0, + "max_total_iterations": 71.0, + "median_n_passive": 1236.5, + "mean_n_passive": 1235.5, + "p90_n_passive": 1245.7, + "max_n_passive": 1255.0, + "median_warm_start_errors": 48.0, + "mean_warm_start_errors": 49.05, + "p90_warm_start_errors": 62.50000000000001, + "max_warm_start_errors": 69.0, + "median_solve_s": 0.3415094499996485, + "mean_solve_s": 0.3642573899999661, + "p90_solve_s": 0.46094931000052386, + "max_solve_s": 0.5359252999996897, + "median_reconstruction_s": 2.089465800000198, + "mean_reconstruction_s": 2.1032667349999885, + "p90_reconstruction_s": 2.279345079999894, + "max_reconstruction_s": 2.505739700000049, + "median_eval_s": 2.1639732500002538, + "mean_eval_s": 2.1849989200001803, + "p90_eval_s": 2.3636561499999518, + "max_eval_s": 2.5841135000000577 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.051908396946564885, + "mean_warm_start_error_fraction": 0.05309160305343512, + "p90_warm_start_error_fraction": 0.06374045801526718, + "max_warm_start_error_fraction": 0.07786259541984733, + "median_outer_iterations": 58.0, + "mean_outer_iterations": 55.3, + "p90_outer_iterations": 71.4, + "max_outer_iterations": 76.0, + "median_inner_iterations": 13.0, + "mean_inner_iterations": 13.55, + "p90_inner_iterations": 21.1, + "max_inner_iterations": 24.0, + "median_total_iterations": 73.0, + "mean_total_iterations": 68.85, + "p90_total_iterations": 88.10000000000002, + "max_total_iterations": 99.0, + "median_n_passive": 1236.0, + "mean_n_passive": 1235.25, + "p90_n_passive": 1245.7, + "max_n_passive": 1255.0, + "median_warm_start_errors": 68.0, + "mean_warm_start_errors": 69.55, + "p90_warm_start_errors": 83.5, + "max_warm_start_errors": 102.0, + "median_solve_s": 0.3408449499993367, + "mean_solve_s": 0.35078165000004446, + "p90_solve_s": 0.47214221000067486, + "max_solve_s": 0.5150430000003325, + "median_reconstruction_s": 2.2260051499997644, + "mean_reconstruction_s": 2.2769230549999064, + "p90_reconstruction_s": 2.687547350000114, + "max_reconstruction_s": 3.241809600000124, + "median_eval_s": 2.319138849999945, + "mean_eval_s": 2.3702159250000934, + "p90_eval_s": 2.784962380000616, + "max_eval_s": 3.334204800000407 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.0900162853000532e-14, + "max_abs_reconstruction_deviation": 1.3794121400678705e-09 + }, + "iteration_reduction_ratio": 0.589041095890411, + "solve_speedup_ratio": 1.0019495668054144, + "eval_speedup_ratio": 0.9330934411280745, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 2.505739700000049, + "eval_s": 2.5841135000000577, + "log_likelihood": 19610.36039125053, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4437389999993684, + "n": 1310, + "solve_s": 0.4437389999993684, + "outer_iterations": 55, + "inner_iterations": 11, + "total_iterations": 66, + "n_passive": 1228, + "warm_start_errors": 67, + "index": 0 + }, + { + "warm_start_error_fraction": 0.038931297709923665, + "reconstruction_s": 2.3914129000004323, + "eval_s": 2.484673300000395, + "log_likelihood": 25182.415110795213, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41187209999952756, + "n": 1310, + "solve_s": 0.41187209999952756, + "outer_iterations": 32, + "inner_iterations": 15, + "total_iterations": 47, + "n_passive": 1226, + "warm_start_errors": 51, + "index": 1 + }, + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 2.266893099999834, + "eval_s": 2.3502097999999023, + "log_likelihood": 28569.062384810884, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3417506999994657, + "n": 1310, + "solve_s": 0.3417506999994657, + "outer_iterations": 21, + "inner_iterations": 14, + "total_iterations": 35, + "n_passive": 1245, + "warm_start_errors": 31, + "index": 2 + }, + { + "warm_start_error_fraction": 0.04122137404580153, + "reconstruction_s": 2.253704700000526, + "eval_s": 2.3458747000004223, + "log_likelihood": 25459.91825679345, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.40509130000009463, + "n": 1310, + "solve_s": 0.40509130000009463, + "outer_iterations": 30, + "inner_iterations": 10, + "total_iterations": 40, + "n_passive": 1237, + "warm_start_errors": 54, + "index": 3 + }, + { + "warm_start_error_fraction": 0.03206106870229008, + "reconstruction_s": 2.2231166999999914, + "eval_s": 2.301701000000321, + "log_likelihood": 27874.240680363288, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31764170000042213, + "n": 1310, + "solve_s": 0.31764170000042213, + "outer_iterations": 27, + "inner_iterations": 13, + "total_iterations": 40, + "n_passive": 1244, + "warm_start_errors": 42, + "index": 4 + }, + { + "warm_start_error_fraction": 0.035114503816793895, + "reconstruction_s": 2.2612041000002137, + "eval_s": 2.3291269000001193, + "log_likelihood": 27810.19866298291, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34126819999983127, + "n": 1310, + "solve_s": 0.34126819999983127, + "outer_iterations": 36, + "inner_iterations": 13, + "total_iterations": 49, + "n_passive": 1237, + "warm_start_errors": 46, + "index": 5 + }, + { + "warm_start_error_fraction": 0.04198473282442748, + "reconstruction_s": 2.048921099999461, + "eval_s": 2.1430380000001605, + "log_likelihood": 26505.940530222397, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.48896370000056777, + "n": 1310, + "solve_s": 0.48896370000056777, + "outer_iterations": 49, + "inner_iterations": 17, + "total_iterations": 66, + "n_passive": 1226, + "warm_start_errors": 55, + "index": 6 + }, + { + "warm_start_error_fraction": 0.03435114503816794, + "reconstruction_s": 1.8758379000000787, + "eval_s": 1.9660997000000862, + "log_likelihood": 24492.81466635593, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3042983000004824, + "n": 1310, + "solve_s": 0.3042983000004824, + "outer_iterations": 20, + "inner_iterations": 11, + "total_iterations": 31, + "n_passive": 1224, + "warm_start_errors": 45, + "index": 7 + }, + { + "warm_start_error_fraction": 0.0366412213740458, + "reconstruction_s": 2.0563591999998607, + "eval_s": 2.1388965000005555, + "log_likelihood": 24952.954588361725, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35939979999966454, + "n": 1310, + "solve_s": 0.35939979999966454, + "outer_iterations": 32, + "inner_iterations": 11, + "total_iterations": 43, + "n_passive": 1236, + "warm_start_errors": 48, + "index": 8 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 2.1242316999996547, + "eval_s": 2.197442800000317, + "log_likelihood": 17329.396957087614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38488060000054247, + "n": 1310, + "solve_s": 0.38488060000054247, + "outer_iterations": 51, + "inner_iterations": 12, + "total_iterations": 63, + "n_passive": 1230, + "warm_start_errors": 69, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0366412213740458, + "reconstruction_s": 2.073335600000064, + "eval_s": 2.1441868000001705, + "log_likelihood": 27029.986950519487, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2944674999998824, + "n": 1310, + "solve_s": 0.2944674999998824, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1243, + "warm_start_errors": 48, + "index": 10 + }, + { + "warm_start_error_fraction": 0.03740458015267176, + "reconstruction_s": 1.9551859000002878, + "eval_s": 2.0211387999997896, + "log_likelihood": 20357.33822918338, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3345847999999023, + "n": 1310, + "solve_s": 0.3345847999999023, + "outer_iterations": 31, + "inner_iterations": 11, + "total_iterations": 42, + "n_passive": 1230, + "warm_start_errors": 49, + "index": 11 + }, + { + "warm_start_error_fraction": 0.04045801526717557, + "reconstruction_s": 2.205942700000378, + "eval_s": 2.278730600000017, + "log_likelihood": 24520.37486726214, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3265575000004901, + "n": 1310, + "solve_s": 0.3265575000004901, + "outer_iterations": 34, + "inner_iterations": 16, + "total_iterations": 50, + "n_passive": 1209, + "warm_start_errors": 53, + "index": 12 + }, + { + "warm_start_error_fraction": 0.031297709923664124, + "reconstruction_s": 1.908508399999846, + "eval_s": 1.9946766000002754, + "log_likelihood": 28035.381114657645, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37637699999959295, + "n": 1310, + "solve_s": 0.37637699999959295, + "outer_iterations": 31, + "inner_iterations": 16, + "total_iterations": 47, + "n_passive": 1233, + "warm_start_errors": 41, + "index": 13 + }, + { + "warm_start_error_fraction": 0.03282442748091603, + "reconstruction_s": 1.9753190000001268, + "eval_s": 2.0741894000002503, + "log_likelihood": 28358.559706509935, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29351929999938875, + "n": 1310, + "solve_s": 0.29351929999938875, + "outer_iterations": 26, + "inner_iterations": 12, + "total_iterations": 38, + "n_passive": 1242, + "warm_start_errors": 43, + "index": 14 + }, + { + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 1.8382741999994323, + "eval_s": 1.912472500000149, + "log_likelihood": 27753.91959514748, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2826896000005945, + "n": 1310, + "solve_s": 0.2826896000005945, + "outer_iterations": 26, + "inner_iterations": 17, + "total_iterations": 43, + "n_passive": 1242, + "warm_start_errors": 36, + "index": 15 + }, + { + "warm_start_error_fraction": 0.031297709923664124, + "reconstruction_s": 1.835775300000023, + "eval_s": 1.9204037999998036, + "log_likelihood": 28124.568319931564, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26677339999969263, + "n": 1310, + "solve_s": 0.26677339999969263, + "outer_iterations": 24, + "inner_iterations": 13, + "total_iterations": 37, + "n_passive": 1255, + "warm_start_errors": 41, + "index": 16 + }, + { + "warm_start_error_fraction": 0.029770992366412213, + "reconstruction_s": 1.9072801999991498, + "eval_s": 1.9982298000004448, + "log_likelihood": 30068.12156445181, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3175113999996029, + "n": 1310, + "solve_s": 0.3175113999996029, + "outer_iterations": 24, + "inner_iterations": 11, + "total_iterations": 35, + "n_passive": 1252, + "warm_start_errors": 39, + "index": 17 + }, + { + "warm_start_error_fraction": 0.04732824427480916, + "reconstruction_s": 2.252696300000025, + "eval_s": 2.331014200000027, + "log_likelihood": 25172.01783934843, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5359252999996897, + "n": 1310, + "solve_s": 0.5359252999996897, + "outer_iterations": 54, + "inner_iterations": 17, + "total_iterations": 71, + "n_passive": 1240, + "warm_start_errors": 62, + "index": 18 + }, + { + "warm_start_error_fraction": 0.046564885496183206, + "reconstruction_s": 2.1055960000003324, + "eval_s": 2.183759700000337, + "log_likelihood": 23342.637661178655, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.45783660000051896, + "n": 1310, + "solve_s": 0.45783660000051896, + "outer_iterations": 43, + "inner_iterations": 10, + "total_iterations": 53, + "n_passive": 1231, + "warm_start_errors": 61, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 2.318486899999698, + "eval_s": 2.406817499999306, + "log_likelihood": 19610.36039125053, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5150430000003325, + "n": 1310, + "solve_s": 0.5150430000003325, + "outer_iterations": 55, + "inner_iterations": 11, + "total_iterations": 66, + "n_passive": 1228, + "warm_start_errors": 67, + "index": 0 + }, + { + "warm_start_error_fraction": 0.0633587786259542, + "reconstruction_s": 2.0192964999996548, + "eval_s": 2.0902842000004966, + "log_likelihood": 25182.41511079518, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3408581999992748, + "n": 1310, + "solve_s": 0.3408581999992748, + "outer_iterations": 58, + "inner_iterations": 15, + "total_iterations": 73, + "n_passive": 1225, + "warm_start_errors": 83, + "index": 1 + }, + { + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 1.875759100000323, + "eval_s": 1.9455515000008745, + "log_likelihood": 28569.06238481089, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3637735000002067, + "n": 1310, + "solve_s": 0.3637735000002067, + "outer_iterations": 75, + "inner_iterations": 24, + "total_iterations": 99, + "n_passive": 1245, + "warm_start_errors": 80, + "index": 2 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 1.7994077999992442, + "eval_s": 1.8859049000002415, + "log_likelihood": 25459.918256793477, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32140010000057373, + "n": 1310, + "solve_s": 0.32140010000057373, + "outer_iterations": 66, + "inner_iterations": 14, + "total_iterations": 80, + "n_passive": 1237, + "warm_start_errors": 66, + "index": 3 + }, + { + "warm_start_error_fraction": 0.04351145038167939, + "reconstruction_s": 1.9441621999994823, + "eval_s": 2.048215799999525, + "log_likelihood": 27874.240680363324, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2423174999994444, + "n": 1310, + "solve_s": 0.2423174999994444, + "outer_iterations": 37, + "inner_iterations": 8, + "total_iterations": 45, + "n_passive": 1244, + "warm_start_errors": 57, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 2.143727700000454, + "eval_s": 2.25477839999985, + "log_likelihood": 27810.19866298294, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35239350000028935, + "n": 1310, + "solve_s": 0.35239350000028935, + "outer_iterations": 54, + "inner_iterations": 19, + "total_iterations": 73, + "n_passive": 1237, + "warm_start_errors": 59, + "index": 5 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 2.3704060999998546, + "eval_s": 2.4823809999998048, + "log_likelihood": 26505.940530222408, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4034030000002531, + "n": 1310, + "solve_s": 0.4034030000002531, + "outer_iterations": 66, + "inner_iterations": 12, + "total_iterations": 78, + "n_passive": 1226, + "warm_start_errors": 79, + "index": 6 + }, + { + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 2.411452299999837, + "eval_s": 2.5061290000003282, + "log_likelihood": 24492.81466635587, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3025409000001673, + "n": 1310, + "solve_s": 0.3025409000001673, + "outer_iterations": 61, + "inner_iterations": 16, + "total_iterations": 77, + "n_passive": 1224, + "warm_start_errors": 88, + "index": 7 + }, + { + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 1.9176618999999846, + "eval_s": 2.0077408999995896, + "log_likelihood": 24952.954588361656, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29915170000003855, + "n": 1310, + "solve_s": 0.29915170000003855, + "outer_iterations": 58, + "inner_iterations": 7, + "total_iterations": 65, + "n_passive": 1235, + "warm_start_errors": 75, + "index": 8 + }, + { + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 2.0306909999999334, + "eval_s": 2.122649399999318, + "log_likelihood": 17329.396957087614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38621990000046935, + "n": 1310, + "solve_s": 0.38621990000046935, + "outer_iterations": 71, + "inner_iterations": 16, + "total_iterations": 87, + "n_passive": 1230, + "warm_start_errors": 69, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 2.1500740000001315, + "eval_s": 2.23713300000054, + "log_likelihood": 27029.986950519495, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2714893000002121, + "n": 1310, + "solve_s": 0.2714893000002121, + "outer_iterations": 45, + "inner_iterations": 11, + "total_iterations": 56, + "n_passive": 1243, + "warm_start_errors": 59, + "index": 10 + }, + { + "warm_start_error_fraction": 0.044274809160305344, + "reconstruction_s": 2.2327098999994632, + "eval_s": 2.3242958999999246, + "log_likelihood": 20357.338229183373, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.305782100000215, + "n": 1310, + "solve_s": 0.305782100000215, + "outer_iterations": 36, + "inner_iterations": 6, + "total_iterations": 42, + "n_passive": 1229, + "warm_start_errors": 58, + "index": 11 + }, + { + "warm_start_error_fraction": 0.07786259541984733, + "reconstruction_s": 2.6546164000001227, + "eval_s": 2.7528495000005933, + "log_likelihood": 24520.374867262166, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5075590999995256, + "n": 1310, + "solve_s": 0.5075590999995256, + "outer_iterations": 63, + "inner_iterations": 21, + "total_iterations": 84, + "n_passive": 1209, + "warm_start_errors": 102, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05801526717557252, + "reconstruction_s": 3.241809600000124, + "eval_s": 3.334204800000407, + "log_likelihood": 28035.38111465795, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.44143890000032115, + "n": 1310, + "solve_s": 0.44143890000032115, + "outer_iterations": 62, + "inner_iterations": 18, + "total_iterations": 80, + "n_passive": 1233, + "warm_start_errors": 76, + "index": 13 + }, + { + "warm_start_error_fraction": 0.06183206106870229, + "reconstruction_s": 2.4272266999996646, + "eval_s": 2.5212438999997175, + "log_likelihood": 28358.55970650994, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3554652000002534, + "n": 1310, + "solve_s": 0.3554652000002534, + "outer_iterations": 66, + "inner_iterations": 17, + "total_iterations": 83, + "n_passive": 1242, + "warm_start_errors": 81, + "index": 14 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 2.983925900000031, + "eval_s": 3.0739783000008174, + "log_likelihood": 27753.91959514753, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34083169999939855, + "n": 1310, + "solve_s": 0.34083169999939855, + "outer_iterations": 43, + "inner_iterations": 12, + "total_iterations": 55, + "n_passive": 1241, + "warm_start_errors": 65, + "index": 15 + }, + { + "warm_start_error_fraction": 0.05343511450381679, + "reconstruction_s": 2.2193004000000656, + "eval_s": 2.313981799999965, + "log_likelihood": 28124.56831993156, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4682070000008025, + "n": 1310, + "solve_s": 0.4682070000008025, + "outer_iterations": 76, + "inner_iterations": 22, + "total_iterations": 98, + "n_passive": 1255, + "warm_start_errors": 70, + "index": 16 + }, + { + "warm_start_error_fraction": 0.038931297709923665, + "reconstruction_s": 2.298410200000035, + "eval_s": 2.3877188000005845, + "log_likelihood": 30068.1215644518, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.251635399999941, + "n": 1310, + "solve_s": 0.251635399999941, + "outer_iterations": 41, + "inner_iterations": 9, + "total_iterations": 50, + "n_passive": 1252, + "warm_start_errors": 51, + "index": 17 + }, + { + "warm_start_error_fraction": 0.03969465648854962, + "reconstruction_s": 2.568133900000248, + "eval_s": 2.660646999999699, + "log_likelihood": 25172.017839348387, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31683389999943756, + "n": 1310, + "solve_s": 0.31683389999943756, + "outer_iterations": 42, + "inner_iterations": 7, + "total_iterations": 49, + "n_passive": 1240, + "warm_start_errors": 52, + "index": 18 + }, + { + "warm_start_error_fraction": 0.04122137404580153, + "reconstruction_s": 1.9312025999997786, + "eval_s": 2.047812900000281, + "log_likelihood": 23342.637661178658, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22928909999973257, + "n": 1310, + "solve_s": 0.22928909999973257, + "outer_iterations": 31, + "inner_iterations": 6, + "total_iterations": 37, + "n_passive": 1230, + "warm_start_errors": 54, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_no_edge_zeroing_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_no_edge_zeroing_v2026.8.17.1.png new file mode 100644 index 000000000..469f2809f Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_no_edge_zeroing_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_powerlaw_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_powerlaw_v2026.8.17.1.json new file mode 100644 index 000000000..1121bd723 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_powerlaw_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "hst", + "model": "powerlaw", + "configuration": { + "model": "powerlaw", + "dataset": "hst", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.05, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 15361, + "over_sampled_pixels": 17980, + "source_pixels": 1250, + "free_parameters": 12, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 6.655396299999666, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.022137404580152672, + "mean_warm_start_error_fraction": 0.022595419847328248, + "p90_warm_start_error_fraction": 0.02687022900763359, + "max_warm_start_error_fraction": 0.028244274809160305, + "median_outer_iterations": 20.0, + "mean_outer_iterations": 20.7, + "p90_outer_iterations": 25.200000000000003, + "max_outer_iterations": 29.0, + "median_inner_iterations": 9.0, + "mean_inner_iterations": 8.55, + "p90_inner_iterations": 10.0, + "max_inner_iterations": 12.0, + "median_total_iterations": 29.0, + "mean_total_iterations": 29.25, + "p90_total_iterations": 37.0, + "max_total_iterations": 37.0, + "median_n_passive": 1257.5, + "mean_n_passive": 1257.2, + "p90_n_passive": 1260.1, + "max_n_passive": 1264.0, + "median_warm_start_errors": 29.0, + "mean_warm_start_errors": 29.6, + "p90_warm_start_errors": 35.2, + "max_warm_start_errors": 37.0, + "median_solve_s": 0.24767779999956474, + "mean_solve_s": 0.2694649799999752, + "p90_solve_s": 0.37640014000025984, + "max_solve_s": 0.39313380000021425, + "median_reconstruction_s": 1.9097034500000518, + "mean_reconstruction_s": 2.002004815000146, + "p90_reconstruction_s": 2.2958839200003243, + "max_reconstruction_s": 2.466360800000075, + "median_eval_s": 1.9807245500005592, + "mean_eval_s": 2.0813928849999686, + "p90_eval_s": 2.373834609999358, + "max_eval_s": 2.551812499999869 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.009923664122137405, + "mean_warm_start_error_fraction": 0.009961832061068704, + "p90_warm_start_error_fraction": 0.01625954198473283, + "max_warm_start_error_fraction": 0.02366412213740458, + "median_outer_iterations": 7.0, + "mean_outer_iterations": 7.0, + "p90_outer_iterations": 11.100000000000001, + "max_outer_iterations": 20.0, + "median_inner_iterations": 2.0, + "mean_inner_iterations": 1.95, + "p90_inner_iterations": 3.0, + "max_inner_iterations": 8.0, + "median_total_iterations": 9.0, + "mean_total_iterations": 8.95, + "p90_total_iterations": 13.0, + "max_total_iterations": 28.0, + "median_n_passive": 1257.5, + "mean_n_passive": 1257.05, + "p90_n_passive": 1260.1, + "max_n_passive": 1264.0, + "median_warm_start_errors": 13.0, + "mean_warm_start_errors": 13.05, + "p90_warm_start_errors": 21.300000000000004, + "max_warm_start_errors": 31.0, + "median_solve_s": 0.08807860000024448, + "mean_solve_s": 0.10640251000004355, + "p90_solve_s": 0.1524920099998781, + "max_solve_s": 0.346443200000067, + "median_reconstruction_s": 1.8727113499999177, + "mean_reconstruction_s": 2.035670615000072, + "p90_reconstruction_s": 2.7196914199998905, + "max_reconstruction_s": 3.222928000000138, + "median_eval_s": 1.9659336499998972, + "mean_eval_s": 2.1280566050000287, + "p90_eval_s": 2.8177533800004317, + "max_eval_s": 3.338736499999868 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.7262372922247078e-15, + "max_abs_reconstruction_deviation": 1.145310246641884e-08 + }, + "iteration_reduction_ratio": 3.2222222222222223, + "solve_speedup_ratio": 2.812008819382657, + "eval_speedup_ratio": 1.0075236008095507, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.7481863000002704, + "eval_s": 1.8302185000002282, + "log_likelihood": 29030.46251019762, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21601119999922958, + "n": 1310, + "solve_s": 0.21601119999922958, + "outer_iterations": 20, + "inner_iterations": 8, + "total_iterations": 28, + "n_passive": 1261, + "warm_start_errors": 31, + "index": 0 + }, + { + "warm_start_error_fraction": 0.01984732824427481, + "reconstruction_s": 1.885213499999736, + "eval_s": 1.9619071000006443, + "log_likelihood": 29374.500756720605, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20900330000040412, + "n": 1310, + "solve_s": 0.20900330000040412, + "outer_iterations": 16, + "inner_iterations": 8, + "total_iterations": 24, + "n_passive": 1258, + "warm_start_errors": 26, + "index": 1 + }, + { + "warm_start_error_fraction": 0.022900763358778626, + "reconstruction_s": 2.24166680000053, + "eval_s": 2.3164120000001276, + "log_likelihood": 29448.411474992085, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39313380000021425, + "n": 1310, + "solve_s": 0.39313380000021425, + "outer_iterations": 22, + "inner_iterations": 10, + "total_iterations": 32, + "n_passive": 1256, + "warm_start_errors": 30, + "index": 2 + }, + { + "warm_start_error_fraction": 0.025954198473282442, + "reconstruction_s": 2.052171499999531, + "eval_s": 2.13021099999969, + "log_likelihood": 29469.307202749496, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24704359999941516, + "n": 1310, + "solve_s": 0.24704359999941516, + "outer_iterations": 19, + "inner_iterations": 9, + "total_iterations": 28, + "n_passive": 1259, + "warm_start_errors": 34, + "index": 3 + }, + { + "warm_start_error_fraction": 0.020610687022900764, + "reconstruction_s": 2.3894022000004043, + "eval_s": 2.471777199999451, + "log_likelihood": 29351.52274423297, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33009320000019216, + "n": 1310, + "solve_s": 0.33009320000019216, + "outer_iterations": 21, + "inner_iterations": 6, + "total_iterations": 27, + "n_passive": 1258, + "warm_start_errors": 27, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0183206106870229, + "reconstruction_s": 2.211844800000108, + "eval_s": 2.32455630000004, + "log_likelihood": 29491.289991099686, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3044476000004579, + "n": 1310, + "solve_s": 0.3044476000004579, + "outer_iterations": 21, + "inner_iterations": 8, + "total_iterations": 29, + "n_passive": 1257, + "warm_start_errors": 24, + "index": 5 + }, + { + "warm_start_error_fraction": 0.022137404580152672, + "reconstruction_s": 2.2718064000000595, + "eval_s": 2.3447738000004392, + "log_likelihood": 29548.050516097668, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27011059999949794, + "n": 1310, + "solve_s": 0.27011059999949794, + "outer_iterations": 17, + "inner_iterations": 9, + "total_iterations": 26, + "n_passive": 1254, + "warm_start_errors": 29, + "index": 6 + }, + { + "warm_start_error_fraction": 0.019083969465648856, + "reconstruction_s": 1.8622793000004094, + "eval_s": 1.936475900000005, + "log_likelihood": 29323.894480159317, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2358012000004237, + "n": 1310, + "solve_s": 0.2358012000004237, + "outer_iterations": 24, + "inner_iterations": 7, + "total_iterations": 31, + "n_passive": 1255, + "warm_start_errors": 25, + "index": 7 + }, + { + "warm_start_error_fraction": 0.019083969465648856, + "reconstruction_s": 1.7779583999999886, + "eval_s": 1.8470820000002277, + "log_likelihood": 29480.885528341623, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24831199999971432, + "n": 1310, + "solve_s": 0.24831199999971432, + "outer_iterations": 20, + "inner_iterations": 10, + "total_iterations": 30, + "n_passive": 1264, + "warm_start_errors": 25, + "index": 8 + }, + { + "warm_start_error_fraction": 0.022137404580152672, + "reconstruction_s": 1.8264306000000943, + "eval_s": 1.890513600000304, + "log_likelihood": 29347.862823963813, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20458510000025854, + "n": 1310, + "solve_s": 0.20458510000025854, + "outer_iterations": 20, + "inner_iterations": 9, + "total_iterations": 29, + "n_passive": 1260, + "warm_start_errors": 29, + "index": 9 + }, + { + "warm_start_error_fraction": 0.020610687022900764, + "reconstruction_s": 1.7047935000000507, + "eval_s": 1.7732094999992114, + "log_likelihood": 29063.818301664818, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.16915199999948527, + "n": 1310, + "solve_s": 0.16915199999948527, + "outer_iterations": 13, + "inner_iterations": 5, + "total_iterations": 18, + "n_passive": 1257, + "warm_start_errors": 27, + "index": 10 + }, + { + "warm_start_error_fraction": 0.021374045801526718, + "reconstruction_s": 1.9341934000003675, + "eval_s": 1.9995420000004742, + "log_likelihood": 29024.17778222286, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3321200000000317, + "n": 1310, + "solve_s": 0.3321200000000317, + "outer_iterations": 27, + "inner_iterations": 10, + "total_iterations": 37, + "n_passive": 1252, + "warm_start_errors": 28, + "index": 11 + }, + { + "warm_start_error_fraction": 0.020610687022900764, + "reconstruction_s": 1.7594020999995337, + "eval_s": 1.8276688999994803, + "log_likelihood": 29173.372366686912, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.19580669999959355, + "n": 1310, + "solve_s": 0.19580669999959355, + "outer_iterations": 16, + "inner_iterations": 7, + "total_iterations": 23, + "n_passive": 1259, + "warm_start_errors": 27, + "index": 12 + }, + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.6548774000002595, + "eval_s": 1.7234402999993108, + "log_likelihood": 29042.713896459867, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18770240000048943, + "n": 1310, + "solve_s": 0.18770240000048943, + "outer_iterations": 18, + "inner_iterations": 7, + "total_iterations": 25, + "n_passive": 1260, + "warm_start_errors": 31, + "index": 13 + }, + { + "warm_start_error_fraction": 0.022900763358778626, + "reconstruction_s": 1.8269286000004286, + "eval_s": 1.9287646000002496, + "log_likelihood": 29310.96162436084, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2328114000001733, + "n": 1310, + "solve_s": 0.2328114000001733, + "outer_iterations": 19, + "inner_iterations": 10, + "total_iterations": 29, + "n_passive": 1259, + "warm_start_errors": 30, + "index": 14 + }, + { + "warm_start_error_fraction": 0.025954198473282442, + "reconstruction_s": 2.1985113999999157, + "eval_s": 2.2673955999998725, + "log_likelihood": 29240.481383671337, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2702183999999761, + "n": 1310, + "solve_s": 0.2702183999999761, + "outer_iterations": 25, + "inner_iterations": 9, + "total_iterations": 34, + "n_passive": 1254, + "warm_start_errors": 34, + "index": 15 + }, + { + "warm_start_error_fraction": 0.01984732824427481, + "reconstruction_s": 1.8073087000002488, + "eval_s": 1.916815200000201, + "log_likelihood": 29214.555593318924, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22672239999974408, + "n": 1310, + "solve_s": 0.22672239999974408, + "outer_iterations": 17, + "inner_iterations": 10, + "total_iterations": 27, + "n_passive": 1260, + "warm_start_errors": 26, + "index": 16 + }, + { + "warm_start_error_fraction": 0.026717557251908396, + "reconstruction_s": 2.285493000000315, + "eval_s": 2.3629520999993474, + "log_likelihood": 29504.462410057873, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3574705000000904, + "n": 1310, + "solve_s": 0.3574705000000904, + "outer_iterations": 25, + "inner_iterations": 9, + "total_iterations": 34, + "n_passive": 1255, + "warm_start_errors": 35, + "index": 17 + }, + { + "warm_start_error_fraction": 0.028244274809160305, + "reconstruction_s": 2.1352676000005886, + "eval_s": 2.2223296000001938, + "log_likelihood": 29082.245137242364, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3830982999998014, + "n": 1310, + "solve_s": 0.3830982999998014, + "outer_iterations": 29, + "inner_iterations": 8, + "total_iterations": 37, + "n_passive": 1255, + "warm_start_errors": 37, + "index": 18 + }, + { + "warm_start_error_fraction": 0.028244274809160305, + "reconstruction_s": 2.466360800000075, + "eval_s": 2.551812499999869, + "log_likelihood": 28905.65309207483, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3756559000003108, + "n": 1310, + "solve_s": 0.3756559000003108, + "outer_iterations": 25, + "inner_iterations": 12, + "total_iterations": 37, + "n_passive": 1251, + "warm_start_errors": 37, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 2.311286499999369, + "eval_s": 2.404512600000089, + "log_likelihood": 29030.46251019762, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.346443200000067, + "n": 1310, + "solve_s": 0.346443200000067, + "outer_iterations": 20, + "inner_iterations": 8, + "total_iterations": 28, + "n_passive": 1261, + "warm_start_errors": 31, + "index": 0 + }, + { + "warm_start_error_fraction": 0.0053435114503816794, + "reconstruction_s": 3.102676700000302, + "eval_s": 3.256623799999943, + "log_likelihood": 29374.500756720598, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18422699999973702, + "n": 1310, + "solve_s": 0.18422699999973702, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 1258, + "warm_start_errors": 7, + "index": 1 + }, + { + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 2.2103759000001446, + "eval_s": 2.2803726000001916, + "log_likelihood": 29448.411474992077, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06508819999999105, + "n": 1310, + "solve_s": 0.06508819999999105, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1255, + "warm_start_errors": 9, + "index": 2 + }, + { + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 1.7319709000003058, + "eval_s": 1.8279428999994707, + "log_likelihood": 29469.30720274946, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11006300000008196, + "n": 1310, + "solve_s": 0.11006300000008196, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 1259, + "warm_start_errors": 14, + "index": 3 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 3.222928000000138, + "eval_s": 3.338736499999868, + "log_likelihood": 29351.522744232963, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.14896589999989374, + "n": 1310, + "solve_s": 0.14896589999989374, + "outer_iterations": 7, + "inner_iterations": 2, + "total_iterations": 9, + "n_passive": 1258, + "warm_start_errors": 13, + "index": 4 + }, + { + "warm_start_error_fraction": 0.008396946564885497, + "reconstruction_s": 2.449044899999535, + "eval_s": 2.5865451999998186, + "log_likelihood": 29491.289991099675, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08103089999985968, + "n": 1310, + "solve_s": 0.08103089999985968, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 1257, + "warm_start_errors": 11, + "index": 5 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 2.6771374999998443, + "eval_s": 2.7689900000004855, + "log_likelihood": 29548.050516097668, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10170920000018668, + "n": 1310, + "solve_s": 0.10170920000018668, + "outer_iterations": 4, + "inner_iterations": 2, + "total_iterations": 6, + "n_passive": 1253, + "warm_start_errors": 10, + "index": 6 + }, + { + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 2.0106696000002557, + "eval_s": 2.1036027999998623, + "log_likelihood": 29323.89448015933, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08225620000030176, + "n": 1310, + "solve_s": 0.08225620000030176, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1255, + "warm_start_errors": 6, + "index": 7 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 1.81443709999985, + "eval_s": 1.9096469000005527, + "log_likelihood": 29480.885528341627, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11449290000018664, + "n": 1310, + "solve_s": 0.11449290000018664, + "outer_iterations": 11, + "inner_iterations": 1, + "total_iterations": 12, + "n_passive": 1264, + "warm_start_errors": 13, + "index": 8 + }, + { + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 1.6467328999997335, + "eval_s": 1.7229489999999714, + "log_likelihood": 29347.862823963813, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07971910000014759, + "n": 1310, + "solve_s": 0.07971910000014759, + "outer_iterations": 4, + "inner_iterations": 2, + "total_iterations": 6, + "n_passive": 1260, + "warm_start_errors": 10, + "index": 9 + }, + { + "warm_start_error_fraction": 0.0183206106870229, + "reconstruction_s": 1.5916622000004281, + "eval_s": 1.6737965000002077, + "log_likelihood": 29063.818301664847, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08898810000027879, + "n": 1310, + "solve_s": 0.08898810000027879, + "outer_iterations": 11, + "inner_iterations": 2, + "total_iterations": 13, + "n_passive": 1256, + "warm_start_errors": 24, + "index": 10 + }, + { + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 1.7421864999996615, + "eval_s": 1.8104185000001962, + "log_likelihood": 29024.177782222865, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.060190700000021025, + "n": 1310, + "solve_s": 0.060190700000021025, + "outer_iterations": 0, + "inner_iterations": 1, + "total_iterations": 1, + "n_passive": 1252, + "warm_start_errors": 4, + "index": 11 + }, + { + "warm_start_error_fraction": 0.012977099236641221, + "reconstruction_s": 1.4686791000003723, + "eval_s": 1.5386679000002914, + "log_likelihood": 29173.372366686912, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08849329999975453, + "n": 1310, + "solve_s": 0.08849329999975453, + "outer_iterations": 12, + "inner_iterations": 1, + "total_iterations": 13, + "n_passive": 1259, + "warm_start_errors": 17, + "index": 12 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 1.6461127000002307, + "eval_s": 1.7383417999999438, + "log_likelihood": 29042.713896459863, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07534360000045126, + "n": 1310, + "solve_s": 0.07534360000045126, + "outer_iterations": 7, + "inner_iterations": 2, + "total_iterations": 9, + "n_passive": 1260, + "warm_start_errors": 13, + "index": 13 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 1.6103060000004916, + "eval_s": 1.6825975000001563, + "log_likelihood": 29310.961624360876, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10019049999937124, + "n": 1310, + "solve_s": 0.10019049999937124, + "outer_iterations": 9, + "inner_iterations": 3, + "total_iterations": 12, + "n_passive": 1259, + "warm_start_errors": 13, + "index": 14 + }, + { + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 1.485310600000048, + "eval_s": 1.5578238999996756, + "log_likelihood": 29240.481383671366, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06248250000044209, + "n": 1310, + "solve_s": 0.06248250000044209, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1254, + "warm_start_errors": 13, + "index": 15 + }, + { + "warm_start_error_fraction": 0.00916030534351145, + "reconstruction_s": 2.1692190000003393, + "eval_s": 2.260515900000428, + "log_likelihood": 29214.55559331891, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09890739999991638, + "n": 1310, + "solve_s": 0.09890739999991638, + "outer_iterations": 9, + "inner_iterations": 2, + "total_iterations": 11, + "n_passive": 1260, + "warm_start_errors": 12, + "index": 16 + }, + { + "warm_start_error_fraction": 0.01603053435114504, + "reconstruction_s": 1.83429679999972, + "eval_s": 1.9355178000005253, + "log_likelihood": 29504.462410057822, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08766390000073443, + "n": 1310, + "solve_s": 0.08766390000073443, + "outer_iterations": 8, + "inner_iterations": 3, + "total_iterations": 11, + "n_passive": 1255, + "warm_start_errors": 21, + "index": 17 + }, + { + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 1.9111259000001155, + "eval_s": 1.996349499999269, + "log_likelihood": 29082.245137242375, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08621749999929307, + "n": 1310, + "solve_s": 0.08621749999929307, + "outer_iterations": 8, + "inner_iterations": 2, + "total_iterations": 10, + "n_passive": 1255, + "warm_start_errors": 14, + "index": 18 + }, + { + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 2.077253500000552, + "eval_s": 2.167180499999631, + "log_likelihood": 28905.653092074812, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06557710000015504, + "n": 1310, + "solve_s": 0.06557710000015504, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1251, + "warm_start_errors": 6, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.037022900763358776, + "mean_warm_start_error_fraction": 0.03652671755725191, + "p90_warm_start_error_fraction": 0.045572519083969476, + "max_warm_start_error_fraction": 0.05419847328244275, + "median_outer_iterations": 35.0, + "mean_outer_iterations": 34.8, + "p90_outer_iterations": 49.0, + "max_outer_iterations": 57.0, + "median_inner_iterations": 11.0, + "mean_inner_iterations": 11.1, + "p90_inner_iterations": 14.100000000000001, + "max_inner_iterations": 18.0, + "median_total_iterations": 47.0, + "mean_total_iterations": 45.9, + "p90_total_iterations": 63.1, + "max_total_iterations": 69.0, + "median_n_passive": 1237.0, + "mean_n_passive": 1234.4, + "p90_n_passive": 1251.4, + "max_n_passive": 1257.0, + "median_warm_start_errors": 48.5, + "mean_warm_start_errors": 47.85, + "p90_warm_start_errors": 59.70000000000001, + "max_warm_start_errors": 71.0, + "median_solve_s": 0.35698415000024397, + "mean_solve_s": 0.3636216250000416, + "p90_solve_s": 0.44676568000013506, + "max_solve_s": 0.4952321999999185, + "median_reconstruction_s": 2.226370649999808, + "mean_reconstruction_s": 2.2637816399999338, + "p90_reconstruction_s": 2.7120006699997248, + "max_reconstruction_s": 2.862619100000302, + "median_eval_s": 2.298028649999651, + "mean_eval_s": 2.3475029599998836, + "p90_eval_s": 2.8030936099999963, + "max_eval_s": 2.9510639999998602 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.05, + "mean_warm_start_error_fraction": 0.051717557251908394, + "p90_warm_start_error_fraction": 0.06297709923664123, + "max_warm_start_error_fraction": 0.06870229007633588, + "median_outer_iterations": 51.0, + "mean_outer_iterations": 53.05, + "p90_outer_iterations": 74.30000000000001, + "max_outer_iterations": 87.0, + "median_inner_iterations": 13.0, + "mean_inner_iterations": 13.5, + "p90_inner_iterations": 20.300000000000004, + "max_inner_iterations": 28.0, + "median_total_iterations": 62.5, + "mean_total_iterations": 66.55, + "p90_total_iterations": 90.2, + "max_total_iterations": 115.0, + "median_n_passive": 1236.5, + "mean_n_passive": 1234.2, + "p90_n_passive": 1250.4, + "max_n_passive": 1257.0, + "median_warm_start_errors": 65.5, + "mean_warm_start_errors": 67.75, + "p90_warm_start_errors": 82.5, + "max_warm_start_errors": 90.0, + "median_solve_s": 0.3012117000002945, + "mean_solve_s": 0.3074447700000292, + "p90_solve_s": 0.37471267000019, + "max_solve_s": 0.5517626000000746, + "median_reconstruction_s": 1.9508664999998473, + "mean_reconstruction_s": 2.037296655, + "p90_reconstruction_s": 2.427082679999512, + "max_reconstruction_s": 2.9234402000001865, + "median_eval_s": 2.0270746500000314, + "mean_eval_s": 2.1183361649999823, + "p90_eval_s": 2.517494150000221, + "max_eval_s": 3.025518700000248 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 2.7199385522855696e-14, + "max_abs_reconstruction_deviation": 3.1048024151658637e-09 + }, + "iteration_reduction_ratio": 0.752, + "solve_speedup_ratio": 1.1851603041976622, + "eval_speedup_ratio": 1.1336674996155742, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 2.862619100000302, + "eval_s": 2.9510639999998602, + "log_likelihood": 19611.517157894556, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4952321999999185, + "n": 1310, + "solve_s": 0.4952321999999185, + "outer_iterations": 49, + "inner_iterations": 14, + "total_iterations": 63, + "n_passive": 1227, + "warm_start_errors": 59, + "index": 0 + }, + { + "warm_start_error_fraction": 0.029770992366412213, + "reconstruction_s": 2.3411026000003403, + "eval_s": 2.44404779999968, + "log_likelihood": 26288.324526589393, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3163589999994656, + "n": 1310, + "solve_s": 0.3163589999994656, + "outer_iterations": 24, + "inner_iterations": 6, + "total_iterations": 30, + "n_passive": 1237, + "warm_start_errors": 39, + "index": 1 + }, + { + "warm_start_error_fraction": 0.025190839694656488, + "reconstruction_s": 2.059044300000096, + "eval_s": 2.1345634999997856, + "log_likelihood": 27157.168296699077, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3379246000004059, + "n": 1310, + "solve_s": 0.3379246000004059, + "outer_iterations": 26, + "inner_iterations": 10, + "total_iterations": 36, + "n_passive": 1255, + "warm_start_errors": 33, + "index": 2 + }, + { + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 2.234422900000027, + "eval_s": 2.3067614999999932, + "log_likelihood": 19707.61768047339, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.42389819999971223, + "n": 1310, + "solve_s": 0.42389819999971223, + "outer_iterations": 57, + "inner_iterations": 12, + "total_iterations": 69, + "n_passive": 1237, + "warm_start_errors": 71, + "index": 3 + }, + { + "warm_start_error_fraction": 0.03969465648854962, + "reconstruction_s": 2.0464977000001454, + "eval_s": 2.1179603000000498, + "log_likelihood": 20145.39025408475, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.303940099999636, + "n": 1310, + "solve_s": 0.303940099999636, + "outer_iterations": 42, + "inner_iterations": 11, + "total_iterations": 53, + "n_passive": 1237, + "warm_start_errors": 52, + "index": 4 + }, + { + "warm_start_error_fraction": 0.035114503816793895, + "reconstruction_s": 2.2878498000000036, + "eval_s": 2.3643050999999105, + "log_likelihood": 25379.647062825014, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32731560000047466, + "n": 1310, + "solve_s": 0.32731560000047466, + "outer_iterations": 35, + "inner_iterations": 12, + "total_iterations": 47, + "n_passive": 1236, + "warm_start_errors": 46, + "index": 5 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 1.9758100999997623, + "eval_s": 2.0490011000001687, + "log_likelihood": 17165.456938119223, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.45714159999988624, + "n": 1310, + "solve_s": 0.45714159999988624, + "outer_iterations": 49, + "inner_iterations": 15, + "total_iterations": 64, + "n_passive": 1215, + "warm_start_errors": 66, + "index": 6 + }, + { + "warm_start_error_fraction": 0.03740458015267176, + "reconstruction_s": 2.3267269999996643, + "eval_s": 2.402168300000085, + "log_likelihood": 25348.544820021336, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3907854999997653, + "n": 1310, + "solve_s": 0.3907854999997653, + "outer_iterations": 30, + "inner_iterations": 11, + "total_iterations": 41, + "n_passive": 1240, + "warm_start_errors": 49, + "index": 7 + }, + { + "warm_start_error_fraction": 0.038931297709923665, + "reconstruction_s": 2.218318399999589, + "eval_s": 2.2892957999993087, + "log_likelihood": 25011.12829469129, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.370872600000439, + "n": 1310, + "solve_s": 0.370872600000439, + "outer_iterations": 37, + "inner_iterations": 18, + "total_iterations": 55, + "n_passive": 1240, + "warm_start_errors": 51, + "index": 8 + }, + { + "warm_start_error_fraction": 0.042748091603053436, + "reconstruction_s": 2.030986799999482, + "eval_s": 2.109142199999951, + "log_likelihood": 22029.129377472647, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32809699999961595, + "n": 1310, + "solve_s": 0.32809699999961595, + "outer_iterations": 36, + "inner_iterations": 11, + "total_iterations": 47, + "n_passive": 1212, + "warm_start_errors": 56, + "index": 9 + }, + { + "warm_start_error_fraction": 0.031297709923664124, + "reconstruction_s": 2.71431339999981, + "eval_s": 2.834374099999877, + "log_likelihood": 24985.77911837881, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39481359999990673, + "n": 1310, + "solve_s": 0.39481359999990673, + "outer_iterations": 21, + "inner_iterations": 8, + "total_iterations": 29, + "n_passive": 1227, + "warm_start_errors": 41, + "index": 10 + }, + { + "warm_start_error_fraction": 0.02900763358778626, + "reconstruction_s": 2.7117436999997153, + "eval_s": 2.7996180000000095, + "log_likelihood": 24744.164854254584, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41824330000054033, + "n": 1310, + "solve_s": 0.41824330000054033, + "outer_iterations": 33, + "inner_iterations": 14, + "total_iterations": 47, + "n_passive": 1241, + "warm_start_errors": 38, + "index": 11 + }, + { + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 2.082159900000079, + "eval_s": 2.175968599999578, + "log_likelihood": 27138.654656474093, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23598170000059326, + "n": 1310, + "solve_s": 0.23598170000059326, + "outer_iterations": 12, + "inner_iterations": 6, + "total_iterations": 18, + "n_passive": 1242, + "warm_start_errors": 31, + "index": 12 + }, + { + "warm_start_error_fraction": 0.029770992366412213, + "reconstruction_s": 2.4592861000000994, + "eval_s": 2.554855899999893, + "log_likelihood": 28430.846800947846, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2963121000002502, + "n": 1310, + "solve_s": 0.2963121000002502, + "outer_iterations": 22, + "inner_iterations": 10, + "total_iterations": 32, + "n_passive": 1257, + "warm_start_errors": 39, + "index": 13 + }, + { + "warm_start_error_fraction": 0.04351145038167939, + "reconstruction_s": 2.215895699999237, + "eval_s": 2.286062199999833, + "log_likelihood": 22637.75789310233, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34309570000004896, + "n": 1310, + "solve_s": 0.34309570000004896, + "outer_iterations": 35, + "inner_iterations": 11, + "total_iterations": 46, + "n_passive": 1210, + "warm_start_errors": 57, + "index": 14 + }, + { + "warm_start_error_fraction": 0.01984732824427481, + "reconstruction_s": 1.8779684999999517, + "eval_s": 1.986396400000558, + "log_likelihood": 27010.65665798475, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2801620999998704, + "n": 1310, + "solve_s": 0.2801620999998704, + "outer_iterations": 29, + "inner_iterations": 13, + "total_iterations": 42, + "n_passive": 1251, + "warm_start_errors": 26, + "index": 15 + }, + { + "warm_start_error_fraction": 0.03282442748091603, + "reconstruction_s": 2.079521200000272, + "eval_s": 2.1585683999992398, + "log_likelihood": 28769.688442378487, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27906849999999395, + "n": 1310, + "solve_s": 0.27906849999999395, + "outer_iterations": 29, + "inner_iterations": 9, + "total_iterations": 38, + "n_passive": 1241, + "warm_start_errors": 43, + "index": 16 + }, + { + "warm_start_error_fraction": 0.04351145038167939, + "reconstruction_s": 2.4433380999998917, + "eval_s": 2.5263678000001164, + "log_likelihood": 16937.319824448496, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.44048710000060964, + "n": 1310, + "solve_s": 0.44048710000060964, + "outer_iterations": 43, + "inner_iterations": 8, + "total_iterations": 51, + "n_passive": 1216, + "warm_start_errors": 57, + "index": 17 + }, + { + "warm_start_error_fraction": 0.04198473282442748, + "reconstruction_s": 2.2516586000001553, + "eval_s": 2.3276144999999815, + "log_likelihood": 19981.428697564752, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4456128000001627, + "n": 1310, + "solve_s": 0.4456128000001627, + "outer_iterations": 47, + "inner_iterations": 11, + "total_iterations": 58, + "n_passive": 1234, + "warm_start_errors": 55, + "index": 18 + }, + { + "warm_start_error_fraction": 0.0366412213740458, + "reconstruction_s": 2.0563689000000522, + "eval_s": 2.1319236999997884, + "log_likelihood": 20516.940163743897, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3870891999995365, + "n": 1310, + "solve_s": 0.3870891999995365, + "outer_iterations": 40, + "inner_iterations": 12, + "total_iterations": 52, + "n_passive": 1233, + "warm_start_errors": 48, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 2.101958299999751, + "eval_s": 2.1757002999993347, + "log_likelihood": 19611.517157894556, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35947589999977936, + "n": 1310, + "solve_s": 0.35947589999977936, + "outer_iterations": 49, + "inner_iterations": 14, + "total_iterations": 63, + "n_passive": 1227, + "warm_start_errors": 59, + "index": 0 + }, + { + "warm_start_error_fraction": 0.048091603053435114, + "reconstruction_s": 1.942344399999456, + "eval_s": 2.0150014000000738, + "log_likelihood": 26288.32452658934, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2711634999996022, + "n": 1310, + "solve_s": 0.2711634999996022, + "outer_iterations": 47, + "inner_iterations": 10, + "total_iterations": 57, + "n_passive": 1236, + "warm_start_errors": 63, + "index": 1 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 2.1264155999997456, + "eval_s": 2.21422259999963, + "log_likelihood": 27157.168296699092, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3419192999999723, + "n": 1310, + "solve_s": 0.3419192999999723, + "outer_iterations": 61, + "inner_iterations": 17, + "total_iterations": 78, + "n_passive": 1254, + "warm_start_errors": 66, + "index": 2 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 2.1169466000001194, + "eval_s": 2.1837240000004385, + "log_likelihood": 19707.617680473384, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28228800000033516, + "n": 1310, + "solve_s": 0.28228800000033516, + "outer_iterations": 52, + "inner_iterations": 10, + "total_iterations": 62, + "n_passive": 1237, + "warm_start_errors": 65, + "index": 3 + }, + { + "warm_start_error_fraction": 0.03969465648854962, + "reconstruction_s": 1.5964439999997921, + "eval_s": 1.6816892999995616, + "log_likelihood": 20145.390254084807, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.19894269999986136, + "n": 1310, + "solve_s": 0.19894269999986136, + "outer_iterations": 37, + "inner_iterations": 9, + "total_iterations": 46, + "n_passive": 1237, + "warm_start_errors": 52, + "index": 4 + }, + { + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 1.879098100000192, + "eval_s": 1.9627694000000702, + "log_likelihood": 25379.64706282496, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24647839999943244, + "n": 1310, + "solve_s": 0.24647839999943244, + "outer_iterations": 40, + "inner_iterations": 8, + "total_iterations": 48, + "n_passive": 1236, + "warm_start_errors": 59, + "index": 5 + }, + { + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 2.056038400000034, + "eval_s": 2.1250465000002805, + "log_likelihood": 17165.456938119256, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3127997999999934, + "n": 1310, + "solve_s": 0.3127997999999934, + "outer_iterations": 51, + "inner_iterations": 8, + "total_iterations": 59, + "n_passive": 1214, + "warm_start_errors": 80, + "index": 6 + }, + { + "warm_start_error_fraction": 0.06259541984732825, + "reconstruction_s": 1.8809029000003648, + "eval_s": 1.9594028999999864, + "log_likelihood": 25348.54482002133, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3786966999996366, + "n": 1310, + "solve_s": 0.3786966999996366, + "outer_iterations": 77, + "inner_iterations": 15, + "total_iterations": 92, + "n_passive": 1240, + "warm_start_errors": 82, + "index": 7 + }, + { + "warm_start_error_fraction": 0.044274809160305344, + "reconstruction_s": 1.9593886000002385, + "eval_s": 2.039147899999989, + "log_likelihood": 25011.128294691425, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2896236000005956, + "n": 1310, + "solve_s": 0.2896236000005956, + "outer_iterations": 54, + "inner_iterations": 17, + "total_iterations": 71, + "n_passive": 1240, + "warm_start_errors": 58, + "index": 8 + }, + { + "warm_start_error_fraction": 0.06259541984732825, + "reconstruction_s": 1.9301242000001366, + "eval_s": 2.005203700000493, + "log_likelihood": 22029.12937747265, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3185990000001766, + "n": 1310, + "solve_s": 0.3185990000001766, + "outer_iterations": 41, + "inner_iterations": 12, + "total_iterations": 53, + "n_passive": 1212, + "warm_start_errors": 82, + "index": 9 + }, + { + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 1.8445064999996248, + "eval_s": 1.9432584999995015, + "log_likelihood": 24985.779118378603, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.26230690000011236, + "n": 1310, + "solve_s": 0.26230690000011236, + "outer_iterations": 57, + "inner_iterations": 11, + "total_iterations": 68, + "n_passive": 1227, + "warm_start_errors": 79, + "index": 10 + }, + { + "warm_start_error_fraction": 0.0549618320610687, + "reconstruction_s": 2.1612427000000025, + "eval_s": 2.239158500000485, + "log_likelihood": 24744.16485425391, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3534224000004542, + "n": 1310, + "solve_s": 0.3534224000004542, + "outer_iterations": 74, + "inner_iterations": 16, + "total_iterations": 90, + "n_passive": 1241, + "warm_start_errors": 72, + "index": 11 + }, + { + "warm_start_error_fraction": 0.04351145038167939, + "reconstruction_s": 1.7924548999999388, + "eval_s": 1.8791512999996485, + "log_likelihood": 27138.65465647404, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27352309999969293, + "n": 1310, + "solve_s": 0.27352309999969293, + "outer_iterations": 48, + "inner_iterations": 14, + "total_iterations": 62, + "n_passive": 1242, + "warm_start_errors": 57, + "index": 12 + }, + { + "warm_start_error_fraction": 0.038931297709923665, + "reconstruction_s": 1.845757300000514, + "eval_s": 1.931683300000259, + "log_likelihood": 28430.846800947842, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3483389000002717, + "n": 1310, + "solve_s": 0.3483389000002717, + "outer_iterations": 62, + "inner_iterations": 9, + "total_iterations": 71, + "n_passive": 1257, + "warm_start_errors": 51, + "index": 13 + }, + { + "warm_start_error_fraction": 0.06641221374045801, + "reconstruction_s": 1.9301399000005404, + "eval_s": 2.0087581999996473, + "log_likelihood": 22637.757893102385, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23185250000005908, + "n": 1310, + "solve_s": 0.23185250000005908, + "outer_iterations": 38, + "inner_iterations": 9, + "total_iterations": 47, + "n_passive": 1210, + "warm_start_errors": 87, + "index": 14 + }, + { + "warm_start_error_fraction": 0.06870229007633588, + "reconstruction_s": 2.4203990999994858, + "eval_s": 2.5100608000002467, + "log_likelihood": 27010.656657984928, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5517626000000746, + "n": 1310, + "solve_s": 0.5517626000000746, + "outer_iterations": 87, + "inner_iterations": 28, + "total_iterations": 115, + "n_passive": 1250, + "warm_start_errors": 90, + "index": 15 + }, + { + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 2.4872348999997484, + "eval_s": 2.5843942999999854, + "log_likelihood": 28769.68844237843, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37427000000025146, + "n": 1310, + "solve_s": 0.37427000000025146, + "outer_iterations": 51, + "inner_iterations": 20, + "total_iterations": 71, + "n_passive": 1241, + "warm_start_errors": 71, + "index": 16 + }, + { + "warm_start_error_fraction": 0.04961832061068702, + "reconstruction_s": 2.9234402000001865, + "eval_s": 3.025518700000248, + "log_likelihood": 16937.31982444849, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.37176369999997405, + "n": 1310, + "solve_s": 0.37176369999997405, + "outer_iterations": 41, + "inner_iterations": 14, + "total_iterations": 55, + "n_passive": 1216, + "warm_start_errors": 65, + "index": 17 + }, + { + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 2.1288554000002478, + "eval_s": 2.1962954000000536, + "log_likelihood": 19981.42869756473, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25205780000032973, + "n": 1310, + "solve_s": 0.25205780000032973, + "outer_iterations": 64, + "inner_iterations": 23, + "total_iterations": 87, + "n_passive": 1234, + "warm_start_errors": 66, + "index": 18 + }, + { + "warm_start_error_fraction": 0.038931297709923665, + "reconstruction_s": 1.6222410999998829, + "eval_s": 1.686536299999716, + "log_likelihood": 20516.940163743748, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.12961059999997815, + "n": 1310, + "solve_s": 0.12961059999997815, + "outer_iterations": 30, + "inner_iterations": 6, + "total_iterations": 36, + "n_passive": 1233, + "warm_start_errors": 51, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_powerlaw_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_powerlaw_v2026.8.17.1.png new file mode 100644 index 000000000..f7da97fb3 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_powerlaw_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_rectangular_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_rectangular_v2026.8.17.1.json new file mode 100644 index 000000000..bd2b4f4e1 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_rectangular_v2026.8.17.1.json @@ -0,0 +1,1601 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "hst", + "model": "rectangular", + "configuration": { + "model": "rectangular", + "dataset": "hst", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.05, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 15361, + "over_sampled_pixels": 17980, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "rectangular_uniform_35x35", + "regularization": "constant", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 7.7079187999997885, + "total_memo_seed_retries": 0, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.6984334203655352, + "mean_warm_start_error_fraction": 0.6972149695387293, + "p90_warm_start_error_fraction": 0.703307223672759, + "max_warm_start_error_fraction": 0.7075718015665796, + "median_outer_iterations": 860.5, + "mean_outer_iterations": 857.5, + "p90_outer_iterations": 868.1, + "max_outer_iterations": 876.0, + "median_inner_iterations": 22.0, + "mean_inner_iterations": 29.75, + "p90_inner_iterations": 50.2, + "max_inner_iterations": 54.0, + "median_total_iterations": 885.5, + "mean_total_iterations": 887.25, + "p90_total_iterations": 897.2, + "max_total_iterations": 906.0, + "median_n_passive": 942.0, + "mean_n_passive": 941.85, + "p90_n_passive": 945.0, + "max_n_passive": 952.0, + "median_warm_start_errors": 802.5, + "mean_warm_start_errors": 801.1, + "p90_warm_start_errors": 808.1, + "max_warm_start_errors": 813.0, + "median_solve_s": 1.2124096499996995, + "mean_solve_s": 1.2170695099999194, + "p90_solve_s": 1.4840982300004726, + "max_solve_s": 1.6190808000001198, + "median_reconstruction_s": 3.0725932499999544, + "mean_reconstruction_s": 3.120451034999951, + "p90_reconstruction_s": 3.538171939999757, + "max_reconstruction_s": 3.6955441999998584, + "median_eval_s": 3.1703301499996996, + "mean_eval_s": 3.2118285949999974, + "p90_eval_s": 3.6502051599997687, + "max_eval_s": 3.803729000000203 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.011314186248912098, + "mean_warm_start_error_fraction": 0.044604003481288075, + "p90_warm_start_error_fraction": 0.021235857267188865, + "max_warm_start_error_fraction": 0.6814621409921671, + "median_outer_iterations": 6.0, + "mean_outer_iterations": 48.5, + "p90_outer_iterations": 15.200000000000003, + "max_outer_iterations": 833.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 3.6, + "p90_inner_iterations": 2.1000000000000014, + "max_inner_iterations": 48.0, + "median_total_iterations": 7.5, + "mean_total_iterations": 52.1, + "p90_total_iterations": 17.1, + "max_total_iterations": 881.0, + "median_n_passive": 942.0, + "mean_n_passive": 941.65, + "p90_n_passive": 944.1, + "max_n_passive": 952.0, + "median_warm_start_errors": 13.0, + "mean_warm_start_errors": 51.25, + "p90_warm_start_errors": 24.400000000000006, + "max_warm_start_errors": 783.0, + "median_solve_s": 0.04032645000006596, + "mean_solve_s": 0.09226416500014238, + "p90_solve_s": 0.09045156000038335, + "max_solve_s": 0.9052021999996214, + "median_reconstruction_s": 1.8040333999997529, + "mean_reconstruction_s": 1.9276211250001325, + "p90_reconstruction_s": 2.2811350200005114, + "max_reconstruction_s": 2.9006846000002042, + "median_eval_s": 1.8875243500001488, + "mean_eval_s": 2.009634369999958, + "p90_eval_s": 2.428005609999582, + "max_eval_s": 2.9695162999996683 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 2.871015492144613e-15, + "max_abs_reconstruction_deviation": 3.036877416207062e-09 + }, + "iteration_reduction_ratio": 118.06666666666666, + "solve_speedup_ratio": 30.064874294606057, + "eval_speedup_ratio": 1.6796234443277247, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.6814621409921671, + "reconstruction_s": 3.520744499999637, + "eval_s": 3.6288971000003585, + "log_likelihood": 26586.6726590554, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.5037931999995635, + "n": 1149, + "solve_s": 1.5037931999995635, + "outer_iterations": 833, + "inner_iterations": 48, + "total_iterations": 881, + "n_passive": 936, + "warm_start_errors": 783, + "index": 0 + }, + { + "warm_start_error_fraction": 0.6988685813751088, + "reconstruction_s": 2.7429645000001983, + "eval_s": 2.827743499999997, + "log_likelihood": 26442.5539852937, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.947303699999793, + "n": 1149, + "solve_s": 0.947303699999793, + "outer_iterations": 869, + "inner_iterations": 19, + "total_iterations": 888, + "n_passive": 943, + "warm_start_errors": 803, + "index": 1 + }, + { + "warm_start_error_fraction": 0.6971279373368147, + "reconstruction_s": 2.7994971000007354, + "eval_s": 2.869189600000027, + "log_likelihood": 26281.70948089614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.1532514000000447, + "n": 1149, + "solve_s": 1.1532514000000447, + "outer_iterations": 849, + "inner_iterations": 50, + "total_iterations": 899, + "n_passive": 944, + "warm_start_errors": 801, + "index": 2 + }, + { + "warm_start_error_fraction": 0.6971279373368147, + "reconstruction_s": 3.0957504999996672, + "eval_s": 3.2000147999997353, + "log_likelihood": 26466.214397577514, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2421018999993976, + "n": 1149, + "solve_s": 1.2421018999993976, + "outer_iterations": 864, + "inner_iterations": 17, + "total_iterations": 881, + "n_passive": 942, + "warm_start_errors": 801, + "index": 3 + }, + { + "warm_start_error_fraction": 0.6962576153176675, + "reconstruction_s": 3.3726587999999538, + "eval_s": 3.459068699999989, + "log_likelihood": 26476.421197245312, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.299070699999902, + "n": 1149, + "solve_s": 1.299070699999902, + "outer_iterations": 861, + "inner_iterations": 19, + "total_iterations": 880, + "n_passive": 942, + "warm_start_errors": 800, + "index": 4 + }, + { + "warm_start_error_fraction": 0.6988685813751088, + "reconstruction_s": 3.695018900000832, + "eval_s": 3.803729000000203, + "log_likelihood": 26283.81284464895, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4819099000005735, + "n": 1149, + "solve_s": 1.4819099000005735, + "outer_iterations": 846, + "inner_iterations": 50, + "total_iterations": 896, + "n_passive": 939, + "warm_start_errors": 803, + "index": 5 + }, + { + "warm_start_error_fraction": 0.6884247171453438, + "reconstruction_s": 3.261168400000315, + "eval_s": 3.330897499999992, + "log_likelihood": 26559.931391644204, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.1080246000001353, + "n": 1149, + "solve_s": 1.1080246000001353, + "outer_iterations": 836, + "inner_iterations": 46, + "total_iterations": 882, + "n_passive": 940, + "warm_start_errors": 791, + "index": 6 + }, + { + "warm_start_error_fraction": 0.6979982593559617, + "reconstruction_s": 3.371270999999979, + "eval_s": 3.484079400000155, + "log_likelihood": 26609.941728965714, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.6190808000001198, + "n": 1149, + "solve_s": 1.6190808000001198, + "outer_iterations": 854, + "inner_iterations": 24, + "total_iterations": 878, + "n_passive": 940, + "warm_start_errors": 802, + "index": 7 + }, + { + "warm_start_error_fraction": 0.7023498694516971, + "reconstruction_s": 3.6955441999998584, + "eval_s": 3.772889199999554, + "log_likelihood": 26705.41235068809, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.212084099999629, + "n": 1149, + "solve_s": 1.212084099999629, + "outer_iterations": 858, + "inner_iterations": 39, + "total_iterations": 897, + "n_passive": 942, + "warm_start_errors": 807, + "index": 8 + }, + { + "warm_start_error_fraction": 0.6953872932985204, + "reconstruction_s": 3.335457399999541, + "eval_s": 3.4276848999998037, + "log_likelihood": 26319.485902028813, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.37749199999962, + "n": 1149, + "solve_s": 1.37749199999962, + "outer_iterations": 852, + "inner_iterations": 20, + "total_iterations": 872, + "n_passive": 935, + "warm_start_errors": 799, + "index": 9 + }, + { + "warm_start_error_fraction": 0.6988685813751088, + "reconstruction_s": 3.0494360000002416, + "eval_s": 3.140645499999664, + "log_likelihood": 26253.727585132234, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3194565000003422, + "n": 1149, + "solve_s": 1.3194565000003422, + "outer_iterations": 852, + "inner_iterations": 54, + "total_iterations": 906, + "n_passive": 942, + "warm_start_errors": 803, + "index": 10 + }, + { + "warm_start_error_fraction": 0.7032201914708442, + "reconstruction_s": 2.7145763999997143, + "eval_s": 2.8083486000004996, + "log_likelihood": 26172.241099616447, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.9431651999993846, + "n": 1149, + "solve_s": 0.9431651999993846, + "outer_iterations": 863, + "inner_iterations": 14, + "total_iterations": 877, + "n_passive": 943, + "warm_start_errors": 808, + "index": 11 + }, + { + "warm_start_error_fraction": 0.7040905134899913, + "reconstruction_s": 2.7716663000001063, + "eval_s": 2.8769185000001016, + "log_likelihood": 26125.28464573013, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.252798299999995, + "n": 1149, + "solve_s": 1.252798299999995, + "outer_iterations": 868, + "inner_iterations": 14, + "total_iterations": 882, + "n_passive": 945, + "warm_start_errors": 809, + "index": 12 + }, + { + "warm_start_error_fraction": 0.7032201914708442, + "reconstruction_s": 2.959957399999439, + "eval_s": 3.0384070999998585, + "log_likelihood": 26173.95735526784, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0616484000001947, + "n": 1149, + "solve_s": 1.0616484000001947, + "outer_iterations": 868, + "inner_iterations": 20, + "total_iterations": 888, + "n_passive": 940, + "warm_start_errors": 808, + "index": 13 + }, + { + "warm_start_error_fraction": 0.6892950391644909, + "reconstruction_s": 2.766369599999962, + "eval_s": 2.8648853000004237, + "log_likelihood": 26096.2558229986, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0249244999995426, + "n": 1149, + "solve_s": 1.0249244999995426, + "outer_iterations": 845, + "inner_iterations": 52, + "total_iterations": 897, + "n_passive": 942, + "warm_start_errors": 792, + "index": 14 + }, + { + "warm_start_error_fraction": 0.6866840731070496, + "reconstruction_s": 2.865574200000083, + "eval_s": 2.9471886000001177, + "log_likelihood": 26149.714184353157, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.186569800000143, + "n": 1149, + "solve_s": 1.186569800000143, + "outer_iterations": 861, + "inner_iterations": 20, + "total_iterations": 881, + "n_passive": 943, + "warm_start_errors": 789, + "index": 15 + }, + { + "warm_start_error_fraction": 0.6979982593559617, + "reconstruction_s": 2.860731699999633, + "eval_s": 2.9334175000003597, + "log_likelihood": 26286.19982257089, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0301724000000831, + "n": 1149, + "solve_s": 1.0301724000000831, + "outer_iterations": 868, + "inner_iterations": 20, + "total_iterations": 888, + "n_passive": 937, + "warm_start_errors": 802, + "index": 16 + }, + { + "warm_start_error_fraction": 0.6988685813751088, + "reconstruction_s": 2.7146318999994037, + "eval_s": 2.8019367999995666, + "log_likelihood": 26637.45980405758, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.052050500000405, + "n": 1149, + "solve_s": 1.052050500000405, + "outer_iterations": 860, + "inner_iterations": 23, + "total_iterations": 883, + "n_passive": 945, + "warm_start_errors": 803, + "index": 17 + }, + { + "warm_start_error_fraction": 0.700609225413403, + "reconstruction_s": 3.2968392000002495, + "eval_s": 3.3840566999997463, + "log_likelihood": 26773.283233815608, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.21273519999977, + "n": 1149, + "solve_s": 1.21273519999977, + "outer_iterations": 867, + "inner_iterations": 25, + "total_iterations": 892, + "n_passive": 945, + "warm_start_errors": 805, + "index": 18 + }, + { + "warm_start_error_fraction": 0.7075718015665796, + "reconstruction_s": 3.5191626999994696, + "eval_s": 3.6365735999997923, + "log_likelihood": 27215.797874499214, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3137570999997479, + "n": 1149, + "solve_s": 1.3137570999997479, + "outer_iterations": 876, + "inner_iterations": 21, + "total_iterations": 897, + "n_passive": 952, + "warm_start_errors": 813, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.6814621409921671, + "reconstruction_s": 2.9006846000002042, + "eval_s": 2.9695162999996683, + "log_likelihood": 26586.6726590554, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.9052021999996214, + "n": 1149, + "solve_s": 0.9052021999996214, + "outer_iterations": 833, + "inner_iterations": 48, + "total_iterations": 881, + "n_passive": 936, + "warm_start_errors": 783, + "index": 0 + }, + { + "warm_start_error_fraction": 0.011314186248912098, + "reconstruction_s": 1.7736594999996669, + "eval_s": 1.8713754000000336, + "log_likelihood": 26442.55398529364, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08817720000024565, + "n": 1149, + "solve_s": 0.08817720000024565, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 943, + "warm_start_errors": 13, + "index": 1 + }, + { + "warm_start_error_fraction": 0.007832898172323759, + "reconstruction_s": 1.6801415999998426, + "eval_s": 1.7418416999998954, + "log_likelihood": 26281.709480896145, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03322110000044631, + "n": 1149, + "solve_s": 0.03322110000044631, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 944, + "warm_start_errors": 9, + "index": 2 + }, + { + "warm_start_error_fraction": 0.008703220191470844, + "reconstruction_s": 1.6497617000004539, + "eval_s": 1.73026260000006, + "log_likelihood": 26466.214397577536, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03542410000045493, + "n": 1149, + "solve_s": 0.03542410000045493, + "outer_iterations": 4, + "inner_iterations": 2, + "total_iterations": 6, + "n_passive": 942, + "warm_start_errors": 10, + "index": 3 + }, + { + "warm_start_error_fraction": 0.0026109660574412533, + "reconstruction_s": 1.7349186000001282, + "eval_s": 1.824351399999614, + "log_likelihood": 26476.421197245327, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03083520000018325, + "n": 1149, + "solve_s": 0.03083520000018325, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 941, + "warm_start_errors": 3, + "index": 4 + }, + { + "warm_start_error_fraction": 0.016536118363794605, + "reconstruction_s": 2.014072099999794, + "eval_s": 2.0926915999998528, + "log_likelihood": 26283.812844648954, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0576483000004373, + "n": 1149, + "solve_s": 0.0576483000004373, + "outer_iterations": 9, + "inner_iterations": 2, + "total_iterations": 11, + "n_passive": 938, + "warm_start_errors": 19, + "index": 5 + }, + { + "warm_start_error_fraction": 0.01392515230635335, + "reconstruction_s": 1.9788367000001017, + "eval_s": 2.0441859000002296, + "log_likelihood": 26559.931391644197, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.045142500000110886, + "n": 1149, + "solve_s": 0.045142500000110886, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 940, + "warm_start_errors": 16, + "index": 6 + }, + { + "warm_start_error_fraction": 0.024369016536118365, + "reconstruction_s": 1.6252689000002647, + "eval_s": 1.692067500000121, + "log_likelihood": 26609.941728965638, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05812500000047294, + "n": 1149, + "solve_s": 0.05812500000047294, + "outer_iterations": 14, + "inner_iterations": 3, + "total_iterations": 17, + "n_passive": 940, + "warm_start_errors": 28, + "index": 7 + }, + { + "warm_start_error_fraction": 0.006962576153176675, + "reconstruction_s": 1.7501556999995955, + "eval_s": 1.8162771000006614, + "log_likelihood": 26705.412350688093, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03512259999934031, + "n": 1149, + "solve_s": 0.03512259999934031, + "outer_iterations": 5, + "inner_iterations": 2, + "total_iterations": 7, + "n_passive": 942, + "warm_start_errors": 8, + "index": 8 + }, + { + "warm_start_error_fraction": 0.011314186248912098, + "reconstruction_s": 1.9665999000008014, + "eval_s": 2.0497120999998515, + "log_likelihood": 26319.485902028828, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03408320000016829, + "n": 1149, + "solve_s": 0.03408320000016829, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 935, + "warm_start_errors": 13, + "index": 9 + }, + { + "warm_start_error_fraction": 0.013054830287206266, + "reconstruction_s": 2.2261545000001206, + "eval_s": 2.399957099999483, + "log_likelihood": 26253.727585132234, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10345170000073267, + "n": 1149, + "solve_s": 0.10345170000073267, + "outer_iterations": 12, + "inner_iterations": 2, + "total_iterations": 14, + "n_passive": 942, + "warm_start_errors": 15, + "index": 10 + }, + { + "warm_start_error_fraction": 0.004351610095735422, + "reconstruction_s": 2.153035700000146, + "eval_s": 2.2450316999993447, + "log_likelihood": 26172.241099616436, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03747459999976854, + "n": 1149, + "solve_s": 0.03747459999976854, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 943, + "warm_start_errors": 5, + "index": 11 + }, + { + "warm_start_error_fraction": 0.0008703220191470844, + "reconstruction_s": 1.609593600000153, + "eval_s": 1.6858702999998059, + "log_likelihood": 26125.284645730128, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.02792600000066159, + "n": 1149, + "solve_s": 0.02792600000066159, + "outer_iterations": 1, + "inner_iterations": 0, + "total_iterations": 1, + "n_passive": 944, + "warm_start_errors": 1, + "index": 12 + }, + { + "warm_start_error_fraction": 0.007832898172323759, + "reconstruction_s": 1.5965238000007957, + "eval_s": 1.6645217000004777, + "log_likelihood": 26173.957355267798, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0298384999996415, + "n": 1149, + "solve_s": 0.0298384999996415, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 939, + "warm_start_errors": 9, + "index": 13 + }, + { + "warm_start_error_fraction": 0.013054830287206266, + "reconstruction_s": 1.7227340000008553, + "eval_s": 1.7908576999998331, + "log_likelihood": 26096.255822998573, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.051238899999589194, + "n": 1149, + "solve_s": 0.051238899999589194, + "outer_iterations": 10, + "inner_iterations": 1, + "total_iterations": 11, + "n_passive": 942, + "warm_start_errors": 15, + "index": 14 + }, + { + "warm_start_error_fraction": 0.005221932114882507, + "reconstruction_s": 1.6548780000002807, + "eval_s": 1.7331835999993928, + "log_likelihood": 26149.714184353157, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.037044499999865366, + "n": 1149, + "solve_s": 0.037044499999865366, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 944, + "warm_start_errors": 6, + "index": 15 + }, + { + "warm_start_error_fraction": 0.018276762402088774, + "reconstruction_s": 1.8344072999998389, + "eval_s": 1.9036733000002641, + "log_likelihood": 26286.199822570896, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04198710000036954, + "n": 1149, + "solve_s": 0.04198710000036954, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 937, + "warm_start_errors": 21, + "index": 16 + }, + { + "warm_start_error_fraction": 0.020887728459530026, + "reconstruction_s": 2.2462810000006357, + "eval_s": 2.3399090000002616, + "log_likelihood": 26637.459804057577, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08900710000034451, + "n": 1149, + "solve_s": 0.08900710000034451, + "outer_iterations": 17, + "inner_iterations": 1, + "total_iterations": 18, + "n_passive": 945, + "warm_start_errors": 24, + "index": 17 + }, + { + "warm_start_error_fraction": 0.006092254134029591, + "reconstruction_s": 2.5948211999993873, + "eval_s": 2.6804422000004706, + "log_likelihood": 26773.283233815622, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.038665799999762385, + "n": 1149, + "solve_s": 0.038665799999762385, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 944, + "warm_start_errors": 7, + "index": 18 + }, + { + "warm_start_error_fraction": 0.017406440382941687, + "reconstruction_s": 1.839894099999583, + "eval_s": 1.9169591999998374, + "log_likelihood": 27215.797874499214, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06566770000063116, + "n": 1149, + "solve_s": 0.06566770000063116, + "outer_iterations": 15, + "inner_iterations": 1, + "total_iterations": 16, + "n_passive": 952, + "warm_start_errors": 20, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.6640557006092254, + "mean_warm_start_error_fraction": 0.6041340295909485, + "p90_warm_start_error_fraction": 0.7182767624020888, + "max_warm_start_error_fraction": 0.741514360313316, + "median_outer_iterations": 840.0, + "mean_outer_iterations": 763.9, + "p90_outer_iterations": 900.1, + "max_outer_iterations": 901.0, + "median_inner_iterations": 25.5, + "mean_inner_iterations": 33.25, + "p90_inner_iterations": 58.200000000000045, + "max_inner_iterations": 92.0, + "median_total_iterations": 872.5, + "mean_total_iterations": 797.15, + "p90_total_iterations": 919.1, + "max_total_iterations": 993.0, + "median_n_passive": 942.0, + "mean_n_passive": 941.75, + "p90_n_passive": 959.1, + "max_n_passive": 970.0, + "median_warm_start_errors": 763.0, + "mean_warm_start_errors": 694.15, + "p90_warm_start_errors": 825.3, + "max_warm_start_errors": 852.0, + "median_solve_s": 1.0896490499999345, + "mean_solve_s": 1.0924774300000535, + "p90_solve_s": 1.3426963299997625, + "max_solve_s": 1.4625286999998934, + "median_reconstruction_s": 3.025467650000337, + "mean_reconstruction_s": 2.9962168699999894, + "p90_reconstruction_s": 3.42077739000024, + "max_reconstruction_s": 3.6150117999995928, + "median_eval_s": 3.098132650000025, + "mean_eval_s": 3.0796476899999563, + "p90_eval_s": 3.5038055899999567, + "max_eval_s": 3.7004257999997208 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "median_warm_start_error_fraction": 0.08616187989556136, + "mean_warm_start_error_fraction": 0.10922541340295908, + "p90_warm_start_error_fraction": 0.11122715404699744, + "max_warm_start_error_fraction": 0.6370757180156658, + "median_outer_iterations": 139.5, + "mean_outer_iterations": 191.05, + "p90_outer_iterations": 289.1000000000002, + "max_outer_iterations": 799.0, + "median_inner_iterations": 15.0, + "mean_inner_iterations": 40.2, + "p90_inner_iterations": 70.00000000000021, + "max_inner_iterations": 234.0, + "median_total_iterations": 171.5, + "mean_total_iterations": 231.25, + "p90_total_iterations": 458.70000000000005, + "max_total_iterations": 854.0, + "median_n_passive": 942.0, + "mean_n_passive": 941.5, + "p90_n_passive": 958.2, + "max_n_passive": 969.0, + "median_warm_start_errors": 99.0, + "mean_warm_start_errors": 125.5, + "p90_warm_start_errors": 127.80000000000005, + "max_warm_start_errors": 732.0, + "median_solve_s": 0.32652390000021114, + "mean_solve_s": 0.4475663999999597, + "p90_solve_s": 0.7376749599998222, + "max_solve_s": 1.3070234999995591, + "median_reconstruction_s": 1.961074550000376, + "mean_reconstruction_s": 2.1438345300001855, + "p90_reconstruction_s": 2.550950790000298, + "max_reconstruction_s": 3.5893910000004325, + "median_eval_s": 2.030718500000148, + "mean_eval_s": 2.222532784999976, + "p90_eval_s": 2.6288225400002383, + "max_eval_s": 3.669363699999849 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 9.875127171974489e-15, + "max_abs_reconstruction_deviation": 3.070110596303799e-09 + }, + "iteration_reduction_ratio": 5.087463556851312, + "solve_speedup_ratio": 3.3371188142712676, + "eval_speedup_ratio": 1.5256337350547597, + "rows": { + "off": [ + { + "warm_start_error_fraction": 0.6370757180156658, + "reconstruction_s": 3.578960400000142, + "eval_s": 3.65947849999975, + "log_likelihood": 17485.50796713436, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2082608000000619, + "n": 1149, + "solve_s": 1.2082608000000619, + "outer_iterations": 799, + "inner_iterations": 55, + "total_iterations": 854, + "n_passive": 945, + "warm_start_errors": 732, + "index": 0 + }, + { + "warm_start_error_fraction": 0.7180156657963447, + "reconstruction_s": 3.025677600000563, + "eval_s": 3.0988328000003094, + "log_likelihood": 21615.441835135316, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.9861988000002384, + "n": 1149, + "solve_s": 0.9861988000002384, + "outer_iterations": 900, + "inner_iterations": 20, + "total_iterations": 920, + "n_passive": 959, + "warm_start_errors": 825, + "index": 1 + }, + { + "warm_start_error_fraction": 0.720626631853786, + "reconstruction_s": 2.973436899999797, + "eval_s": 3.059610399999656, + "log_likelihood": 25591.769261275505, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.2287762999994811, + "n": 1149, + "solve_s": 1.2287762999994811, + "outer_iterations": 857, + "inner_iterations": 22, + "total_iterations": 879, + "n_passive": 916, + "warm_start_errors": 828, + "index": 2 + }, + { + "warm_start_error_fraction": 0.6057441253263708, + "reconstruction_s": 3.6150117999995928, + "eval_s": 3.7004257999997208, + "log_likelihood": 23407.5479301935, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.4625286999998934, + "n": 1149, + "solve_s": 1.4625286999998934, + "outer_iterations": 792, + "inner_iterations": 87, + "total_iterations": 879, + "n_passive": 957, + "warm_start_errors": 696, + "index": 3 + }, + { + "warm_start_error_fraction": 0.6431679721496953, + "reconstruction_s": 3.3197689999997237, + "eval_s": 3.476454200000262, + "log_likelihood": 25051.07575569312, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3389208999997209, + "n": 1149, + "solve_s": 1.3389208999997209, + "outer_iterations": 806, + "inner_iterations": 22, + "total_iterations": 828, + "n_passive": 939, + "warm_start_errors": 739, + "index": 4 + }, + { + "warm_start_error_fraction": 0.6875543951261966, + "reconstruction_s": 3.347220699999525, + "eval_s": 3.424292499999865, + "log_likelihood": 24825.281072409372, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0970552000007956, + "n": 1149, + "solve_s": 1.0970552000007956, + "outer_iterations": 848, + "inner_iterations": 47, + "total_iterations": 895, + "n_passive": 935, + "warm_start_errors": 790, + "index": 5 + }, + { + "warm_start_error_fraction": 0.7162750217580505, + "reconstruction_s": 2.695093999999699, + "eval_s": 2.763699600000109, + "log_likelihood": 22356.591295712056, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0389463999999862, + "n": 1149, + "solve_s": 1.0389463999999862, + "outer_iterations": 901, + "inner_iterations": 18, + "total_iterations": 919, + "n_passive": 951, + "warm_start_errors": 823, + "index": 6 + }, + { + "warm_start_error_fraction": 0.660574412532637, + "reconstruction_s": 2.598107699999673, + "eval_s": 2.6744106999994983, + "log_likelihood": 21401.45162782251, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.9699830999998085, + "n": 1149, + "solve_s": 0.9699830999998085, + "outer_iterations": 788, + "inner_iterations": 23, + "total_iterations": 811, + "n_passive": 902, + "warm_start_errors": 759, + "index": 7 + }, + { + "warm_start_error_fraction": 0.6892950391644909, + "reconstruction_s": 3.025257700000111, + "eval_s": 3.10935150000023, + "log_likelihood": 21654.932007862913, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.1167421000000104, + "n": 1149, + "solve_s": 1.1167421000000104, + "outer_iterations": 846, + "inner_iterations": 30, + "total_iterations": 876, + "n_passive": 921, + "warm_start_errors": 792, + "index": 8 + }, + { + "warm_start_error_fraction": 0.741514360313316, + "reconstruction_s": 2.59972420000031, + "eval_s": 2.673071700000037, + "log_likelihood": 15446.649038322048, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.9979735000006258, + "n": 1149, + "solve_s": 0.9979735000006258, + "outer_iterations": 883, + "inner_iterations": 29, + "total_iterations": 912, + "n_passive": 942, + "warm_start_errors": 852, + "index": 9 + }, + { + "warm_start_error_fraction": 0.6501305483028721, + "reconstruction_s": 2.659315699999752, + "eval_s": 2.745274199999585, + "log_likelihood": 24313.00807431178, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8653031000003466, + "n": 1149, + "solve_s": 0.8653031000003466, + "outer_iterations": 841, + "inner_iterations": 28, + "total_iterations": 869, + "n_passive": 926, + "warm_start_errors": 747, + "index": 10 + }, + { + "warm_start_error_fraction": 0.6840731070496083, + "reconstruction_s": 2.685583299999962, + "eval_s": 2.7729980000003707, + "log_likelihood": 18450.248452275675, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0876044000006004, + "n": 1149, + "solve_s": 1.0876044000006004, + "outer_iterations": 901, + "inner_iterations": 92, + "total_iterations": 993, + "n_passive": 946, + "warm_start_errors": 786, + "index": 11 + }, + { + "warm_start_error_fraction": 0.6640557006092254, + "reconstruction_s": 3.2948005999996894, + "eval_s": 3.3785364000004847, + "log_likelihood": 20520.077188526822, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3249738000004072, + "n": 1149, + "solve_s": 1.3249738000004072, + "outer_iterations": 816, + "inner_iterations": 26, + "total_iterations": 842, + "n_passive": 941, + "warm_start_errors": 763, + "index": 12 + }, + { + "warm_start_error_fraction": 0.17493472584856398, + "reconstruction_s": 3.4032015000002502, + "eval_s": 3.486508599999979, + "log_likelihood": 23692.010687028232, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.376675200000136, + "n": 1149, + "solve_s": 1.376675200000136, + "outer_iterations": 476, + "inner_iterations": 32, + "total_iterations": 508, + "n_passive": 960, + "warm_start_errors": 201, + "index": 13 + }, + { + "warm_start_error_fraction": 0.22019147084421237, + "reconstruction_s": 3.2496623000006366, + "eval_s": 3.327976900000067, + "log_likelihood": 25380.785427249546, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0133077000000412, + "n": 1149, + "solve_s": 1.0133077000000412, + "outer_iterations": 300, + "inner_iterations": 16, + "total_iterations": 316, + "n_passive": 958, + "warm_start_errors": 253, + "index": 14 + }, + { + "warm_start_error_fraction": 0.1409921671018277, + "reconstruction_s": 2.4831230999998297, + "eval_s": 2.5624474999995073, + "log_likelihood": 23839.966868313422, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4985583999996379, + "n": 1149, + "solve_s": 0.4985583999996379, + "outer_iterations": 131, + "inner_iterations": 9, + "total_iterations": 140, + "n_passive": 970, + "warm_start_errors": 162, + "index": 15 + }, + { + "warm_start_error_fraction": 0.6640557006092254, + "reconstruction_s": 3.1325282000007064, + "eval_s": 3.2035292999999, + "log_likelihood": 25432.2575657524, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.0916936999992686, + "n": 1149, + "solve_s": 1.0916936999992686, + "outer_iterations": 839, + "inner_iterations": 23, + "total_iterations": 862, + "n_passive": 931, + "warm_start_errors": 763, + "index": 16 + }, + { + "warm_start_error_fraction": 0.7110530896431679, + "reconstruction_s": 2.6766168000003745, + "eval_s": 2.7750851999999213, + "log_likelihood": 27302.229027713605, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.9997489999996105, + "n": 1149, + "solve_s": 0.9997489999996105, + "outer_iterations": 853, + "inner_iterations": 25, + "total_iterations": 878, + "n_passive": 942, + "warm_start_errors": 817, + "index": 17 + }, + { + "warm_start_error_fraction": 0.6910356832027851, + "reconstruction_s": 3.025920000000042, + "eval_s": 3.0974324999997407, + "log_likelihood": 22914.052021782307, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.1973248000003878, + "n": 1149, + "solve_s": 1.1973248000003878, + "outer_iterations": 866, + "inner_iterations": 37, + "total_iterations": 903, + "n_passive": 938, + "warm_start_errors": 794, + "index": 18 + }, + { + "warm_start_error_fraction": 0.6623150565709313, + "reconstruction_s": 2.535325899999407, + "eval_s": 2.6035375000001295, + "log_likelihood": 20254.667765403323, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.948972700000013, + "n": 1149, + "solve_s": 0.948972700000013, + "outer_iterations": 835, + "inner_iterations": 24, + "total_iterations": 859, + "n_passive": 956, + "warm_start_errors": 761, + "index": 19 + } + ], + "on": [ + { + "warm_start_error_fraction": 0.6370757180156658, + "reconstruction_s": 3.5893910000004325, + "eval_s": 3.669363699999849, + "log_likelihood": 17485.50796713436, + "n_solves": 1, + "retries": 0, + "solve_s_total": 1.3070234999995591, + "n": 1149, + "solve_s": 1.3070234999995591, + "outer_iterations": 799, + "inner_iterations": 55, + "total_iterations": 854, + "n_passive": 945, + "warm_start_errors": 732, + "index": 0 + }, + { + "warm_start_error_fraction": 0.072236727589208, + "reconstruction_s": 1.8842515999995157, + "eval_s": 1.9563852999999654, + "log_likelihood": 21615.44183513532, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24575000000004366, + "n": 1149, + "solve_s": 0.24575000000004366, + "outer_iterations": 82, + "inner_iterations": 17, + "total_iterations": 99, + "n_passive": 958, + "warm_start_errors": 83, + "index": 1 + }, + { + "warm_start_error_fraction": 0.09399477806788512, + "reconstruction_s": 2.3621679000007134, + "eval_s": 2.4615169999997306, + "log_likelihood": 25591.76926127549, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5409538000003522, + "n": 1149, + "solve_s": 0.5409538000003522, + "outer_iterations": 176, + "inner_iterations": 13, + "total_iterations": 189, + "n_passive": 916, + "warm_start_errors": 108, + "index": 2 + }, + { + "warm_start_error_fraction": 0.10356832027850305, + "reconstruction_s": 2.4598359999999957, + "eval_s": 2.5362451999999394, + "log_likelihood": 23407.54793019364, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5765602999999828, + "n": 1149, + "solve_s": 0.5765602999999828, + "outer_iterations": 251, + "inner_iterations": 18, + "total_iterations": 269, + "n_passive": 957, + "warm_start_errors": 119, + "index": 3 + }, + { + "warm_start_error_fraction": 0.08181026979982593, + "reconstruction_s": 1.890073599999596, + "eval_s": 1.9647040999998353, + "log_likelihood": 25051.075755692873, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3020072999997865, + "n": 1149, + "solve_s": 0.3020072999997865, + "outer_iterations": 113, + "inner_iterations": 11, + "total_iterations": 124, + "n_passive": 939, + "warm_start_errors": 94, + "index": 4 + }, + { + "warm_start_error_fraction": 0.09225413402959094, + "reconstruction_s": 2.2248174000005747, + "eval_s": 2.345101899999463, + "log_likelihood": 24825.28107240938, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.673157799999899, + "n": 1149, + "solve_s": 0.673157799999899, + "outer_iterations": 276, + "inner_iterations": 36, + "total_iterations": 312, + "n_passive": 935, + "warm_start_errors": 106, + "index": 5 + }, + { + "warm_start_error_fraction": 0.0557006092254134, + "reconstruction_s": 1.912907100000666, + "eval_s": 2.0025888999998642, + "log_likelihood": 22356.59129571208, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25005010000040784, + "n": 1149, + "solve_s": 0.25005010000040784, + "outer_iterations": 108, + "inner_iterations": 8, + "total_iterations": 116, + "n_passive": 951, + "warm_start_errors": 64, + "index": 6 + }, + { + "warm_start_error_fraction": 0.10704960835509138, + "reconstruction_s": 1.8931059000005916, + "eval_s": 1.9626570999998876, + "log_likelihood": 21401.451627822393, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32980940000015835, + "n": 1149, + "solve_s": 0.32980940000015835, + "outer_iterations": 140, + "inner_iterations": 11, + "total_iterations": 151, + "n_passive": 902, + "warm_start_errors": 123, + "index": 7 + }, + { + "warm_start_error_fraction": 0.1409921671018277, + "reconstruction_s": 2.4972082000003866, + "eval_s": 2.5744345000002795, + "log_likelihood": 21654.932007862906, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.7221934999997757, + "n": 1149, + "solve_s": 0.7221934999997757, + "outer_iterations": 246, + "inner_iterations": 31, + "total_iterations": 277, + "n_passive": 920, + "warm_start_errors": 162, + "index": 8 + }, + { + "warm_start_error_fraction": 0.0731070496083551, + "reconstruction_s": 1.9673836000001756, + "eval_s": 2.0382724000000962, + "log_likelihood": 15446.649038322048, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3490406000000803, + "n": 1149, + "solve_s": 0.3490406000000803, + "outer_iterations": 172, + "inner_iterations": 11, + "total_iterations": 183, + "n_passive": 942, + "warm_start_errors": 84, + "index": 9 + }, + { + "warm_start_error_fraction": 0.06962576153176675, + "reconstruction_s": 2.2733871000000363, + "eval_s": 2.3652609000000666, + "log_likelihood": 24313.008074311838, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.683017800000016, + "n": 1149, + "solve_s": 0.683017800000016, + "outer_iterations": 196, + "inner_iterations": 205, + "total_iterations": 401, + "n_passive": 926, + "warm_start_errors": 80, + "index": 10 + }, + { + "warm_start_error_fraction": 0.08964316797214969, + "reconstruction_s": 1.9547655000005761, + "eval_s": 2.0231646000002, + "log_likelihood": 18450.248452275617, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2739878000002136, + "n": 1149, + "solve_s": 0.2739878000002136, + "outer_iterations": 139, + "inner_iterations": 2, + "total_iterations": 141, + "n_passive": 945, + "warm_start_errors": 103, + "index": 11 + }, + { + "warm_start_error_fraction": 0.10791993037423847, + "reconstruction_s": 2.0871071999999913, + "eval_s": 2.1533102000003055, + "log_likelihood": 20520.07718852685, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.32323840000026394, + "n": 1149, + "solve_s": 0.32323840000026394, + "outer_iterations": 138, + "inner_iterations": 39, + "total_iterations": 177, + "n_passive": 941, + "warm_start_errors": 124, + "index": 12 + }, + { + "warm_start_error_fraction": 0.05657093124456049, + "reconstruction_s": 1.7150289000001067, + "eval_s": 1.779740099999799, + "log_likelihood": 23692.010687028196, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18995489999997517, + "n": 1149, + "solve_s": 0.18995489999997517, + "outer_iterations": 82, + "inner_iterations": 8, + "total_iterations": 90, + "n_passive": 960, + "warm_start_errors": 65, + "index": 13 + }, + { + "warm_start_error_fraction": 0.04351610095735422, + "reconstruction_s": 1.758915700000216, + "eval_s": 1.8202594000003955, + "log_likelihood": 25380.78542724962, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1538660999995045, + "n": 1149, + "solve_s": 0.1538660999995045, + "outer_iterations": 61, + "inner_iterations": 7, + "total_iterations": 68, + "n_passive": 958, + "warm_start_errors": 50, + "index": 14 + }, + { + "warm_start_error_fraction": 0.03220191470844212, + "reconstruction_s": 1.542606799999703, + "eval_s": 1.6092707999996492, + "log_likelihood": 23839.966868313502, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06834929999968153, + "n": 1149, + "solve_s": 0.06834929999968153, + "outer_iterations": 24, + "inner_iterations": 1, + "total_iterations": 25, + "n_passive": 969, + "warm_start_errors": 37, + "index": 15 + }, + { + "warm_start_error_fraction": 0.10791993037423847, + "reconstruction_s": 2.460299500000474, + "eval_s": 2.556220999999823, + "log_likelihood": 25432.257565752396, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6766998999992211, + "n": 1149, + "solve_s": 0.6766998999992211, + "outer_iterations": 249, + "inner_iterations": 234, + "total_iterations": 483, + "n_passive": 931, + "warm_start_errors": 124, + "index": 16 + }, + { + "warm_start_error_fraction": 0.08268059181897303, + "reconstruction_s": 1.7319836000006035, + "eval_s": 1.7973085000003266, + "log_likelihood": 27302.229027713613, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28471130000070843, + "n": 1149, + "solve_s": 0.28471130000070843, + "outer_iterations": 123, + "inner_iterations": 43, + "total_iterations": 166, + "n_passive": 942, + "warm_start_errors": 95, + "index": 17 + }, + { + "warm_start_error_fraction": 0.09573542210617929, + "reconstruction_s": 3.0346340999994936, + "eval_s": 3.1183148999998593, + "log_likelihood": 22914.052021782307, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.8770081000002392, + "n": 1149, + "solve_s": 0.8770081000002392, + "outer_iterations": 407, + "inner_iterations": 49, + "total_iterations": 456, + "n_passive": 938, + "warm_start_errors": 110, + "index": 18 + }, + { + "warm_start_error_fraction": 0.040905134899912966, + "reconstruction_s": 1.636819899999864, + "eval_s": 1.7165352000001803, + "log_likelihood": 20254.66776540329, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.12394809999932477, + "n": 1149, + "solve_s": 0.12394809999932477, + "outer_iterations": 39, + "inner_iterations": 5, + "total_iterations": 44, + "n_passive": 955, + "warm_start_errors": 47, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_rectangular_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_rectangular_v2026.8.17.1.png new file mode 100644 index 000000000..3508173b8 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_rectangular_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_source_complex_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_source_complex_v2026.8.17.1.json new file mode 100644 index 000000000..7d5f8b1e5 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_source_complex_v2026.8.17.1.json @@ -0,0 +1,1767 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "hst", + "model": "source_complex", + "configuration": { + "model": "source_complex", + "dataset": "source_complex", + "adapt_image_source": "positive_clipped_data", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.05, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 15380, + "over_sampled_pixels": 18068, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 4.449207200001183, + "total_memo_seed_retries": 0, + "total_warm_start_fallbacks": 0, + "nnls_warm_start_error_tolerance": 1.5, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.3751908396946565, + "mean_warm_start_error_fraction": 0.37751908396946565, + "p90_warm_start_error_fraction": 0.3895419847328244, + "max_warm_start_error_fraction": 0.3946564885496183, + "median_outer_iterations": 237.0, + "mean_outer_iterations": 237.2, + "p90_outer_iterations": 253.3, + "max_outer_iterations": 259.0, + "median_inner_iterations": 86.5, + "mean_inner_iterations": 87.35, + "p90_inner_iterations": 101.4, + "max_inner_iterations": 106.0, + "median_total_iterations": 322.5, + "mean_total_iterations": 324.55, + "p90_total_iterations": 359.1, + "max_total_iterations": 361.0, + "median_n_passive": 513.0, + "mean_n_passive": 512.65, + "p90_n_passive": 518.1, + "max_n_passive": 522.0, + "median_warm_start_errors": 491.5, + "mean_warm_start_errors": 494.55, + "p90_warm_start_errors": 510.3, + "max_warm_start_errors": 517.0, + "median_solve_s": 0.3837411000013162, + "mean_solve_s": 0.40739792500007754, + "p90_solve_s": 0.5192327699995077, + "max_solve_s": 0.6496409000010317, + "median_reconstruction_s": 1.1844636499999979, + "mean_reconstruction_s": 1.205945425000209, + "p90_reconstruction_s": 1.3384509499999695, + "max_reconstruction_s": 1.494651499999236, + "median_eval_s": 1.2523866999999882, + "mean_eval_s": 1.2729460749998907, + "p90_eval_s": 1.436062659999334, + "max_eval_s": 1.5612433000005694 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.07213740458015266, + "mean_warm_start_error_fraction": 0.07916030534351146, + "p90_warm_start_error_fraction": 0.10786259541984733, + "max_warm_start_error_fraction": 0.3648854961832061, + "median_outer_iterations": 70.0, + "mean_outer_iterations": 69.6, + "p90_outer_iterations": 105.2, + "max_outer_iterations": 236.0, + "median_inner_iterations": 22.5, + "mean_inner_iterations": 23.65, + "p90_inner_iterations": 42.1, + "max_inner_iterations": 75.0, + "median_total_iterations": 95.0, + "mean_total_iterations": 93.25, + "p90_total_iterations": 141.9, + "max_total_iterations": 311.0, + "median_n_passive": 513.0, + "mean_n_passive": 512.65, + "p90_n_passive": 518.1, + "max_n_passive": 522.0, + "median_warm_start_errors": 94.5, + "mean_warm_start_errors": 103.7, + "p90_warm_start_errors": 141.3, + "max_warm_start_errors": 478.0, + "median_solve_s": 0.09004585000002407, + "mean_solve_s": 0.09224797999995644, + "p90_solve_s": 0.11261627999920165, + "max_solve_s": 0.4039963000013813, + "median_reconstruction_s": 0.8006323500003418, + "mean_reconstruction_s": 0.8563341449999825, + "p90_reconstruction_s": 1.0220723400003409, + "max_reconstruction_s": 1.345832099999825, + "median_eval_s": 0.86646114999985, + "mean_eval_s": 0.9193965950001257, + "p90_eval_s": 1.0974535899995317, + "max_eval_s": 1.4150939000010112 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 4.0712881142276965e-16, + "max_abs_reconstruction_deviation": 9.159339953157541e-13 + }, + "iteration_reduction_ratio": 3.3947368421052633, + "solve_speedup_ratio": 4.261618941919185, + "eval_speedup_ratio": 1.4454043323237342, + "rows": { + "off": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3648854961832061, + "reconstruction_s": 1.2406069000007847, + "eval_s": 1.304079499999716, + "log_likelihood": 35768.2627616579, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.44208049999906507, + "n": 1310, + "solve_s": 0.44208049999906507, + "outer_iterations": 236, + "inner_iterations": 75, + "total_iterations": 311, + "n_passive": 510, + "warm_start_errors": 478, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3763358778625954, + "reconstruction_s": 1.494651499999236, + "eval_s": 1.5612433000005694, + "log_likelihood": 35399.889099577056, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.6496409000010317, + "n": 1310, + "solve_s": 0.6496409000010317, + "outer_iterations": 245, + "inner_iterations": 94, + "total_iterations": 339, + "n_passive": 518, + "warm_start_errors": 493, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.366412213740458, + "reconstruction_s": 1.0743476999996346, + "eval_s": 1.1406221999986883, + "log_likelihood": 35592.067901500166, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.35133529999984603, + "n": 1310, + "solve_s": 0.35133529999984603, + "outer_iterations": 253, + "inner_iterations": 106, + "total_iterations": 359, + "n_passive": 517, + "warm_start_errors": 480, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3801526717557252, + "reconstruction_s": 1.168156200001249, + "eval_s": 1.2374347999993915, + "log_likelihood": 35554.535385205556, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43924320000041917, + "n": 1310, + "solve_s": 0.43924320000041917, + "outer_iterations": 222, + "inner_iterations": 71, + "total_iterations": 293, + "n_passive": 512, + "warm_start_errors": 498, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3885496183206107, + "reconstruction_s": 1.2185799000017141, + "eval_s": 1.278549299999213, + "log_likelihood": 35440.254952261836, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4606650999994599, + "n": 1310, + "solve_s": 0.4606650999994599, + "outer_iterations": 247, + "inner_iterations": 87, + "total_iterations": 334, + "n_passive": 519, + "warm_start_errors": 509, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3702290076335878, + "reconstruction_s": 1.293038300000262, + "eval_s": 1.3520473000007769, + "log_likelihood": 35617.94689386428, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4119401999996626, + "n": 1310, + "solve_s": 0.4119401999996626, + "outer_iterations": 241, + "inner_iterations": 89, + "total_iterations": 330, + "n_passive": 518, + "warm_start_errors": 485, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.36793893129770994, + "reconstruction_s": 1.1823304000008648, + "eval_s": 1.2411926999993739, + "log_likelihood": 35389.998907722176, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3467263999991701, + "n": 1310, + "solve_s": 0.3467263999991701, + "outer_iterations": 259, + "inner_iterations": 101, + "total_iterations": 360, + "n_passive": 522, + "warm_start_errors": 482, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.36717557251908395, + "reconstruction_s": 1.1574090000012802, + "eval_s": 1.213904599999296, + "log_likelihood": 35742.77923862256, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30524549999972805, + "n": 1310, + "solve_s": 0.30524549999972805, + "outer_iterations": 229, + "inner_iterations": 82, + "total_iterations": 311, + "n_passive": 502, + "warm_start_errors": 481, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.38778625954198476, + "reconstruction_s": 1.0798608000004606, + "eval_s": 1.13854419999916, + "log_likelihood": 35583.53825680174, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3181220000005851, + "n": 1310, + "solve_s": 0.3181220000005851, + "outer_iterations": 238, + "inner_iterations": 88, + "total_iterations": 326, + "n_passive": 510, + "warm_start_errors": 508, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.36870229007633587, + "reconstruction_s": 1.2095815000011498, + "eval_s": 1.2737049999996088, + "log_likelihood": 35505.74719855371, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3619222000015725, + "n": 1310, + "solve_s": 0.3619222000015725, + "outer_iterations": 256, + "inner_iterations": 105, + "total_iterations": 361, + "n_passive": 513, + "warm_start_errors": 483, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3893129770992366, + "reconstruction_s": 0.992758700000195, + "eval_s": 1.0481638000001112, + "log_likelihood": 36291.68120711284, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31407410000065283, + "n": 1310, + "solve_s": 0.31407410000065283, + "outer_iterations": 230, + "inner_iterations": 85, + "total_iterations": 315, + "n_passive": 512, + "warm_start_errors": 510, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.37404580152671757, + "reconstruction_s": 0.9642986999988352, + "eval_s": 1.0295461000005162, + "log_likelihood": 35651.34551616697, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30896559999928286, + "n": 1310, + "solve_s": 0.30896559999928286, + "outer_iterations": 226, + "inner_iterations": 85, + "total_iterations": 311, + "n_passive": 516, + "warm_start_errors": 490, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.36793893129770994, + "reconstruction_s": 1.1529989999999088, + "eval_s": 1.2292541999995592, + "log_likelihood": 35901.521078982376, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4485213999996631, + "n": 1310, + "solve_s": 0.4485213999996631, + "outer_iterations": 230, + "inner_iterations": 91, + "total_iterations": 321, + "n_passive": 513, + "warm_start_errors": 482, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.38244274809160306, + "reconstruction_s": 1.3287688000000344, + "eval_s": 1.4295731999991403, + "log_likelihood": 35877.69374367331, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.46182760000010603, + "n": 1310, + "solve_s": 0.46182760000010603, + "outer_iterations": 230, + "inner_iterations": 90, + "total_iterations": 320, + "n_passive": 510, + "warm_start_errors": 501, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3946564885496183, + "reconstruction_s": 1.425590299999385, + "eval_s": 1.4944678000010754, + "log_likelihood": 35494.409302838365, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5095273999995698, + "n": 1310, + "solve_s": 0.5095273999995698, + "outer_iterations": 227, + "inner_iterations": 83, + "total_iterations": 310, + "n_passive": 507, + "warm_start_errors": 517, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3732824427480916, + "reconstruction_s": 1.178741200001241, + "eval_s": 1.2575347000001784, + "log_likelihood": 35525.321912254374, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.40556000000105996, + "n": 1310, + "solve_s": 0.40556000000105996, + "outer_iterations": 224, + "inner_iterations": 76, + "total_iterations": 300, + "n_passive": 510, + "warm_start_errors": 489, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3709923664122137, + "reconstruction_s": 1.186596899999131, + "eval_s": 1.247238699999798, + "log_likelihood": 35500.701019232896, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33292790000086825, + "n": 1310, + "solve_s": 0.33292790000086825, + "outer_iterations": 211, + "inner_iterations": 78, + "total_iterations": 289, + "n_passive": 514, + "warm_start_errors": 486, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3893129770992366, + "reconstruction_s": 1.161285800000769, + "eval_s": 1.217432600000393, + "log_likelihood": 35249.97034530486, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3427652999998827, + "n": 1310, + "solve_s": 0.3427652999998827, + "outer_iterations": 250, + "inner_iterations": 92, + "total_iterations": 342, + "n_passive": 513, + "warm_start_errors": 510, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3786259541984733, + "reconstruction_s": 1.2956997999990563, + "eval_s": 1.3533187000011822, + "log_likelihood": 35473.98999557113, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33028680000097665, + "n": 1310, + "solve_s": 0.33028680000097665, + "outer_iterations": 241, + "inner_iterations": 83, + "total_iterations": 324, + "n_passive": 513, + "warm_start_errors": 496, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3916030534351145, + "reconstruction_s": 1.3136070999989897, + "eval_s": 1.4110688000000664, + "log_likelihood": 35716.7045006313, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.606581099998948, + "n": 1310, + "solve_s": 0.606581099998948, + "outer_iterations": 249, + "inner_iterations": 86, + "total_iterations": 335, + "n_passive": 504, + "warm_start_errors": 513, + "index": 19 + } + ], + "on": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3648854961832061, + "reconstruction_s": 1.345832099999825, + "eval_s": 1.4150939000010112, + "log_likelihood": 35768.2627616579, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4039963000013813, + "n": 1310, + "solve_s": 0.4039963000013813, + "outer_iterations": 236, + "inner_iterations": 75, + "total_iterations": 311, + "n_passive": 510, + "warm_start_errors": 478, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.08396946564885496, + "reconstruction_s": 0.9418606999988697, + "eval_s": 1.004224999998769, + "log_likelihood": 35399.889099577056, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09855990000141901, + "n": 1310, + "solve_s": 0.09855990000141901, + "outer_iterations": 81, + "inner_iterations": 25, + "total_iterations": 106, + "n_passive": 518, + "warm_start_errors": 110, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.04198473282442748, + "reconstruction_s": 0.8217248000000836, + "eval_s": 0.8778476999996201, + "log_likelihood": 35592.067901500166, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04990580000048794, + "n": 1310, + "solve_s": 0.04990580000048794, + "outer_iterations": 42, + "inner_iterations": 16, + "total_iterations": 58, + "n_passive": 517, + "warm_start_errors": 55, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.10763358778625955, + "reconstruction_s": 0.7834660000007716, + "eval_s": 0.8414521000013337, + "log_likelihood": 35554.535385205556, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10387019999870972, + "n": 1310, + "solve_s": 0.10387019999870972, + "outer_iterations": 99, + "inner_iterations": 42, + "total_iterations": 141, + "n_passive": 512, + "warm_start_errors": 141, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 1.0201101000002382, + "eval_s": 1.0966671999995015, + "log_likelihood": 35440.25495226183, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.096420500000022, + "n": 1310, + "solve_s": 0.096420500000022, + "outer_iterations": 58, + "inner_iterations": 16, + "total_iterations": 74, + "n_passive": 519, + "warm_start_errors": 79, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.083206106870229, + "reconstruction_s": 1.0397325000012643, + "eval_s": 1.1045310999998037, + "log_likelihood": 35617.94689386428, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08719969999947352, + "n": 1310, + "solve_s": 0.08719969999947352, + "outer_iterations": 70, + "inner_iterations": 31, + "total_iterations": 101, + "n_passive": 518, + "warm_start_errors": 109, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0717557251908397, + "reconstruction_s": 0.9403805999991164, + "eval_s": 1.0049217000014323, + "log_likelihood": 35389.998907722176, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08283579999988433, + "n": 1310, + "solve_s": 0.08283579999988433, + "outer_iterations": 72, + "inner_iterations": 23, + "total_iterations": 95, + "n_passive": 522, + "warm_start_errors": 94, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.08854961832061069, + "reconstruction_s": 0.7366507000006095, + "eval_s": 0.7995692000004055, + "log_likelihood": 35742.77923862255, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.08336159999998927, + "n": 1310, + "solve_s": 0.08336159999998927, + "outer_iterations": 70, + "inner_iterations": 25, + "total_iterations": 95, + "n_passive": 502, + "warm_start_errors": 116, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.1099236641221374, + "reconstruction_s": 0.8484232999999222, + "eval_s": 0.9073384999992413, + "log_likelihood": 35583.53825680174, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11255779999919469, + "n": 1310, + "solve_s": 0.11255779999919469, + "outer_iterations": 105, + "inner_iterations": 36, + "total_iterations": 141, + "n_passive": 510, + "warm_start_errors": 144, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 0.8126375000010739, + "eval_s": 0.8784231999998156, + "log_likelihood": 35505.74719855371, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04672559999926307, + "n": 1310, + "solve_s": 0.04672559999926307, + "outer_iterations": 33, + "inner_iterations": 8, + "total_iterations": 41, + "n_passive": 513, + "warm_start_errors": 59, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.07251908396946564, + "reconstruction_s": 0.9835654999988037, + "eval_s": 1.0503905999994458, + "log_likelihood": 36291.68120711283, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09588739999890095, + "n": 1310, + "solve_s": 0.09588739999890095, + "outer_iterations": 70, + "inner_iterations": 30, + "total_iterations": 100, + "n_passive": 512, + "warm_start_errors": 95, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.10534351145038168, + "reconstruction_s": 0.788126999999804, + "eval_s": 0.8468922000010934, + "log_likelihood": 35651.34551616698, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.11314259999926435, + "n": 1310, + "solve_s": 0.11314259999926435, + "outer_iterations": 107, + "inner_iterations": 43, + "total_iterations": 150, + "n_passive": 516, + "warm_start_errors": 138, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.029770992366412213, + "reconstruction_s": 0.7041489000002912, + "eval_s": 0.7616145000010874, + "log_likelihood": 35901.521078982376, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04070499999943422, + "n": 1310, + "solve_s": 0.04070499999943422, + "outer_iterations": 33, + "inner_iterations": 10, + "total_iterations": 43, + "n_passive": 513, + "warm_start_errors": 39, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.01450381679389313, + "reconstruction_s": 0.7092085000003863, + "eval_s": 0.7847432999988087, + "log_likelihood": 35877.693743673306, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03856129999985569, + "n": 1310, + "solve_s": 0.03856129999985569, + "outer_iterations": 11, + "inner_iterations": 3, + "total_iterations": 14, + "n_passive": 510, + "warm_start_errors": 19, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.04198473282442748, + "reconstruction_s": 0.767075499999919, + "eval_s": 0.8252725000002101, + "log_likelihood": 35494.409302838365, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04137079999964044, + "n": 1310, + "solve_s": 0.04137079999964044, + "outer_iterations": 31, + "inner_iterations": 8, + "total_iterations": 39, + "n_passive": 507, + "warm_start_errors": 55, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0633587786259542, + "reconstruction_s": 0.776125200000024, + "eval_s": 0.85507460000008, + "log_likelihood": 35525.32191225437, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10420589999921503, + "n": 1310, + "solve_s": 0.10420589999921503, + "outer_iterations": 66, + "inner_iterations": 18, + "total_iterations": 84, + "n_passive": 510, + "warm_start_errors": 83, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 0.7886271999996097, + "eval_s": 0.8443126000001939, + "log_likelihood": 35500.7010192329, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.017612200001167366, + "n": 1310, + "solve_s": 0.017612200001167366, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 514, + "warm_start_errors": 14, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.083206106870229, + "reconstruction_s": 0.8225899999997637, + "eval_s": 0.8791997999996966, + "log_likelihood": 35249.97034530486, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.10760260000097333, + "n": 1310, + "solve_s": 0.10760260000097333, + "outer_iterations": 100, + "inner_iterations": 39, + "total_iterations": 139, + "n_passive": 513, + "warm_start_errors": 109, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 0.7665348999998969, + "eval_s": 0.822780700000294, + "log_likelihood": 35473.98999557113, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.027546600000277977, + "n": 1310, + "solve_s": 0.027546600000277977, + "outer_iterations": 21, + "inner_iterations": 2, + "total_iterations": 23, + "n_passive": 513, + "warm_start_errors": 36, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.07709923664122137, + "reconstruction_s": 0.7298618999993778, + "eval_s": 0.7875815000006696, + "log_likelihood": 35716.7045006313, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09289200000057463, + "n": 1310, + "solve_s": 0.09289200000057463, + "outer_iterations": 78, + "inner_iterations": 22, + "total_iterations": 100, + "n_passive": 504, + "warm_start_errors": 101, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.39274809160305346, + "mean_warm_start_error_fraction": 0.38927480916030527, + "p90_warm_start_error_fraction": 0.4016793893129771, + "max_warm_start_error_fraction": 0.41755725190839693, + "median_outer_iterations": 249.5, + "mean_outer_iterations": 251.4, + "p90_outer_iterations": 273.1, + "max_outer_iterations": 309.0, + "median_inner_iterations": 91.5, + "mean_inner_iterations": 89.75, + "p90_inner_iterations": 108.0, + "max_inner_iterations": 118.0, + "median_total_iterations": 344.5, + "mean_total_iterations": 341.15, + "p90_total_iterations": 371.20000000000005, + "max_total_iterations": 427.0, + "median_n_passive": 503.5, + "mean_n_passive": 506.25, + "p90_n_passive": 524.0, + "max_n_passive": 524.0, + "median_warm_start_errors": 514.5, + "mean_warm_start_errors": 509.95, + "p90_warm_start_errors": 526.2, + "max_warm_start_errors": 547.0, + "median_solve_s": 0.3555950500003746, + "mean_solve_s": 0.36566885499996715, + "p90_solve_s": 0.4346389699996508, + "max_solve_s": 0.5398652000003494, + "median_reconstruction_s": 1.0914915499997733, + "mean_reconstruction_s": 1.1155484000000797, + "p90_reconstruction_s": 1.2564077199991517, + "max_reconstruction_s": 1.4100257999998576, + "median_eval_s": 1.1585505999992165, + "mean_eval_s": 1.17798378999978, + "p90_eval_s": 1.3178049500005726, + "max_eval_s": 1.4842185999987123 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.3, + "mean_warm_start_error_fraction": 0.28774809160305337, + "p90_warm_start_error_fraction": 0.3431297709923664, + "max_warm_start_error_fraction": 0.4015267175572519, + "median_outer_iterations": 312.5, + "mean_outer_iterations": 301.95, + "p90_outer_iterations": 362.6, + "max_outer_iterations": 381.0, + "median_inner_iterations": 164.5, + "mean_inner_iterations": 161.05, + "p90_inner_iterations": 204.8, + "max_inner_iterations": 224.0, + "median_total_iterations": 475.0, + "mean_total_iterations": 463.0, + "p90_total_iterations": 560.3000000000001, + "max_total_iterations": 605.0, + "median_n_passive": 503.5, + "mean_n_passive": 506.25, + "p90_n_passive": 524.0, + "max_n_passive": 524.0, + "median_warm_start_errors": 393.0, + "mean_warm_start_errors": 376.95, + "p90_warm_start_errors": 449.5, + "max_warm_start_errors": 526.0, + "median_solve_s": 0.37166625000008935, + "mean_solve_s": 0.37052852499991784, + "p90_solve_s": 0.4617843100000755, + "max_solve_s": 0.5312439000008453, + "median_reconstruction_s": 1.1412270999999237, + "mean_reconstruction_s": 1.1405403849998037, + "p90_reconstruction_s": 1.2402727600003347, + "max_reconstruction_s": 1.4572016000001895, + "median_eval_s": 1.2095759499998167, + "mean_eval_s": 1.2072750299999826, + "p90_eval_s": 1.3093303600002404, + "max_eval_s": 1.5280774000002566 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 2.1812755997733406e-16, + "max_abs_reconstruction_deviation": 9.862666239257578e-13 + }, + "iteration_reduction_ratio": 0.7252631578947368, + "solve_speedup_ratio": 0.9567590546633953, + "eval_speedup_ratio": 0.9578155055078534, + "rows": { + "off": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.4015267175572519, + "reconstruction_s": 1.1283605999997235, + "eval_s": 1.1869747000000643, + "log_likelihood": 33066.84209140426, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.36886670000058075, + "n": 1310, + "solve_s": 0.36886670000058075, + "outer_iterations": 263, + "inner_iterations": 88, + "total_iterations": 351, + "n_passive": 506, + "warm_start_errors": 526, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.36870229007633587, + "reconstruction_s": 0.975274200000058, + "eval_s": 1.0316736999993736, + "log_likelihood": 34658.47203298578, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2921225999998569, + "n": 1310, + "solve_s": 0.2921225999998569, + "outer_iterations": 230, + "inner_iterations": 75, + "total_iterations": 305, + "n_passive": 501, + "warm_start_errors": 483, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.39541984732824426, + "reconstruction_s": 1.0066088999992644, + "eval_s": 1.0685850999998365, + "log_likelihood": 35273.3366639478, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34327359999952023, + "n": 1310, + "solve_s": 0.34327359999952023, + "outer_iterations": 271, + "inner_iterations": 97, + "total_iterations": 368, + "n_passive": 524, + "warm_start_errors": 518, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.366412213740458, + "reconstruction_s": 0.9831781000011688, + "eval_s": 1.044987699999183, + "log_likelihood": 34138.6849972875, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33047679999981483, + "n": 1310, + "solve_s": 0.33047679999981483, + "outer_iterations": 220, + "inner_iterations": 89, + "total_iterations": 309, + "n_passive": 513, + "warm_start_errors": 480, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.37786259541984735, + "reconstruction_s": 1.4100257999998576, + "eval_s": 1.4842185999987123, + "log_likelihood": 38134.64345110007, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5398652000003494, + "n": 1310, + "solve_s": 0.5398652000003494, + "outer_iterations": 232, + "inner_iterations": 88, + "total_iterations": 320, + "n_passive": 524, + "warm_start_errors": 495, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.37709923664122136, + "reconstruction_s": 1.230855199999496, + "eval_s": 1.2895203999996738, + "log_likelihood": 38237.8487142431, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3675050000001647, + "n": 1310, + "solve_s": 0.3675050000001647, + "outer_iterations": 262, + "inner_iterations": 96, + "total_iterations": 358, + "n_passive": 516, + "warm_start_errors": 494, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.39083969465648855, + "reconstruction_s": 1.2079762000012124, + "eval_s": 1.2807539000004908, + "log_likelihood": 33356.43425773195, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4285364999996091, + "n": 1310, + "solve_s": 0.4285364999996091, + "outer_iterations": 248, + "inner_iterations": 92, + "total_iterations": 340, + "n_passive": 491, + "warm_start_errors": 512, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3938931297709924, + "reconstruction_s": 1.1988856000007218, + "eval_s": 1.2595448000010947, + "log_likelihood": 35726.98004068463, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.48956120000002556, + "n": 1310, + "solve_s": 0.48956120000002556, + "outer_iterations": 268, + "inner_iterations": 92, + "total_iterations": 360, + "n_passive": 514, + "warm_start_errors": 516, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3748091603053435, + "reconstruction_s": 1.053601499999786, + "eval_s": 1.1111029999992752, + "log_likelihood": 38406.0375730783, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3178797000000486, + "n": 1310, + "solve_s": 0.3178797000000486, + "outer_iterations": 246, + "inner_iterations": 82, + "total_iterations": 328, + "n_passive": 523, + "warm_start_errors": 491, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.38549618320610685, + "reconstruction_s": 0.9907896999993682, + "eval_s": 1.047067600000446, + "log_likelihood": 34231.82690933779, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28779109999959473, + "n": 1310, + "solve_s": 0.28779109999959473, + "outer_iterations": 240, + "inner_iterations": 78, + "total_iterations": 318, + "n_passive": 500, + "warm_start_errors": 505, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3916030534351145, + "reconstruction_s": 1.0023958000001585, + "eval_s": 1.0603033999996114, + "log_likelihood": 34163.884100841526, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.30574610000076063, + "n": 1310, + "solve_s": 0.30574610000076063, + "outer_iterations": 258, + "inner_iterations": 91, + "total_iterations": 349, + "n_passive": 495, + "warm_start_errors": 513, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.36412213740458016, + "reconstruction_s": 1.0244329000015568, + "eval_s": 1.0886744999988878, + "log_likelihood": 36070.883766715415, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3242302999988169, + "n": 1310, + "solve_s": 0.3242302999988169, + "outer_iterations": 180, + "inner_iterations": 70, + "total_iterations": 250, + "n_passive": 513, + "warm_start_errors": 477, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.39618320610687024, + "reconstruction_s": 1.2550092999990738, + "eval_s": 1.3162592000007862, + "log_likelihood": 34089.51631839466, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3724899999997433, + "n": 1310, + "solve_s": 0.3724899999997433, + "outer_iterations": 250, + "inner_iterations": 108, + "total_iterations": 358, + "n_passive": 492, + "warm_start_errors": 519, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.39923664122137403, + "reconstruction_s": 1.044637099999818, + "eval_s": 1.10222210000029, + "log_likelihood": 33584.66553008446, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3379569000007905, + "n": 1310, + "solve_s": 0.3379569000007905, + "outer_iterations": 259, + "inner_iterations": 93, + "total_iterations": 352, + "n_passive": 489, + "warm_start_errors": 523, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3984732824427481, + "reconstruction_s": 1.0722860999994737, + "eval_s": 1.1323855999999068, + "log_likelihood": 38870.66347430777, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.38027430000147433, + "n": 1310, + "solve_s": 0.38027430000147433, + "outer_iterations": 292, + "inner_iterations": 108, + "total_iterations": 400, + "n_passive": 524, + "warm_start_errors": 522, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.3977099236641221, + "reconstruction_s": 1.1171676000012667, + "eval_s": 1.1843497000008938, + "log_likelihood": 36897.666205087415, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.41178239999862853, + "n": 1310, + "solve_s": 0.41178239999862853, + "outer_iterations": 309, + "inner_iterations": 118, + "total_iterations": 427, + "n_passive": 522, + "warm_start_errors": 521, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.41755725190839693, + "reconstruction_s": 1.1106970000000729, + "eval_s": 1.168027899999288, + "log_likelihood": 35024.04230310099, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31470679999983986, + "n": 1310, + "solve_s": 0.31470679999983986, + "outer_iterations": 262, + "inner_iterations": 78, + "total_iterations": 340, + "n_passive": 494, + "warm_start_errors": 547, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.4015267175572519, + "reconstruction_s": 1.0714017000009335, + "eval_s": 1.149073299999145, + "log_likelihood": 36266.77531717943, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3718766999991203, + "n": 1310, + "solve_s": 0.3718766999991203, + "outer_iterations": 247, + "inner_iterations": 99, + "total_iterations": 346, + "n_passive": 488, + "warm_start_errors": 526, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.383969465648855, + "reconstruction_s": 1.2689934999998513, + "eval_s": 1.3317166999986512, + "log_likelihood": 35238.33578544314, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3436851000005845, + "n": 1310, + "solve_s": 0.3436851000005845, + "outer_iterations": 249, + "inner_iterations": 94, + "total_iterations": 343, + "n_passive": 497, + "warm_start_errors": 503, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.4030534351145038, + "reconstruction_s": 1.1583911999987322, + "eval_s": 1.222233899999992, + "log_likelihood": 34102.91395509731, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3847501000000193, + "n": 1310, + "solve_s": 0.3847501000000193, + "outer_iterations": 242, + "inner_iterations": 59, + "total_iterations": 301, + "n_passive": 499, + "warm_start_errors": 528, + "index": 19 + } + ], + "on": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.4015267175572519, + "reconstruction_s": 1.2112357999994856, + "eval_s": 1.2905644000002212, + "log_likelihood": 33066.84209140426, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.46091100000012375, + "n": 1310, + "solve_s": 0.46091100000012375, + "outer_iterations": 263, + "inner_iterations": 88, + "total_iterations": 351, + "n_passive": 506, + "warm_start_errors": 526, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.25725190839694656, + "reconstruction_s": 1.1970201000003726, + "eval_s": 1.2741027000010945, + "log_likelihood": 34658.47203298579, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34651779999876453, + "n": 1310, + "solve_s": 0.34651779999876453, + "outer_iterations": 300, + "inner_iterations": 156, + "total_iterations": 456, + "n_passive": 501, + "warm_start_errors": 337, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3320610687022901, + "reconstruction_s": 1.1429336000001058, + "eval_s": 1.2035029999988183, + "log_likelihood": 35273.336663947804, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.40860950000023877, + "n": 1310, + "solve_s": 0.40860950000023877, + "outer_iterations": 381, + "inner_iterations": 224, + "total_iterations": 605, + "n_passive": 524, + "warm_start_errors": 435, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.2862595419847328, + "reconstruction_s": 1.011813400000392, + "eval_s": 1.0690360000007786, + "log_likelihood": 34138.684997287506, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.33862589999989723, + "n": 1310, + "solve_s": 0.33862589999989723, + "outer_iterations": 302, + "inner_iterations": 162, + "total_iterations": 464, + "n_passive": 513, + "warm_start_errors": 375, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3045801526717557, + "reconstruction_s": 1.1845525999997335, + "eval_s": 1.245665899999949, + "log_likelihood": 38134.64345110007, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3628855000006297, + "n": 1310, + "solve_s": 0.3628855000006297, + "outer_iterations": 326, + "inner_iterations": 160, + "total_iterations": 486, + "n_passive": 524, + "warm_start_errors": 399, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.15572519083969466, + "reconstruction_s": 1.0097463999991305, + "eval_s": 1.0718867000014143, + "log_likelihood": 38237.8487142431, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2014417999998841, + "n": 1310, + "solve_s": 0.2014417999998841, + "outer_iterations": 139, + "inner_iterations": 55, + "total_iterations": 194, + "n_passive": 516, + "warm_start_errors": 204, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3, + "reconstruction_s": 1.033071399999244, + "eval_s": 1.0964232999995147, + "log_likelihood": 33356.43425773194, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3398596999995789, + "n": 1310, + "solve_s": 0.3398596999995789, + "outer_iterations": 323, + "inner_iterations": 187, + "total_iterations": 510, + "n_passive": 491, + "warm_start_errors": 393, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3015267175572519, + "reconstruction_s": 1.1775562999991962, + "eval_s": 1.2540002999994613, + "log_likelihood": 35726.98004068463, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43260239999835903, + "n": 1310, + "solve_s": 0.43260239999835903, + "outer_iterations": 362, + "inner_iterations": 193, + "total_iterations": 555, + "n_passive": 514, + "warm_start_errors": 395, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.2099236641221374, + "reconstruction_s": 1.0810411000002205, + "eval_s": 1.1487397999990208, + "log_likelihood": 38406.0375730783, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2803490000005695, + "n": 1310, + "solve_s": 0.2803490000005695, + "outer_iterations": 215, + "inner_iterations": 104, + "total_iterations": 319, + "n_passive": 523, + "warm_start_errors": 275, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3152671755725191, + "reconstruction_s": 1.2210947000003216, + "eval_s": 1.2800337000007858, + "log_likelihood": 34231.82690933779, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.380446999999549, + "n": 1310, + "solve_s": 0.380446999999549, + "outer_iterations": 351, + "inner_iterations": 199, + "total_iterations": 550, + "n_passive": 500, + "warm_start_errors": 413, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.32748091603053436, + "reconstruction_s": 1.0955160000012256, + "eval_s": 1.1597927000002528, + "log_likelihood": 34163.88410084153, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3991882999998779, + "n": 1310, + "solve_s": 0.3991882999998779, + "outer_iterations": 347, + "inner_iterations": 212, + "total_iterations": 559, + "n_passive": 495, + "warm_start_errors": 429, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.2687022900763359, + "reconstruction_s": 1.0646483000000444, + "eval_s": 1.1298859999988053, + "log_likelihood": 36070.88376671542, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.34302589999970223, + "n": 1310, + "solve_s": 0.34302589999970223, + "outer_iterations": 295, + "inner_iterations": 167, + "total_iterations": 462, + "n_passive": 513, + "warm_start_errors": 352, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3427480916030534, + "reconstruction_s": 1.4572016000001895, + "eval_s": 1.5280774000002566, + "log_likelihood": 34089.51631839466, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4696440999996412, + "n": 1310, + "solve_s": 0.4696440999996412, + "outer_iterations": 346, + "inner_iterations": 190, + "total_iterations": 536, + "n_passive": 492, + "warm_start_errors": 449, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.21603053435114503, + "reconstruction_s": 0.9579748999985895, + "eval_s": 1.0208726000000752, + "log_likelihood": 33584.66553008446, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24992530000054103, + "n": 1310, + "solve_s": 0.24992530000054103, + "outer_iterations": 232, + "inner_iterations": 118, + "total_iterations": 350, + "n_passive": 489, + "warm_start_errors": 283, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3, + "reconstruction_s": 1.1489833999985422, + "eval_s": 1.215648900000815, + "log_likelihood": 38870.66347430777, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.4365305999999691, + "n": 1310, + "solve_s": 0.4365305999999691, + "outer_iterations": 337, + "inner_iterations": 173, + "total_iterations": 510, + "n_passive": 524, + "warm_start_errors": 393, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.26106870229007634, + "reconstruction_s": 1.1395205999997415, + "eval_s": 1.2266572999997152, + "log_likelihood": 36897.66620508742, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.43212409999978263, + "n": 1310, + "solve_s": 0.43212409999978263, + "outer_iterations": 268, + "inner_iterations": 146, + "total_iterations": 414, + "n_passive": 522, + "warm_start_errors": 342, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3465648854961832, + "reconstruction_s": 1.4128753000004508, + "eval_s": 1.4782240000004094, + "log_likelihood": 35024.042303101, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.5312439000008453, + "n": 1310, + "solve_s": 0.5312439000008453, + "outer_iterations": 351, + "inner_iterations": 198, + "total_iterations": 549, + "n_passive": 494, + "warm_start_errors": 454, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.2732824427480916, + "reconstruction_s": 1.180581800001164, + "eval_s": 1.2418065999991086, + "log_likelihood": 36266.77531717943, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3132663999986107, + "n": 1310, + "solve_s": 0.3132663999986107, + "outer_iterations": 280, + "inner_iterations": 161, + "total_iterations": 441, + "n_passive": 488, + "warm_start_errors": 358, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.2450381679389313, + "reconstruction_s": 0.9811236999994435, + "eval_s": 1.0431037999987893, + "log_likelihood": 35238.33578544314, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2840339000013046, + "n": 1310, + "solve_s": 0.2840339000013046, + "outer_iterations": 253, + "inner_iterations": 124, + "total_iterations": 377, + "n_passive": 497, + "warm_start_errors": 321, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.3099236641221374, + "reconstruction_s": 1.1023166999984824, + "eval_s": 1.1674755000003643, + "log_likelihood": 34102.91395509731, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39933840000048804, + "n": 1310, + "solve_s": 0.39933840000048804, + "outer_iterations": 368, + "inner_iterations": 204, + "total_iterations": 572, + "n_passive": 499, + "warm_start_errors": 406, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_source_complex_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_source_complex_v2026.8.17.1.png new file mode 100644 index 000000000..a881f7b3c Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_source_complex_v2026.8.17.1.png differ diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_v2026.8.17.1.json b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_v2026.8.17.1.json new file mode 100644 index 000000000..d96045b73 --- /dev/null +++ b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_v2026.8.17.1.json @@ -0,0 +1,1767 @@ +{ + "autolens_version": "2026.8.17.1", + "device": { + "backend": "cpu", + "device": "cpu:0", + "xla_flags": "--xla_disable_hlo_passes=constant_folding --xla_gpu_autotune_level=0", + "omp_num_threads": "1", + "cpu_count": 8 + }, + "instrument": "hst", + "model": "fiducial", + "configuration": { + "model": "fiducial", + "dataset": "hst", + "adapt_image_source": "lensed_source_from_truth_tracer", + "settings_overrides": { + "use_border_relocator": true + }, + "pixel_scale_arcsec": 0.05, + "mask_radius_arcsec": 3.5, + "image_pixels_masked": 15361, + "over_sampled_pixels": 17980, + "source_pixels": 1250, + "free_parameters": 11, + "inversion_path": "sparse_numba", + "use_jax": false, + "mesh": "delaunay_hilbert_1250", + "regularization": "constant_split", + "lens_light": "mge_60", + "n_instances_per_sequence": 20, + "random_walk_step_fraction_of_prior_width": 0.01, + "iid_unit_range": [ + 0.4, + 0.6 + ], + "seed": 498, + "omp_num_threads": "1", + "autoarray_numba_operated_memo": "0" + }, + "warmup_incl_numba_compile_s": 4.367720099999133, + "total_memo_seed_retries": 0, + "total_warm_start_fallbacks": 3, + "nnls_warm_start_error_tolerance": 1.5, + "parity_violations": [], + "sequences": { + "random_walk": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.02366412213740458, + "mean_warm_start_error_fraction": 0.023702290076335876, + "p90_warm_start_error_fraction": 0.02748091603053435, + "max_warm_start_error_fraction": 0.028244274809160305, + "median_outer_iterations": 24.5, + "mean_outer_iterations": 23.45, + "p90_outer_iterations": 27.0, + "max_outer_iterations": 29.0, + "median_inner_iterations": 8.0, + "mean_inner_iterations": 8.45, + "p90_inner_iterations": 11.0, + "max_inner_iterations": 12.0, + "median_total_iterations": 32.0, + "mean_total_iterations": 31.9, + "p90_total_iterations": 38.0, + "max_total_iterations": 41.0, + "median_n_passive": 1257.0, + "mean_n_passive": 1258.05, + "p90_n_passive": 1263.0, + "max_n_passive": 1263.0, + "median_warm_start_errors": 31.0, + "mean_warm_start_errors": 31.05, + "p90_warm_start_errors": 36.0, + "max_warm_start_errors": 37.0, + "median_solve_s": 0.20455000000129075, + "mean_solve_s": 0.2174330849998114, + "p90_solve_s": 0.265152469998975, + "max_solve_s": 0.31412769999951706, + "median_reconstruction_s": 1.549097350000011, + "mean_reconstruction_s": 1.5272900050000318, + "p90_reconstruction_s": 1.6335341899990454, + "max_reconstruction_s": 1.6693316000000777, + "median_eval_s": 1.61068830000022, + "mean_eval_s": 1.589810470000066, + "p90_eval_s": 1.6946705499998644, + "max_eval_s": 1.7396595000009256 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.007251908396946565, + "mean_warm_start_error_fraction": 0.008625954198473281, + "p90_warm_start_error_fraction": 0.013816793893129772, + "max_warm_start_error_fraction": 0.02366412213740458, + "median_outer_iterations": 5.5, + "mean_outer_iterations": 8.15, + "p90_outer_iterations": 17.400000000000006, + "max_outer_iterations": 29.0, + "median_inner_iterations": 1.0, + "mean_inner_iterations": 2.0, + "p90_inner_iterations": 2.200000000000003, + "max_inner_iterations": 12.0, + "median_total_iterations": 7.5, + "mean_total_iterations": 10.15, + "p90_total_iterations": 19.400000000000006, + "max_total_iterations": 41.0, + "median_n_passive": 1256.5, + "mean_n_passive": 1257.75, + "p90_n_passive": 1262.1, + "max_n_passive": 1263.0, + "median_warm_start_errors": 9.5, + "mean_warm_start_errors": 11.3, + "p90_warm_start_errors": 18.1, + "max_warm_start_errors": 31.0, + "median_solve_s": 0.06251510000038252, + "mean_solve_s": 0.07504798999989362, + "p90_solve_s": 0.10216445000005489, + "max_solve_s": 0.21680879999985336, + "median_reconstruction_s": 1.3270503499998085, + "mean_reconstruction_s": 1.3536076250000406, + "p90_reconstruction_s": 1.4933162299992546, + "max_reconstruction_s": 1.7331817000012961, + "median_eval_s": 1.3885196500004895, + "mean_eval_s": 1.4181188900000052, + "p90_eval_s": 1.5621165499998826, + "max_eval_s": 1.8074825999992754 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 1.2731920227827472e-15, + "max_abs_reconstruction_deviation": 2.462922310542126e-09 + }, + "iteration_reduction_ratio": 4.266666666666667, + "solve_speedup_ratio": 3.272009482509652, + "eval_speedup_ratio": 1.160003965373952, + "rows": { + "off": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.6018423999994411, + "eval_s": 1.6633984000000055, + "log_likelihood": 29034.99992341116, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23194130000047153, + "n": 1310, + "solve_s": 0.23194130000047153, + "outer_iterations": 29, + "inner_iterations": 12, + "total_iterations": 41, + "n_passive": 1262, + "warm_start_errors": 31, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.024427480916030534, + "reconstruction_s": 1.5429378000008, + "eval_s": 1.606390700000702, + "log_likelihood": 28849.582048876684, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23540019999927608, + "n": 1310, + "solve_s": 0.23540019999927608, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1260, + "warm_start_errors": 32, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.024427480916030534, + "reconstruction_s": 1.6107963000013115, + "eval_s": 1.676357399999688, + "log_likelihood": 28670.06086112797, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.25241490000007616, + "n": 1310, + "solve_s": 0.25241490000007616, + "outer_iterations": 25, + "inner_iterations": 11, + "total_iterations": 36, + "n_passive": 1257, + "warm_start_errors": 32, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.021374045801526718, + "reconstruction_s": 1.3674193000006198, + "eval_s": 1.4267228999997315, + "log_likelihood": 28883.348745978776, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.193254199999501, + "n": 1310, + "solve_s": 0.193254199999501, + "outer_iterations": 27, + "inner_iterations": 7, + "total_iterations": 34, + "n_passive": 1261, + "warm_start_errors": 28, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.019083969465648856, + "reconstruction_s": 1.6136846000008518, + "eval_s": 1.6722758999985672, + "log_likelihood": 28888.219309590837, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.27034160000039265, + "n": 1310, + "solve_s": 0.27034160000039265, + "outer_iterations": 22, + "inner_iterations": 7, + "total_iterations": 29, + "n_passive": 1262, + "warm_start_errors": 25, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 1.5767421000000468, + "eval_s": 1.6357803999999305, + "log_likelihood": 28682.766464090804, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20135239999945043, + "n": 1310, + "solve_s": 0.20135239999945043, + "outer_iterations": 24, + "inner_iterations": 7, + "total_iterations": 31, + "n_passive": 1255, + "warm_start_errors": 36, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.022900763358778626, + "reconstruction_s": 1.537735599998996, + "eval_s": 1.598865100000694, + "log_likelihood": 29003.39603966668, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2237035999987711, + "n": 1310, + "solve_s": 0.2237035999987711, + "outer_iterations": 25, + "inner_iterations": 10, + "total_iterations": 35, + "n_passive": 1262, + "warm_start_errors": 30, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.021374045801526718, + "reconstruction_s": 1.3538377000004402, + "eval_s": 1.4173842999989574, + "log_likelihood": 29098.354207998218, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.158084400000007, + "n": 1310, + "solve_s": 0.158084400000007, + "outer_iterations": 13, + "inner_iterations": 8, + "total_iterations": 21, + "n_passive": 1254, + "warm_start_errors": 28, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.5552568999992218, + "eval_s": 1.6149858999997377, + "log_likelihood": 29188.767818380577, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20510030000150437, + "n": 1310, + "solve_s": 0.20510030000150437, + "outer_iterations": 20, + "inner_iterations": 9, + "total_iterations": 29, + "n_passive": 1256, + "warm_start_errors": 31, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.022137404580152672, + "reconstruction_s": 1.6693316000000777, + "eval_s": 1.7396595000009256, + "log_likelihood": 28740.24100740815, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21092339999995602, + "n": 1310, + "solve_s": 0.21092339999995602, + "outer_iterations": 16, + "inner_iterations": 6, + "total_iterations": 22, + "n_passive": 1254, + "warm_start_errors": 29, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.4939039000000776, + "eval_s": 1.5609863000008772, + "log_likelihood": 28659.773960522565, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2029867999990529, + "n": 1310, + "solve_s": 0.2029867999990529, + "outer_iterations": 23, + "inner_iterations": 8, + "total_iterations": 31, + "n_passive": 1255, + "warm_start_errors": 31, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 1.6303047999990667, + "eval_s": 1.6908374000013282, + "log_likelihood": 28569.897575338688, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.19666889999825798, + "n": 1310, + "solve_s": 0.19666889999825798, + "outer_iterations": 26, + "inner_iterations": 6, + "total_iterations": 32, + "n_passive": 1257, + "warm_start_errors": 36, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.025190839694656488, + "reconstruction_s": 1.382647200000065, + "eval_s": 1.4447392000001855, + "log_likelihood": 28516.18652537265, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.1917170000015176, + "n": 1310, + "solve_s": 0.1917170000015176, + "outer_iterations": 26, + "inner_iterations": 7, + "total_iterations": 33, + "n_passive": 1257, + "warm_start_errors": 33, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.022900763358778626, + "reconstruction_s": 1.4971621000004234, + "eval_s": 1.5615553000006912, + "log_likelihood": 28593.06738215032, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2113759999992908, + "n": 1310, + "solve_s": 0.2113759999992908, + "outer_iterations": 23, + "inner_iterations": 7, + "total_iterations": 30, + "n_passive": 1257, + "warm_start_errors": 30, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 1.4525755000013305, + "eval_s": 1.511235800000577, + "log_likelihood": 28494.22512230963, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18886770000062825, + "n": 1310, + "solve_s": 0.18886770000062825, + "outer_iterations": 24, + "inner_iterations": 8, + "total_iterations": 32, + "n_passive": 1256, + "warm_start_errors": 36, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.028244274809160305, + "reconstruction_s": 1.4178678999996919, + "eval_s": 1.4808950999995432, + "log_likelihood": 28573.68521003123, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20399970000107714, + "n": 1310, + "solve_s": 0.20399970000107714, + "outer_iterations": 26, + "inner_iterations": 6, + "total_iterations": 32, + "n_passive": 1254, + "warm_start_errors": 37, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.025190839694656488, + "reconstruction_s": 1.6286464999993768, + "eval_s": 1.6913550999997824, + "log_likelihood": 28754.169652775512, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.31412769999951706, + "n": 1310, + "solve_s": 0.31412769999951706, + "outer_iterations": 23, + "inner_iterations": 10, + "total_iterations": 33, + "n_passive": 1253, + "warm_start_errors": 33, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.020610687022900764, + "reconstruction_s": 1.5764548999995895, + "eval_s": 1.6489340999996784, + "log_likelihood": 29100.48957849467, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2645758999988175, + "n": 1310, + "solve_s": 0.2645758999988175, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1263, + "warm_start_errors": 27, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0183206106870229, + "reconstruction_s": 1.6625986999988527, + "eval_s": 1.7245096000006015, + "log_likelihood": 29261.487227838043, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18797389999963343, + "n": 1310, + "solve_s": 0.18797389999963343, + "outer_iterations": 16, + "inner_iterations": 7, + "total_iterations": 23, + "n_passive": 1263, + "warm_start_errors": 24, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.024427480916030534, + "reconstruction_s": 1.3740543000003527, + "eval_s": 1.4293409999991127, + "log_likelihood": 29790.892417880026, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20385179999902903, + "n": 1310, + "solve_s": 0.20385179999902903, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1263, + "warm_start_errors": 32, + "index": 19 + } + ], + "on": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.3905931000008422, + "eval_s": 1.454786599999352, + "log_likelihood": 29034.99992341116, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21680879999985336, + "n": 1310, + "solve_s": 0.21680879999985336, + "outer_iterations": 29, + "inner_iterations": 12, + "total_iterations": 41, + "n_passive": 1262, + "warm_start_errors": 31, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.007633587786259542, + "reconstruction_s": 1.2773264000006748, + "eval_s": 1.3391339000008884, + "log_likelihood": 28849.582048876684, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06514679999963846, + "n": 1310, + "solve_s": 0.06514679999963846, + "outer_iterations": 5, + "inner_iterations": 1, + "total_iterations": 6, + "n_passive": 1260, + "warm_start_errors": 10, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 1.2601400000003196, + "eval_s": 1.3232484000000113, + "log_likelihood": 28670.060861127982, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0567854000000807, + "n": 1310, + "solve_s": 0.0567854000000807, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1257, + "warm_start_errors": 9, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.00916030534351145, + "reconstruction_s": 1.3391947000000073, + "eval_s": 1.407504799999515, + "log_likelihood": 28883.348745978772, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07601330000034068, + "n": 1310, + "solve_s": 0.07601330000034068, + "outer_iterations": 9, + "inner_iterations": 1, + "total_iterations": 10, + "n_passive": 1261, + "warm_start_errors": 12, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 1.4570968999996694, + "eval_s": 1.5196948000011616, + "log_likelihood": 28888.21930959083, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.051777900000161026, + "n": 1310, + "solve_s": 0.051777900000161026, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1261, + "warm_start_errors": 4, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.012977099236641221, + "reconstruction_s": 1.563853099998596, + "eval_s": 1.628988799999206, + "log_likelihood": 28682.76646409081, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.07506149999971967, + "n": 1310, + "solve_s": 0.07506149999971967, + "outer_iterations": 8, + "inner_iterations": 2, + "total_iterations": 10, + "n_passive": 1254, + "warm_start_errors": 17, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.012213740458015267, + "reconstruction_s": 1.7331817000012961, + "eval_s": 1.8074825999992754, + "log_likelihood": 29003.396039666695, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09033070000077714, + "n": 1310, + "solve_s": 0.09033070000077714, + "outer_iterations": 15, + "inner_iterations": 1, + "total_iterations": 16, + "n_passive": 1262, + "warm_start_errors": 16, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.013740458015267175, + "reconstruction_s": 1.2905093999997916, + "eval_s": 1.3510105999994266, + "log_likelihood": 29098.3542079982, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06461600000147882, + "n": 1310, + "solve_s": 0.06461600000147882, + "outer_iterations": 5, + "inner_iterations": 4, + "total_iterations": 9, + "n_passive": 1254, + "warm_start_errors": 18, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.010687022900763359, + "reconstruction_s": 1.3313641000004282, + "eval_s": 1.3873347000007925, + "log_likelihood": 29188.767818380573, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06873029999951541, + "n": 1310, + "solve_s": 0.06873029999951541, + "outer_iterations": 10, + "inner_iterations": 2, + "total_iterations": 12, + "n_passive": 1256, + "warm_start_errors": 14, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 1.4854787999993277, + "eval_s": 1.5546862999999576, + "log_likelihood": 28740.241007408153, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.049365999999281485, + "n": 1310, + "solve_s": 0.049365999999281485, + "outer_iterations": 3, + "inner_iterations": 1, + "total_iterations": 4, + "n_passive": 1254, + "warm_start_errors": 6, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.006870229007633588, + "reconstruction_s": 1.2366916999999376, + "eval_s": 1.2967733000004955, + "log_likelihood": 28659.77396052256, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05708079999931215, + "n": 1310, + "solve_s": 0.05708079999931215, + "outer_iterations": 6, + "inner_iterations": 1, + "total_iterations": 7, + "n_passive": 1255, + "warm_start_errors": 9, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.012213740458015267, + "reconstruction_s": 1.3360057000008965, + "eval_s": 1.4090427000010095, + "log_likelihood": 28569.89757533868, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.0769469000006211, + "n": 1310, + "solve_s": 0.0769469000006211, + "outer_iterations": 10, + "inner_iterations": 2, + "total_iterations": 12, + "n_passive": 1257, + "warm_start_errors": 16, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 1.3017135999998573, + "eval_s": 1.3696952000009333, + "log_likelihood": 28516.18652537265, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04552519999924698, + "n": 1310, + "solve_s": 0.04552519999924698, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1256, + "warm_start_errors": 3, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0030534351145038168, + "reconstruction_s": 1.267309299999397, + "eval_s": 1.327733599999192, + "log_likelihood": 28593.06738215032, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.03743480000048294, + "n": 1310, + "solve_s": 0.03743480000048294, + "outer_iterations": 2, + "inner_iterations": 1, + "total_iterations": 3, + "n_passive": 1256, + "warm_start_errors": 4, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.004580152671755725, + "reconstruction_s": 1.1838262000001123, + "eval_s": 1.2466734999998152, + "log_likelihood": 28494.225122309614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.05134600000019418, + "n": 1310, + "solve_s": 0.05134600000019418, + "outer_iterations": 4, + "inner_iterations": 1, + "total_iterations": 5, + "n_passive": 1256, + "warm_start_errors": 6, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 1.322736599999189, + "eval_s": 1.3897046000001865, + "log_likelihood": 28573.685210031195, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.16597489999912796, + "n": 1310, + "solve_s": 0.16597489999912796, + "outer_iterations": 21, + "inner_iterations": 2, + "total_iterations": 23, + "n_passive": 1254, + "warm_start_errors": 8, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0061068702290076335, + "reconstruction_s": 1.2501953999999387, + "eval_s": 1.3143815999992512, + "log_likelihood": 28754.16965277553, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.050590099999681115, + "n": 1310, + "solve_s": 0.050590099999681115, + "outer_iterations": 3, + "inner_iterations": 2, + "total_iterations": 5, + "n_passive": 1252, + "warm_start_errors": 8, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.01450381679389313, + "reconstruction_s": 1.350868100000298, + "eval_s": 1.4166492000003927, + "log_likelihood": 29100.48957849465, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.09507440000015777, + "n": 1310, + "solve_s": 0.09507440000015777, + "outer_iterations": 17, + "inner_iterations": 2, + "total_iterations": 19, + "n_passive": 1263, + "warm_start_errors": 19, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0022900763358778627, + "reconstruction_s": 1.4301498000004358, + "eval_s": 1.493118799999138, + "log_likelihood": 29261.487227838028, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.04593579999891517, + "n": 1310, + "solve_s": 0.04593579999891517, + "outer_iterations": 1, + "inner_iterations": 1, + "total_iterations": 2, + "n_passive": 1262, + "warm_start_errors": 3, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.009923664122137405, + "reconstruction_s": 1.2639178999997966, + "eval_s": 1.3247338000001037, + "log_likelihood": 29790.892417880037, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.06041419999928621, + "n": 1310, + "solve_s": 0.06041419999928621, + "outer_iterations": 7, + "inner_iterations": 1, + "total_iterations": 8, + "n_passive": 1263, + "warm_start_errors": 13, + "index": 19 + } + ] + } + }, + "iid": { + "summary": { + "off": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 0, + "median_warm_start_error_fraction": 0.0366412213740458, + "mean_warm_start_error_fraction": 0.037442748091603056, + "p90_warm_start_error_fraction": 0.04770992366412214, + "max_warm_start_error_fraction": 0.05267175572519084, + "median_outer_iterations": 31.0, + "mean_outer_iterations": 33.65, + "p90_outer_iterations": 51.300000000000004, + "max_outer_iterations": 55.0, + "median_inner_iterations": 12.5, + "mean_inner_iterations": 13.05, + "p90_inner_iterations": 17.0, + "max_inner_iterations": 17.0, + "median_total_iterations": 43.0, + "mean_total_iterations": 46.7, + "p90_total_iterations": 66.0, + "max_total_iterations": 71.0, + "median_n_passive": 1236.5, + "mean_n_passive": 1235.5, + "p90_n_passive": 1245.7, + "max_n_passive": 1255.0, + "median_warm_start_errors": 48.0, + "mean_warm_start_errors": 49.05, + "p90_warm_start_errors": 62.50000000000001, + "max_warm_start_errors": 69.0, + "median_solve_s": 0.24129214999993565, + "mean_solve_s": 0.253768239999863, + "p90_solve_s": 0.3080220199999531, + "max_solve_s": 0.39767010000105074, + "median_reconstruction_s": 1.5298511000000872, + "mean_reconstruction_s": 1.5437933749997683, + "p90_reconstruction_s": 1.707487570000012, + "max_reconstruction_s": 1.798733199999333, + "median_eval_s": 1.591511200000241, + "mean_eval_s": 1.605753034999816, + "p90_eval_s": 1.771086880000985, + "max_eval_s": 1.8645166999995126 + }, + "on": { + "n_evaluations": 20, + "retries": 0, + "n_fallbacks": 3, + "median_warm_start_error_fraction": 0.05076335877862595, + "mean_warm_start_error_fraction": 0.0502290076335878, + "p90_warm_start_error_fraction": 0.06374045801526718, + "max_warm_start_error_fraction": 0.07786259541984733, + "median_outer_iterations": 56.5, + "mean_outer_iterations": 52.05, + "p90_outer_iterations": 71.4, + "max_outer_iterations": 76.0, + "median_inner_iterations": 14.5, + "mean_inner_iterations": 13.8, + "p90_inner_iterations": 21.1, + "max_inner_iterations": 24.0, + "median_total_iterations": 69.5, + "mean_total_iterations": 65.85, + "p90_total_iterations": 88.10000000000002, + "max_total_iterations": 99.0, + "median_n_passive": 1236.0, + "mean_n_passive": 1235.3, + "p90_n_passive": 1245.7, + "max_n_passive": 1255.0, + "median_warm_start_errors": 66.5, + "mean_warm_start_errors": 65.8, + "p90_warm_start_errors": 83.5, + "max_warm_start_errors": 102.0, + "median_solve_s": 0.23680069999954867, + "mean_solve_s": 0.24643872999995436, + "p90_solve_s": 0.31772869000014914, + "max_solve_s": 0.3866522000007535, + "median_reconstruction_s": 1.370142499999929, + "mean_reconstruction_s": 1.4448884099997485, + "p90_reconstruction_s": 1.7333141999990405, + "max_reconstruction_s": 1.964120600001479, + "median_eval_s": 1.42807404999985, + "mean_eval_s": 1.5092889950000425, + "p90_eval_s": 1.8118448299999728, + "max_eval_s": 2.0764697000013257 + } + }, + "parity": { + "rtol": 1e-06, + "max_rel_log_likelihood_deviation": 9.538631459760166e-15, + "max_abs_reconstruction_deviation": 1.1574030622796272e-09 + }, + "iteration_reduction_ratio": 0.6187050359712231, + "solve_speedup_ratio": 1.0189672158925018, + "eval_speedup_ratio": 1.114445851040013, + "rows": { + "off": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 1.6824587999999494, + "eval_s": 1.7483351999999286, + "log_likelihood": 19610.36039125053, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3421736000000237, + "n": 1310, + "solve_s": 0.3421736000000237, + "outer_iterations": 55, + "inner_iterations": 11, + "total_iterations": 66, + "n_passive": 1228, + "warm_start_errors": 67, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.038931297709923665, + "reconstruction_s": 1.5338107999996282, + "eval_s": 1.5941278999998758, + "log_likelihood": 25182.415110795213, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24058660000082455, + "n": 1310, + "solve_s": 0.24058660000082455, + "outer_iterations": 32, + "inner_iterations": 15, + "total_iterations": 47, + "n_passive": 1226, + "warm_start_errors": 51, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02366412213740458, + "reconstruction_s": 1.5115141000005679, + "eval_s": 1.5709327000004123, + "log_likelihood": 28569.062384810884, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20387990000017453, + "n": 1310, + "solve_s": 0.20387990000017453, + "outer_iterations": 21, + "inner_iterations": 14, + "total_iterations": 35, + "n_passive": 1245, + "warm_start_errors": 31, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.04122137404580153, + "reconstruction_s": 1.6967656999986502, + "eval_s": 1.7599723000002996, + "log_likelihood": 25459.91825679345, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.28868299999885494, + "n": 1310, + "solve_s": 0.28868299999885494, + "outer_iterations": 30, + "inner_iterations": 10, + "total_iterations": 40, + "n_passive": 1237, + "warm_start_errors": 54, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.03206106870229008, + "reconstruction_s": 1.5258914000005461, + "eval_s": 1.588894500000606, + "log_likelihood": 27874.240680363288, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22619609999856038, + "n": 1310, + "solve_s": 0.22619609999856038, + "outer_iterations": 27, + "inner_iterations": 13, + "total_iterations": 40, + "n_passive": 1244, + "warm_start_errors": 42, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.035114503816793895, + "reconstruction_s": 1.5881952999989153, + "eval_s": 1.6549049999994168, + "log_likelihood": 27810.19866298291, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2622446999994281, + "n": 1310, + "solve_s": 0.2622446999994281, + "outer_iterations": 36, + "inner_iterations": 13, + "total_iterations": 49, + "n_passive": 1237, + "warm_start_errors": 46, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.04198473282442748, + "reconstruction_s": 1.4577226999990671, + "eval_s": 1.518062999999529, + "log_likelihood": 26505.940530222397, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3031943999994837, + "n": 1310, + "solve_s": 0.3031943999994837, + "outer_iterations": 49, + "inner_iterations": 17, + "total_iterations": 66, + "n_passive": 1226, + "warm_start_errors": 55, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.03435114503816794, + "reconstruction_s": 1.758975400000054, + "eval_s": 1.8294786999995267, + "log_likelihood": 24492.81466635593, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.29043020000062825, + "n": 1310, + "solve_s": 0.29043020000062825, + "outer_iterations": 20, + "inner_iterations": 11, + "total_iterations": 31, + "n_passive": 1224, + "warm_start_errors": 45, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0366412213740458, + "reconstruction_s": 1.798733199999333, + "eval_s": 1.8645166999995126, + "log_likelihood": 24952.954588361725, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.24199769999904674, + "n": 1310, + "solve_s": 0.24199769999904674, + "outer_iterations": 32, + "inner_iterations": 11, + "total_iterations": 43, + "n_passive": 1236, + "warm_start_errors": 48, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 1.3606325000000652, + "eval_s": 1.4167632999997295, + "log_likelihood": 17329.396957087614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2539459999989049, + "n": 1310, + "solve_s": 0.2539459999989049, + "outer_iterations": 51, + "inner_iterations": 12, + "total_iterations": 63, + "n_passive": 1230, + "warm_start_errors": 69, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.0366412213740458, + "reconstruction_s": 1.5359157000002597, + "eval_s": 1.5943244000009145, + "log_likelihood": 27029.986950519487, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21542109999973036, + "n": 1310, + "solve_s": 0.21542109999973036, + "outer_iterations": 27, + "inner_iterations": 11, + "total_iterations": 38, + "n_passive": 1243, + "warm_start_errors": 48, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.03740458015267176, + "reconstruction_s": 1.3361293999987538, + "eval_s": 1.3922490000004473, + "log_likelihood": 20357.33822918338, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2104390999993484, + "n": 1310, + "solve_s": 0.2104390999993484, + "outer_iterations": 31, + "inner_iterations": 11, + "total_iterations": 42, + "n_passive": 1230, + "warm_start_errors": 49, + "index": 11 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.04045801526717557, + "reconstruction_s": 1.651700800000981, + "eval_s": 1.707923899999514, + "log_likelihood": 24520.37486726214, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.269862799999828, + "n": 1310, + "solve_s": 0.269862799999828, + "outer_iterations": 34, + "inner_iterations": 16, + "total_iterations": 50, + "n_passive": 1209, + "warm_start_errors": 53, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.031297709923664124, + "reconstruction_s": 1.4604659999986325, + "eval_s": 1.5476170999991155, + "log_likelihood": 28035.381114657645, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21161799999936193, + "n": 1310, + "solve_s": 0.21161799999936193, + "outer_iterations": 31, + "inner_iterations": 16, + "total_iterations": 47, + "n_passive": 1233, + "warm_start_errors": 41, + "index": 13 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.03282442748091603, + "reconstruction_s": 1.501687199999651, + "eval_s": 1.5588775999985955, + "log_likelihood": 28358.559706509935, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20914760000050592, + "n": 1310, + "solve_s": 0.20914760000050592, + "outer_iterations": 26, + "inner_iterations": 12, + "total_iterations": 38, + "n_passive": 1242, + "warm_start_errors": 43, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 1.4636938000003283, + "eval_s": 1.5226103999993938, + "log_likelihood": 27753.91959514748, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2155218000007153, + "n": 1310, + "solve_s": 0.2155218000007153, + "outer_iterations": 26, + "inner_iterations": 17, + "total_iterations": 43, + "n_passive": 1242, + "warm_start_errors": 36, + "index": 15 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.031297709923664124, + "reconstruction_s": 1.2933374999993248, + "eval_s": 1.3522322999997414, + "log_likelihood": 28124.568319931564, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18581110000013723, + "n": 1310, + "solve_s": 0.18581110000013723, + "outer_iterations": 24, + "inner_iterations": 13, + "total_iterations": 37, + "n_passive": 1255, + "warm_start_errors": 41, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.029770992366412213, + "reconstruction_s": 1.3740166000006866, + "eval_s": 1.4335742999992362, + "log_likelihood": 30068.12156445181, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20231360000070708, + "n": 1310, + "solve_s": 0.20231360000070708, + "outer_iterations": 24, + "inner_iterations": 11, + "total_iterations": 35, + "n_passive": 1252, + "warm_start_errors": 39, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.04732824427480916, + "reconstruction_s": 1.7017667000000074, + "eval_s": 1.7645989000011468, + "log_likelihood": 25172.01783934843, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.39767010000105074, + "n": 1310, + "solve_s": 0.39767010000105074, + "outer_iterations": 54, + "inner_iterations": 17, + "total_iterations": 71, + "n_passive": 1240, + "warm_start_errors": 62, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.046564885496183206, + "reconstruction_s": 1.642453899999964, + "eval_s": 1.695063499999378, + "log_likelihood": 23342.637661178655, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3042273999999452, + "n": 1310, + "solve_s": 0.3042273999999452, + "outer_iterations": 43, + "inner_iterations": 10, + "total_iterations": 53, + "n_passive": 1231, + "warm_start_errors": 61, + "index": 19 + } + ], + "on": [ + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.05114503816793893, + "reconstruction_s": 1.488545800000793, + "eval_s": 1.5431580000004033, + "log_likelihood": 19610.36039125053, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2838327000008576, + "n": 1310, + "solve_s": 0.2838327000008576, + "outer_iterations": 55, + "inner_iterations": 11, + "total_iterations": 66, + "n_passive": 1228, + "warm_start_errors": 67, + "index": 0 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0633587786259542, + "reconstruction_s": 1.3219896999999037, + "eval_s": 1.3788411999994423, + "log_likelihood": 25182.41511079518, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2505703999995603, + "n": 1310, + "solve_s": 0.2505703999995603, + "outer_iterations": 58, + "inner_iterations": 15, + "total_iterations": 73, + "n_passive": 1225, + "warm_start_errors": 83, + "index": 1 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.061068702290076333, + "reconstruction_s": 1.331089699999211, + "eval_s": 1.3848213000001124, + "log_likelihood": 28569.06238481089, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2540102000002662, + "n": 1310, + "solve_s": 0.2540102000002662, + "outer_iterations": 75, + "inner_iterations": 24, + "total_iterations": 99, + "n_passive": 1245, + "warm_start_errors": 80, + "index": 2 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.050381679389312976, + "reconstruction_s": 1.271063399999548, + "eval_s": 1.332054299999072, + "log_likelihood": 25459.918256793477, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22550520000004326, + "n": 1310, + "solve_s": 0.22550520000004326, + "outer_iterations": 66, + "inner_iterations": 14, + "total_iterations": 80, + "n_passive": 1237, + "warm_start_errors": 66, + "index": 3 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.04351145038167939, + "reconstruction_s": 1.1870689999996102, + "eval_s": 1.2428818000007595, + "log_likelihood": 27874.240680363324, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.14374289999977918, + "n": 1310, + "solve_s": 0.14374289999977918, + "outer_iterations": 37, + "inner_iterations": 8, + "total_iterations": 45, + "n_passive": 1244, + "warm_start_errors": 57, + "index": 4 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 1.3632195999998657, + "eval_s": 1.4221340999993117, + "log_likelihood": 27810.19866298294, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.22439029999986815, + "n": 1310, + "solve_s": 0.22439029999986815, + "outer_iterations": 54, + "inner_iterations": 19, + "total_iterations": 73, + "n_passive": 1237, + "warm_start_errors": 59, + "index": 5 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06030534351145038, + "reconstruction_s": 1.3483595999987301, + "eval_s": 1.4050021000002744, + "log_likelihood": 26505.940530222408, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.23139809999884164, + "n": 1310, + "solve_s": 0.23139809999884164, + "outer_iterations": 66, + "inner_iterations": 12, + "total_iterations": 78, + "n_passive": 1226, + "warm_start_errors": 79, + "index": 6 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.06717557251908397, + "reconstruction_s": 1.3477485999992496, + "eval_s": 1.40608229999998, + "log_likelihood": 24492.81466635587, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21708010000111244, + "n": 1310, + "solve_s": 0.21708010000111244, + "outer_iterations": 61, + "inner_iterations": 16, + "total_iterations": 77, + "n_passive": 1224, + "warm_start_errors": 88, + "index": 7 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.05725190839694656, + "reconstruction_s": 1.3281181999991531, + "eval_s": 1.3858995999999024, + "log_likelihood": 24952.954588361656, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.21086669999931473, + "n": 1310, + "solve_s": 0.21086669999931473, + "outer_iterations": 58, + "inner_iterations": 7, + "total_iterations": 65, + "n_passive": 1235, + "warm_start_errors": 75, + "index": 8 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.05267175572519084, + "reconstruction_s": 1.4191890000001877, + "eval_s": 1.4819931000001816, + "log_likelihood": 17329.396957087614, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2422033000002557, + "n": 1310, + "solve_s": 0.2422033000002557, + "outer_iterations": 71, + "inner_iterations": 16, + "total_iterations": 87, + "n_passive": 1230, + "warm_start_errors": 69, + "index": 9 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.0450381679389313, + "reconstruction_s": 1.3604123999994044, + "eval_s": 1.4176196000007621, + "log_likelihood": 27029.986950519495, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.16975969999839435, + "n": 1310, + "solve_s": 0.16975969999839435, + "outer_iterations": 45, + "inner_iterations": 11, + "total_iterations": 56, + "n_passive": 1243, + "warm_start_errors": 59, + "index": 10 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.044274809160305344, + "reconstruction_s": 1.2657521999990422, + "eval_s": 1.327160200000435, + "log_likelihood": 20357.338229183373, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.18136720000075002, + "n": 1310, + "solve_s": 0.18136720000075002, + "outer_iterations": 36, + "inner_iterations": 6, + "total_iterations": 42, + "n_passive": 1229, + "warm_start_errors": 58, + "index": 11 + }, + { + "n_fallbacks": 1, + "seed_source": "memo", + "warm_start_error_fraction": 0.07786259541984733, + "reconstruction_s": 1.552074100000027, + "eval_s": 1.6205638000010367, + "log_likelihood": 24520.374867262166, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.287711599999966, + "n": 1310, + "solve_s": 0.287711599999966, + "outer_iterations": 63, + "inner_iterations": 21, + "total_iterations": 84, + "n_passive": 1209, + "warm_start_errors": 102, + "index": 12 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.031297709923664124, + "reconstruction_s": 1.4236409999994066, + "eval_s": 1.4797954999994545, + "log_likelihood": 28035.381114657645, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2138161000002583, + "n": 1310, + "solve_s": 0.2138161000002583, + "outer_iterations": 31, + "inner_iterations": 16, + "total_iterations": 47, + "n_passive": 1233, + "warm_start_errors": 41, + "index": 13 + }, + { + "n_fallbacks": 1, + "seed_source": "memo", + "warm_start_error_fraction": 0.06183206106870229, + "reconstruction_s": 1.48821130000033, + "eval_s": 1.5451111999991554, + "log_likelihood": 28358.559706510052, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2868297999993956, + "n": 1310, + "solve_s": 0.2868297999993956, + "outer_iterations": 66, + "inner_iterations": 17, + "total_iterations": 83, + "n_passive": 1242, + "warm_start_errors": 81, + "index": 14 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.02748091603053435, + "reconstruction_s": 1.3770653999999922, + "eval_s": 1.4340140000003885, + "log_likelihood": 27753.91959514748, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.20090779999918595, + "n": 1310, + "solve_s": 0.20090779999918595, + "outer_iterations": 26, + "inner_iterations": 17, + "total_iterations": 43, + "n_passive": 1242, + "warm_start_errors": 36, + "index": 15 + }, + { + "n_fallbacks": 1, + "seed_source": "memo", + "warm_start_error_fraction": 0.05419847328244275, + "reconstruction_s": 1.5289749999992637, + "eval_s": 1.596228999998857, + "log_likelihood": 28124.568319931575, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3512418999998772, + "n": 1310, + "solve_s": 0.3512418999998772, + "outer_iterations": 76, + "inner_iterations": 22, + "total_iterations": 98, + "n_passive": 1255, + "warm_start_errors": 71, + "index": 16 + }, + { + "n_fallbacks": 0, + "seed_source": "dense", + "warm_start_error_fraction": 0.029770992366412213, + "reconstruction_s": 1.7252522999988287, + "eval_s": 1.8015623999999661, + "log_likelihood": 30068.12156445181, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3866522000007535, + "n": 1310, + "solve_s": 0.3866522000007535, + "outer_iterations": 24, + "inner_iterations": 11, + "total_iterations": 35, + "n_passive": 1252, + "warm_start_errors": 39, + "index": 17 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.03969465648854962, + "reconstruction_s": 1.964120600001479, + "eval_s": 2.0764697000013257, + "log_likelihood": 25172.01783934819, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.3140050000001793, + "n": 1310, + "solve_s": 0.3140050000001793, + "outer_iterations": 42, + "inner_iterations": 7, + "total_iterations": 49, + "n_passive": 1240, + "warm_start_errors": 52, + "index": 18 + }, + { + "n_fallbacks": 0, + "seed_source": "memo", + "warm_start_error_fraction": 0.04122137404580153, + "reconstruction_s": 1.8058713000009448, + "eval_s": 1.9043867000000319, + "log_likelihood": 23342.63766117862, + "n_solves": 1, + "retries": 0, + "solve_s_total": 0.2528834000004281, + "n": 1310, + "solve_s": 0.2528834000004281, + "outer_iterations": 31, + "inner_iterations": 6, + "total_iterations": 37, + "n_passive": 1230, + "warm_start_errors": 54, + "index": 19 + } + ] + } + } + } +} \ No newline at end of file diff --git a/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_v2026.8.17.1.png b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_v2026.8.17.1.png new file mode 100644 index 000000000..eeb80aa89 Binary files /dev/null and b/results/breakdown/imaging/delaunay_numba_nnls_iterations_hst_v2026.8.17.1.png differ diff --git a/results/notes/nnls_warm_start_memo.md b/results/notes/nnls_warm_start_memo.md new file mode 100644 index 000000000..c371955a2 --- /dev/null +++ b/results/notes/nnls_warm_start_memo.md @@ -0,0 +1,90 @@ +# NNLS cross-evaluation warm-start memo: should it default on? (2026-08-28) + +Measurement leg of [PyAutoArray#498](https://github.com/PyAutoLabs/PyAutoArray/issues/498) +phase 3a — the default for `aa.Settings(nnls_warm_start_memo=...)`, which seeds the numba-CPU +fnnls active-set loop from the **previous evaluation's** final passive set instead of the sign +of the unconstrained dense solve. Script: +`scripts/imaging/likelihood_breakdown/delaunay_numba_nnls_iterations.py`; results: +`results/breakdown/imaging/delaunay_numba_nnls_iterations_{euclid,hst}_v2026.8.17.1.{json,png}`. +Laptop CPU fp64, `OMP_NUM_THREADS=1`, `AUTOARRAY_NUMBA_OPERATED_MEMO=0`, v2026.8.17.1, Delaunay +Hilbert-1250 + MGE-60 linear lens light, sparse-operator numba path, n = 1310. + +## What was measured + +The memo can only be judged across *different* parameter points — refitting one fixed instance +hands it a 100%-correct seed. Two 30-instance sequences, each run twice (memo OFF then ON, memo +cleared between): **random walk** (seeded, each step N(0, 1% of each parameter's prior width) — +the sampler-like regime) and **iid** (independent draws, unit values uniform in the central 20% +of every prior — the pessimistic, uncorrelated case). `fnnls_cholesky` is wrapped at module +level so every solve's `stats` is captured from inside a real `FitImaging`. Medians over 30. + +## Results + +| instrument | sequence | total iters OFF→ON | ws errors OFF→ON | solve s OFF→ON | eval s OFF→ON | iter ratio | +|---|---|---|---|---|---|---| +| euclid | random walk | 69.5 → **7.0** | 73.5 → 9.5 | 0.544 → **0.051** | 1.510 → **0.564** | **9.9×** | +| euclid | iid | 86.0 → 94.0 | 94.5 → 98.0 | 0.347 → 0.309 | 0.906 → 0.838 | 0.91× | +| hst | random walk | 32.0 → **8.0** | 31.0 → 9.5 | 0.245 → **0.067** | 1.733 → **1.594** | **4.0×** | +| hst | iid | 42.0 → 69.5 | 48.0 → 68.0 | 0.268 → 0.260 | 1.687 → 1.867 | 0.60× | + +**Parity.** Max relative Δlog-likelihood 2.98e-14 (gate rtol 1e-6); max |Δreconstruction| 1.6e-8. +**Retries.** 0 memo-seed failures in 240 evaluations — the fallback in +`reconstruction_positive_only_from` never fired. + +**Step-0 baseline** (`delaunay_numba.py`, same env, post-#453/#497): euclid reconstruction solve +0.470 s of 1.189 s step-total (40%); hst 0.542 s of 3.223 s (17%) — no longer the ~70% it was +when #498 was written (previously recorded euclid: 3.615 s of 4.839 s). Pins PASSED both. + +**Model robustness.** This measured ONE fiducial. The same A/B across 11 model variants and +both instruments is in [`nnls_warm_start_memo_matrix.md`](./nnls_warm_start_memo_matrix.md). + +## Verdict + +**Gate met: `nnls_warm_start_memo` defaults to `true`.** The random-walk median iteration +reduction is 9.9× (euclid) and 4.0× (hst), both ≥ 2×, with zero pin drift — the reconstruction is +identical to round-off and the pinned evidence values are unchanged. + +The **iid column** costs *iterations*, not time. Uncorrelated jumps make the seed a worse guess at +the passive set (0.91× euclid, 0.60× hst), but solve seconds are ≤ memo-off in all four +sequence×instrument cells: the memo also skips the dense unconstrained `np.linalg.solve` that +produces the sign start, and that saving covers the extra active-set moves. The one adverse +eval-level number, hst iid +0.18 s, cannot come out of a 0.26 s solve that got no slower — it is +noise in the other steps. So the default is `true` with no time penalty in the measured worst +case; iid stays the regime to revisit if a search class proves uncorrelated throughout. + +**Shipped default:** `nnls_warm_start_memo: true` in `autoarray/config/general.yaml`, with the +`KeyError` fallback in `Settings.nnls_warm_start_memo` returning `True` for workspaces whose own +`general.yaml` shadows autoarray's without the key. + +## Fallback guard + +`nnls_warm_start_error_tolerance` (config `inversion.nnls_warm_start_error_tolerance: 1.5`; +`inf`, `0` or a negative value disables it) makes the memo self-calibrating: each entry remembers +the error fraction of the most recent **dense-sign-started** solve for its key, and a seeded solve +whose own fraction exceeds `1.5 ×` that reference has its entry dropped — the next solve for that +key restarts dense and refreshes the reference. The rule is *relative* because the absolute +fraction does not separate helpful from harmful seeds (0.048–0.138 overlap) while the seed/dense +ratio does (helpful ≤ 0.89, worst 1.42), and 1.5 sits above that worst measured ratio, so the +guard is protective against unmeasured regimes rather than flapping inside the matrix. + +Re-run with the guard on (`--n-instances 20`, same env, so the fiducial JSONs/PNGs on disk are now +these 20-instance cells, not the 30-instance ones tabulated above): euclid fiducial 10.00×/0.91×, +hst fiducial 4.27×/**0.62×**, hst source_complex 3.39×/0.73× (rw/iid). It fired **3 times in 240 +evaluations**, all in `hst/fiducial/iid` — the one cell whose seed was worse than the dense-sign +start. No cell's ratio, solve seconds or parity moved beyond sampling noise (parity ≤ 9.5e-15). + +## Phase 3b (batched active-set moves) + +The residual iteration counts on the warm path are already 7–8 total (5–6 outer), so batching +outer active-set moves has almost nothing left to remove there. It retains value only for the +**cold / uncorrelated** path — 30–95 outer iterations on the first evaluation and the iid +sequence — which is also exactly the regime the memo does not help. That, not the warm path, is +3b's case. + +## Relation to the JAX solver ledger + +[`nnls_solver_ledger.md`](./nnls_solver_ledger.md) closed the *JAX PDIP* solver, where warm +starting from the unconstrained solve **hurt** (17 → 38 iterations) by un-centering the interior +point. That does not transfer: this is the numpy/numba Bro & De Jong active-set solver, where a +warm start is a discrete guess at the passive set and a wrong guess costs iterations rather than +degrading the iterate. The two solvers stay separately tuned. diff --git a/results/notes/nnls_warm_start_memo_matrix.md b/results/notes/nnls_warm_start_memo_matrix.md new file mode 100644 index 000000000..1d9b20f90 --- /dev/null +++ b/results/notes/nnls_warm_start_memo_matrix.md @@ -0,0 +1,78 @@ +# NNLS warm-start memo: model robustness matrix + +Aggregated from `results/breakdown/imaging/delaunay_numba_nnls_iterations_*_v2026.8.17.1.json` +by `scripts/imaging/likelihood_breakdown/nnls_iterations_matrix.py`. Companion to +[`nnls_warm_start_memo.md`](./nnls_warm_start_memo.md), which measured the single fiducial. + +Each model variant changes exactly ONE thing about that fiducial (Delaunay Hilbert-1250 + +ConstantSplit + MGE-60 linear lens light + Isothermal + shear), memo OFF then ON on the +*same* seeded instance sequences: `rw` = random walk (N(0, 1% of each prior's width) per +step, the sampler-like regime), `iid` = independent draws from the central 20% of every +prior (uncorrelated, pessimistic). Laptop CPU fp64, `OMP_NUM_THREADS=1`, +`AUTOARRAY_NUMBA_OPERATED_MEMO=0`. `ratio` = median iterations OFF / ON (>1 = memo saved +iterations). `err frac` = warm-start errors / solve size — the quantity a fallback +tolerance would threshold on. `Δlnl` = max relative log-likelihood deviation OFF vs ON. + +Three variants could not be a pure one-thing change, and say so here rather than in a +footnote: `rectangular` must drop `ConstantSplit` for `Constant` (split regularization is +structurally incompatible with the rectangular interpolator); `source_complex` ships as +data/noise/PSF with no truth tracer, so its adapt image is the positive-clipped data, and +it is sampled at 0.05"/px, i.e. hst geometry; `no_edge_zeroing` flips the solve onto the +full-system branch but does NOT change the index set — `Delaunay(zeroed_pixels=0)` already +keeps every index, and the only mesh here that zeroes any (`rectangular`, its shape-derived +136-pixel border) zeroes a *static* set. So the memo key never churns in this matrix. + +| instrument | model | seq | n | iters med OFF→ON | iters max OFF→ON | errs med OFF→ON | errs max ON | solve s med OFF→ON | solve s max ON | eval s med OFF→ON | ratio | err frac ON med/p90/max | err frac OFF med | Δlnl | retries | fallbacks | +|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---| +| euclid | fiducial | rw | 20 | 70→7 | 206→84 | 74→9 | 77 | 0.257→0.054 | 0.353 | 0.73→0.51 | 10.00× | 0.007/0.011/0.059 | 0.056 | 1.1e-14 | 0 | 0 | +| euclid | fiducial | iid | 20 | 86→94 | 134→146 | 96→96 | 139 | 0.392→0.269 | 0.495 | 0.99→0.78 | 0.91× | 0.073/0.098/0.106 | 0.073 | 8.6e-15 | 0 | 0 | +| euclid | powerlaw | rw | 20 | 60→8 | 72→72 | 69→10 | 75 | 0.281→0.051 | 0.354 | 0.87→0.56 | 7.93× | 0.007/0.021/0.057 | 0.053 | 2.6e-15 | 0 | – | +| euclid | powerlaw | iid | 20 | 92→85 | 129→117 | 108→96 | 122 | 0.426→0.278 | 0.482 | 0.99→0.81 | 1.09× | 0.073/0.088/0.093 | 0.082 | 7.8e-15 | 0 | – | +| euclid | subhalo | rw | 20 | 76→6 | 88→73 | 81→10 | 73 | 0.404→0.071 | 0.383 | 1.05→0.73 | 11.69× | 0.008/0.016/0.056 | 0.062 | 1.5e-14 | 0 | – | +| euclid | subhalo | iid | 20 | 88→94 | 123→140 | 98→95 | 128 | 0.580→0.370 | 0.610 | 1.41→0.98 | 0.93× | 0.073/0.096/0.098 | 0.074 | 6.4e-15 | 0 | – | +| euclid | no_lens_light | rw | 20 | 59→14 | 67→59 | 96→18 | 97 | 0.289→0.066 | 0.348 | 0.56→0.32 | 4.07× | 0.014/0.026/0.078 | 0.077 | 4.5e-16 | 0 | – | +| euclid | no_lens_light | iid | 20 | 53→101 | 69→160 | 91→117 | 161 | 0.300→0.306 | 0.580 | 0.56→0.58 | 0.52× | 0.094/0.120/0.129 | 0.073 | 2.2e-16 | 0 | – | +| euclid | sersic_light | rw | 20 | 42→5 | 48→37 | 154→6 | 154 | 0.315→0.053 | 0.284 | 0.63→0.33 | 8.50× | 0.005/0.013/0.123 | 0.123 | 3.1e-15 | 0 | – | +| euclid | sersic_light | iid | 20 | 44→41 | 64→63 | 157→62 | 152 | 0.369→0.223 | 0.441 | 0.69→0.65 | 1.06× | 0.049/0.068/0.122 | 0.125 | 4.1e-15 | 0 | – | +| euclid | rectangular | rw | 20 | 598→7 | 630→602 | 507→11 | 501 | 0.727→0.035 | 0.980 | 1.37→0.71 | 85.50× | 0.010/0.025/0.436 | 0.441 | 1.2e-15 | 0 | – | +| euclid | rectangular | iid | 20 | 620→161 | 677→619 | 502→158 | 519 | 0.558→0.233 | 0.480 | 1.13→0.85 | 3.85× | 0.138/0.158/0.452 | 0.436 | 6.7e-15 | 0 | – | +| euclid | adapt_reg | rw | 20 | 66→16 | 86→72 | 71→18 | 66 | 0.512→0.111 | 0.836 | 1.32→0.70 | 4.26× | 0.013/0.022/0.050 | 0.054 | 2.4e-14 | 0 | – | +| euclid | adapt_reg | iid | 20 | 70→70 | 99→165 | 75→65 | 105 | 0.352→0.270 | 0.637 | 1.00→0.85 | 1.00× | 0.050/0.067/0.080 | 0.057 | 4.3e-14 | 0 | – | +| euclid | no_edge_zeroing | rw | 20 | 70→7 | 206→84 | 74→9 | 77 | 0.359→0.057 | 0.593 | 0.99→0.63 | 10.00× | 0.007/0.011/0.059 | 0.056 | 1.1e-14 | 0 | – | +| euclid | no_edge_zeroing | iid | 20 | 86→94 | 134→146 | 96→96 | 139 | 0.523→0.324 | 0.825 | 1.26→0.87 | 0.91× | 0.073/0.098/0.106 | 0.073 | 8.6e-15 | 0 | – | +| euclid | mesh_600 | rw | 20 | 39→6 | 47→38 | 43→8 | 43 | 0.064→0.010 | 0.074 | 0.50→0.43 | 6.50× | 0.013/0.022/0.065 | 0.065 | 7.6e-16 | 0 | – | +| euclid | mesh_600 | iid | 20 | 49→56 | 66→79 | 54→54 | 74 | 0.065→0.050 | 0.122 | 0.50→0.51 | 0.88× | 0.082/0.106/0.112 | 0.082 | 5.3e-15 | 0 | – | +| euclid | mesh_2000 | rw | 20 | 96→13 | 134→85 | 117→17 | 110 | 1.376→0.304 | 1.347 | 2.61→1.30 | 7.38× | 0.008/0.013/0.053 | 0.057 | 7.7e-16 | 0 | – | +| euclid | mesh_2000 | iid | 20 | 114→128 | 179→200 | 132→133 | 184 | 1.601→1.208 | 1.899 | 3.10→2.29 | 0.88× | 0.065/0.083/0.089 | 0.064 | 1.3e-14 | 0 | – | +| hst | fiducial | rw | 20 | 32→8 | 41→41 | 31→10 | 31 | 0.205→0.063 | 0.217 | 1.61→1.39 | 4.27× | 0.007/0.014/0.024 | 0.024 | 1.3e-15 | 0 | 0 | +| hst | fiducial | iid | 20 | 43→70 | 71→99 | 48→66 | 102 | 0.241→0.237 | 0.387 | 1.59→1.43 | 0.62× | 0.051/0.064/0.078 | 0.037 | 9.5e-15 | 0 | 3 | +| hst | powerlaw | rw | 20 | 29→9 | 37→28 | 29→13 | 31 | 0.248→0.088 | 0.346 | 1.98→1.97 | 3.22× | 0.010/0.016/0.024 | 0.022 | 1.7e-15 | 0 | – | +| hst | powerlaw | iid | 20 | 47→62 | 69→115 | 48→66 | 90 | 0.357→0.301 | 0.552 | 2.30→2.03 | 0.75× | 0.050/0.063/0.069 | 0.037 | 2.7e-14 | 0 | – | +| hst | rectangular | rw | 20 | 886→8 | 906→881 | 802→13 | 783 | 1.212→0.040 | 0.905 | 3.17→1.89 | 118.07× | 0.011/0.021/0.681 | 0.698 | 2.9e-15 | 0 | – | +| hst | rectangular | iid | 20 | 872→172 | 993→854 | 763→99 | 732 | 1.090→0.327 | 1.307 | 3.10→2.03 | 5.09× | 0.086/0.111/0.637 | 0.664 | 9.9e-15 | 0 | – | +| hst | no_edge_zeroing | rw | 20 | 32→8 | 41→41 | 31→10 | 31 | 0.284→0.084 | 0.270 | 2.24→1.81 | 4.27× | 0.007/0.014/0.024 | 0.024 | 1.3e-15 | 0 | – | +| hst | no_edge_zeroing | iid | 20 | 43→73 | 71→99 | 48→68 | 102 | 0.342→0.341 | 0.515 | 2.16→2.32 | 0.59× | 0.052/0.064/0.078 | 0.037 | 1.1e-14 | 0 | – | +| hst | source_complex | rw | 20 | 322→95 | 361→311 | 492→94 | 478 | 0.384→0.090 | 0.404 | 1.25→0.87 | 3.39× | 0.072/0.108/0.365 | 0.375 | 4.1e-16 | 0 | 0 | +| hst | source_complex | iid | 20 | 344→475 | 427→605 | 514→393 | 526 | 0.356→0.372 | 0.531 | 1.16→1.21 | 0.73× | 0.300/0.343/0.402 | 0.393 | 2.2e-16 | 0 | 0 | +| hst | mesh_600 | rw | 20 | 24→3 | 29→21 | 21→4 | 20 | 0.059→0.012 | 0.048 | 1.84→1.91 | 8.00× | 0.005/0.017/0.030 | 0.032 | 1.0e-15 | 0 | – | +| hst | mesh_600 | iid | 20 | 30→36 | 51→54 | 34→32 | 47 | 0.069→0.041 | 0.064 | 1.88→1.79 | 0.85× | 0.048/0.067/0.071 | 0.052 | 6.4e-15 | 0 | – | + +## Findings + +- **Worst cell by iteration ratio:** `euclid/no_lens_light/iid` at 0.52× (53 → 101 median iterations). +- **Worst cell by MAX memo-ON iterations:** `hst/rectangular/rw` at 881 (memo OFF max on the same sequence: 906), and its worst evaluation is index 0. In 18/32 cells the memo-ON maximum IS evaluation 0, whose memo is empty and which therefore ran the dense-sign start — a cold start, not a pathology. The cells where it is not: `euclid/fiducial/iid` (idx 12), `euclid/powerlaw/iid` (idx 15), `euclid/subhalo/iid` (idx 13), `euclid/no_lens_light/iid` (idx 12), `euclid/sersic_light/iid` (idx 19), `euclid/adapt_reg/iid` (idx 3), `euclid/no_edge_zeroing/iid` (idx 12), `euclid/mesh_600/iid` (idx 12), `euclid/mesh_2000/iid` (idx 12), `hst/fiducial/iid` (idx 2), `hst/powerlaw/iid` (idx 15), `hst/no_edge_zeroing/iid` (idx 2), `hst/source_complex/iid` (idx 2), `hst/mesh_600/iid` (idx 16). +- **Cells where median memo-ON solve seconds EXCEED memo-OFF (2/32):** `hst/source_complex/iid` (0.356 → 0.372 s), `euclid/no_lens_light/iid` (0.300 → 0.306 s). +- **Retries** (a memo-seeded solve raised and the library fell back to the dense-sign start): none in any cell. +- **Tolerance fallbacks** (a memo seed's error fraction exceeded `nnls_warm_start_error_tolerance` × its entry's dense-sign reference, so the entry was dropped): `hst/fiducial/iid` (3) — of 6/32 cells recorded with the guard. +- **Parity:** no cell exceeds 1e-08; worst is `euclid/adapt_reg/iid` at max rel Δlnl 4.3e-14, max |Δreconstruction| 8.1e-10. + +### Fallback-tolerance calibration (memo-seed error fraction) + +- **random_walk** (16 cells): memo-ON error fraction median 0.005–0.072 (across-cell median 0.008), per-cell max up to 0.681; dense-sign (OFF) median 0.022–0.698. +- **iid** (16 cells): memo-ON error fraction median 0.048–0.300 (across-cell median 0.073), per-cell max up to 0.637; dense-sign (OFF) median 0.037–0.664. + +- **Separating X.** Helpful cells (ratio ≥ 1×): 21; non-helpful (< 1×): 11. The two populations OVERLAP: helpful cells reach a median memo-ON error fraction of 0.138 while the least-wrong non-helpful cell sits at 0.048. **No X separates them**, so a fallback tolerance cannot be calibrated from the error fraction alone. The highest cut that never discards a seed which was still saving iterations is X ≈ 0.14, but it fires on none of the non-helpful cells below it, i.e. the fallback would be inert where it is wanted. + +- **Relative seed quality (ON / OFF median error fraction).** The absolute fraction cannot separate the populations, but the ratio nearly does: helpful cells top out at 0.89, non-helpful cells start at 0.76, with the worst seed in the matrix at 1.42 (`hst/no_edge_zeroing/iid`). A fallback phrased as *"the memo seed is no better than the dense-sign start"* — i.e. **ON/OFF error fraction ≳ 0.9** — is therefore the discriminator the raw fraction is not. Its cost is that computing the OFF fraction means computing the dense-sign start, which is the work the memo exists to skip. + +- **Observability caveat.** The error count is only known once the solve has run, so a fallback cannot threshold on the current evaluation's fraction. Any X would have to be applied to the *previous* evaluation's fraction as a running proxy — which is exactly the quantity the random-walk/iid contrast above shows is stable within a sequence and unstable across regimes. diff --git a/results/simulators/imaging_euclid_summary_v2026.8.17.1.json b/results/simulators/imaging_euclid_summary_v2026.8.17.1.json index 5d7169905..7404ae046 100644 --- a/results/simulators/imaging_euclid_summary_v2026.8.17.1.json +++ b/results/simulators/imaging_euclid_summary_v2026.8.17.1.json @@ -21,25 +21,25 @@ ] }, "phases": { - "setup_grids": 0.04633339999963937, - "setup_psf_simulator": 0.00036360000012791716, - "setup_galaxies": 7.870000081311446e-05, - "setup_tracer": 2.9499999072868377e-05, - "image_2d_eager": 0.26651510000010603, - "image_2d_jit_lower": 0.19283700000050885, - "image_2d_jit_compile": 0.46145789999900444, - "image_2d_jit_first_call": 0.00813890000063111, - "image_2d_jit_steady_x10": 0.03916080000090005, - "via_tracer_from": 0.7143261000001075, - "solver_build": 4.3799998820759356e-05, - "solver_solve_eager": 0.038935500000661705, - "output_fits": 0.18606509999881382, - "output_lensed_source": 0.024783499999102787, - "output_json": 0.0016039999991335208 + "setup_grids": 0.09266010000010283, + "setup_psf_simulator": 0.0006021999997756211, + "setup_galaxies": 0.00012140000035287812, + "setup_tracer": 4.600000011123484e-05, + "image_2d_eager": 0.45647769999959564, + "image_2d_jit_lower": 0.3479296999998951, + "image_2d_jit_compile": 0.6907042000002548, + "image_2d_jit_first_call": 0.009016799999699288, + "image_2d_jit_steady_x10": 0.05976509999982227, + "via_tracer_from": 1.308894500000406, + "solver_build": 7.689999983995222e-05, + "solver_solve_eager": 0.0863312000001315, + "output_fits": 0.3481563999998798, + "output_lensed_source": 0.06316220000007888, + "output_json": 0.003867800000080024 }, "key_timings": { - "image_2d_eager_s": 0.26651510000010603, - "via_tracer_from_s": 0.7143261000001075, - "solver_solve_eager_s": 0.038935500000661705 + "image_2d_eager_s": 0.45647769999959564, + "via_tracer_from_s": 1.308894500000406, + "solver_solve_eager_s": 0.0863312000001315 } } \ No newline at end of file diff --git a/results/simulators/imaging_euclid_summary_v2026.8.17.1.png b/results/simulators/imaging_euclid_summary_v2026.8.17.1.png index e62bc2a9d..3039cf643 100644 Binary files a/results/simulators/imaging_euclid_summary_v2026.8.17.1.png and b/results/simulators/imaging_euclid_summary_v2026.8.17.1.png differ diff --git a/scripts/imaging/likelihood_breakdown/delaunay_numba_nnls_iterations.py b/scripts/imaging/likelihood_breakdown/delaunay_numba_nnls_iterations.py new file mode 100644 index 000000000..6b85a54e6 --- /dev/null +++ b/scripts/imaging/likelihood_breakdown/delaunay_numba_nnls_iterations.py @@ -0,0 +1,879 @@ +""" +Numba CPU Profiling: Delaunay NNLS Active-Set Iterations (Warm-Start Memo A/B) +============================================================================= + +Measures what the cross-evaluation NNLS warm-start memo +(``aa.Settings(nnls_warm_start_memo=...)``, PyAutoArray#498) actually buys on the +production Delaunay numba-CPU fiducial, so the default can be chosen from numbers +rather than from the plausibility of the idea. + +The positive-only reconstruction is ~70% of a numba CPU Delaunay likelihood +evaluation at Euclid resolution. Its Bro & De Jong active-set loop is warm-started +from a guess at the passive set; production guesses it from the sign of the +unconstrained dense solve, and the memo instead reuses the *previous evaluation's* +final passive set. + +Method +------ +The memo only pays off across *different* parameter points — re-fitting one fixed +instance would hand the memo a 100%-correct seed and fake the answer. So two +30-instance sequences are built and each is run **twice**, memo OFF then memo ON: + +* ``random_walk`` — a seeded Gaussian random walk from the prior medians, each step + perturbing every parameter by N(0, 1% of that parameter's prior width). This is + the sampler-like regime the memo is designed for. +* ``iid`` — 30 independent draws with unit values uniform in [0.4, 0.6] (the central + 20% of every prior). Neighbouring evaluations are unrelated: the pessimistic case. + +``autoarray.util.fnnls.fnnls_cholesky`` is wrapped at module level (the caller +imports it at call time) so every solve's ``stats`` dict — outer/inner active-set +iterations, final passive-set size, and how many entries the warm start got wrong — +is captured from inside a real ``FitImaging``. A memo-seeded solve that raises is +retried by the library from the dense-sign start; both attempts are recorded and the +retries are counted and reported. + +Parity is asserted per instance: the NNLS optimum is unique, so the memo must not +move the log likelihood (rtol 1e-6) or the reconstruction. + +Model robustness matrix +----------------------- +One fiducial cannot certify the memo: a warm start is a *discrete guess at the +passive set*, so its quality is a property of the solution's structure, and a +different mass model / lens light / mesh / regularization is a different structure. +``--model `` selects one of ``MODEL_VARIANTS`` — each changes exactly ONE +thing about the fiducial, so a pathological cell is attributable. ``--n-instances`` +overrides the sequence length (30 for the fiducial, 20 for the matrix). + +``nnls_iterations_matrix.py`` aggregates every JSON written here into +``results/notes/nnls_warm_start_memo_matrix.md``. + +Output +------ +``results/breakdown/imaging/delaunay_numba_nnls_iterations_[_]_v.{json,png}`` +(``--model fiducial`` keeps the original, unsuffixed filename.) +""" + +import sys as _sys +from pathlib import Path as _Path + + +def _profiling_root() -> _Path: + for _p in _Path(__file__).resolve().parents: + if (_p / "ruff.toml").exists(): + return _p + raise RuntimeError("autolens_profiling root (ruff.toml) not found") + + +_misc_dir = str(_profiling_root() / "scripts" / "misc") +if _misc_dir not in _sys.path: + _sys.path.insert(0, _misc_dir) + +_sys.path.insert(0, str(_profiling_root())) + +import argparse +import json +import os + +# AUTOLENS_PROFILING_SMOKE=1 short-circuit (CI lint smoke). +import os as _smoke_os +import sys as _smoke_sys +import time +from pathlib import Path + +import autofit as af +import autolens as al +import numpy as np + +if _smoke_os.environ.get("AUTOLENS_PROFILING_SMOKE") == "1": + print(f"[smoke] {__file__}: imports + module setup OK; exiting.") + _smoke_sys.exit(0) + +from simulators.imaging import INSTRUMENTS # noqa: E402 + +from _adapt_image_util import adapt_image_for_dataset # noqa: E402 +from _profile_cli import ( # noqa: E402 + auto_simulate_if_missing, + device_info_dict, + parse_profile_cli, + resolve_output_paths, +) + +# =================================================================== +# Model variants — the robustness matrix +# =================================================================== +# +# Each entry changes exactly ONE thing about the fiducial. Empty dict = fiducial. +# +# mass — "power_law" swaps Isothermal for PowerLaw (+ slope prior) +# subhalo — adds an NFWMCRLudlowSph subhalo galaxy +# lens_light — "sersic" | None (default "mge_60") +# mesh — "rectangular_uniform_35" (default: Delaunay + Hilbert) +# regularization — "adapt_split" (default: ConstantSplit) +# use_edge_zeroed_pixels — False exercises the NON-subset solve index space +# dataset — a dataset directory that is not an instrument preset +# pixels — Hilbert/Delaunay vertex count (default 1250) + +MODEL_VARIANTS: dict[str, dict] = { + "fiducial": {}, + "powerlaw": {"mass": "power_law"}, + "subhalo": {"subhalo": True}, + "no_lens_light": {"lens_light": None}, + "sersic_light": {"lens_light": "sersic"}, + "rectangular": {"mesh": "rectangular_uniform_35"}, + "adapt_reg": {"regularization": "adapt_split"}, + "no_edge_zeroing": {"use_edge_zeroed_pixels": False}, + # Its own dataset, not an instrument preset: a different source morphology is a + # different *type* of NNLS solution. Sampled at 0.05"/px, i.e. hst geometry. + "source_complex": {"dataset": "source_complex", "instrument": "hst"}, + "mesh_600": {"pixels": 600}, + "mesh_2000": {"pixels": 2000}, +} + +# ``parse_profile_cli`` uses ``parse_known_args``, so these extra flags are parsed +# here rather than pushed into the shared helper every other profiling cell imports. +_model_parser = argparse.ArgumentParser(add_help=False) +_model_parser.add_argument("--model", default="fiducial", choices=sorted(MODEL_VARIANTS)) +_model_parser.add_argument("--n-instances", type=int, default=30) +_model_args, _ = _model_parser.parse_known_args() + +MODEL = _model_args.model +VARIANT = MODEL_VARIANTS[MODEL] + +_cli = parse_profile_cli() + +# A variant that ships its own dataset pins the instrument geometry it was sampled at. +instrument = VARIANT.get("instrument") or _cli.instrument or "euclid" + +N_INSTANCES = _model_args.n_instances +RANDOM_WALK_STEP_FRACTION = 0.01 # step sigma as a fraction of each prior's width +IID_UNIT_LOW, IID_UNIT_HIGH = 0.4, 0.6 # central 20% of every prior +SEED = 498 # PyAutoArray#498 +PARITY_RTOL = 1.0e-6 + +# =================================================================== +# Setup — identical fiducial to delaunay_numba.py (the breakdown sibling) +# =================================================================== + +print(f"\n--- Dataset loading & masking [{instrument} | model '{MODEL}'] ---") + +_workspace_root = _profiling_root() +pixel_scale = INSTRUMENTS[instrument]["pixel_scale"] +dataset_name = VARIANT.get("dataset", instrument) +dataset_path = Path("dataset") / "imaging" / dataset_name + +if dataset_name == instrument: + auto_simulate_if_missing( + dataset_path, + dataset_type="imaging", + instrument=instrument, + workspace_root=_workspace_root, + ) +elif not (dataset_path / "data.fits").exists(): + raise FileNotFoundError( + f"Model '{MODEL}' needs the curated dataset at {dataset_path}, which has no " + f"simulator preset in INSTRUMENTS and is not present. Run from the " + f"autolens_profiling root, or drop the variant." + ) + +dataset = al.Imaging.from_fits( + data_path=dataset_path / "data.fits", + psf_path=dataset_path / "psf.fits", + noise_map_path=dataset_path / "noise_map.fits", + pixel_scales=pixel_scale, +) + +mask_radius = 3.5 + +mask = al.Mask2D.circular( + shape_native=dataset.shape_native, + pixel_scales=dataset.pixel_scales, + radius=mask_radius, +) + +dataset = dataset.apply_mask(mask=mask) + +over_sample_size = al.util.over_sample.over_sample_size_via_radial_bins_from( + grid=dataset.grid, + sub_size_list=[4, 2, 1], + radial_list=[0.3, 0.6], + centre_list=[(0.0, 0.0)], +) + +dataset = dataset.apply_over_sampling( + over_sample_size_lp=over_sample_size, + over_sample_size_pixelization=1, +) + +dataset = dataset.apply_sparse_operator_cpu() + +print("\n--- Adapt image + image mesh (one-off) ---") + +n_mesh_vertices = VARIANT.get("pixels", 1250) +RECTANGULAR_SHAPE = (35, 35) # 1225 pixels — the Delaunay-1250 fiducial's nearest square +use_rectangular = VARIANT.get("mesh") == "rectangular_uniform_35" +needs_adapt_data = VARIANT.get("regularization") == "adapt_split" + +if (dataset_path / "lensed_source.fits").exists() or (dataset_path / "tracer.json").exists(): + adapt_image = adapt_image_for_dataset(dataset_path=dataset_path, dataset=dataset) + adapt_image_source = "lensed_source_from_truth_tracer" +else: + # `source_complex` ships as data/noise/psf only — there is no truth tracer to + # derive the lensed source from. The positive-clipped data is the standard + # stand-in, and is what an adapt pass starts from before any fit exists. + adapt_image = al.Array2D(values=np.clip(np.asarray(dataset.data), 0.0, None), mask=dataset.mask) + adapt_image_source = "positive_clipped_data" + +if use_rectangular: + # RectangularUniform lays its own uniform grid over the traced source plane, so + # there is no image-plane mesh grid to precompute or hand through AdaptImages. + image_plane_mesh_grid = None +else: + image_mesh = al.image_mesh.Hilbert(pixels=n_mesh_vertices, weight_power=1.0, weight_floor=0.0) + image_plane_mesh_grid = image_mesh.image_plane_mesh_grid_from( + mask=dataset.mask, adapt_data=adapt_image + ) + +print("\n--- Model construction ---") + +lens_light_kind = VARIANT.get("lens_light", "mge_60") +_lens_light_ell = al.convert.ell_comps_from(axis_ratio=0.9, angle=45.0) + +if lens_light_kind == "mge_60": + lens_bulge = al.model_util.mge_model_from( + mask_radius=mask_radius, + total_gaussians=60, + centre_prior_is_uniform=True, + ) +elif lens_light_kind == "sersic": + # Gaussian priors throughout: the iid sequence draws the central 20% of every + # prior, so a wide uniform prior would draw physically silly light profiles and + # measure the solver on garbage rather than on a different lens light. + lens_bulge = af.Model(al.lp_linear.Sersic) + lens_bulge.centre.centre_0 = af.GaussianPrior(mean=0.0, sigma=0.05) + lens_bulge.centre.centre_1 = af.GaussianPrior(mean=0.0, sigma=0.05) + lens_bulge.ell_comps.ell_comps_0 = af.GaussianPrior(mean=_lens_light_ell[0], sigma=0.02) + lens_bulge.ell_comps.ell_comps_1 = af.GaussianPrior(mean=_lens_light_ell[1], sigma=0.02) + lens_bulge.effective_radius = af.GaussianPrior(mean=1.0, sigma=0.2) + lens_bulge.sersic_index = af.GaussianPrior(mean=4.0, sigma=0.5) +else: + lens_bulge = None + +if VARIANT.get("mass") == "power_law": + mass = af.Model(al.mp.PowerLaw) + mass.slope = af.GaussianPrior(mean=2.0, sigma=0.05) +else: + mass = af.Model(al.mp.Isothermal) +mass.centre.centre_0 = af.GaussianPrior(mean=0.0, sigma=0.005) +mass.centre.centre_1 = af.GaussianPrior(mean=0.0, sigma=0.005) +mass.einstein_radius = af.GaussianPrior(mean=1.6, sigma=0.05) +_lens_mass_ell = al.convert.ell_comps_from(axis_ratio=0.9, angle=45.0) +mass.ell_comps.ell_comps_0 = af.GaussianPrior(mean=_lens_mass_ell[0], sigma=0.01) +mass.ell_comps.ell_comps_1 = af.GaussianPrior(mean=_lens_mass_ell[1], sigma=0.01) + +shear = af.Model(al.mp.ExternalShear) +shear.gamma_1 = af.GaussianPrior(mean=0.05, sigma=0.005) +shear.gamma_2 = af.GaussianPrior(mean=0.05, sigma=0.005) + +lens = af.Model(al.Galaxy, redshift=0.5, bulge=lens_bulge, mass=mass, shear=shear) + +galaxy_models = {"lens": lens} + +if VARIANT.get("subhalo"): + subhalo_mass = af.Model(al.mp.NFWMCRLudlowSph) + subhalo_mass.centre.centre_0 = af.GaussianPrior(mean=0.8, sigma=0.05) + subhalo_mass.centre.centre_1 = af.GaussianPrior(mean=0.8, sigma=0.05) + subhalo_mass.mass_at_200 = af.LogUniformPrior(lower_limit=1.0e9, upper_limit=1.0e10) + subhalo_mass.redshift_object = 0.5 + subhalo_mass.redshift_source = 1.0 + galaxy_models["subhalo"] = af.Model(al.Galaxy, redshift=0.5, mass=subhalo_mass) + +if use_rectangular: + # ConstantSplit is structurally incompatible with the rectangular meshes (their + # interpolator supplies no split mappings), so the rectangular cell uses the + # unsplit Constant — the one forced deviation from "change only the named thing". + mesh = al.mesh.RectangularUniform(shape=RECTANGULAR_SHAPE) + regularization = al.reg.Constant(coefficient=1.0) + mesh_label = f"rectangular_uniform_{RECTANGULAR_SHAPE[0]}x{RECTANGULAR_SHAPE[1]}" + regularization_label = "constant" +else: + mesh = al.mesh.Delaunay(pixels=n_mesh_vertices, zeroed_pixels=0) + mesh_label = f"delaunay_hilbert_{n_mesh_vertices}" + if needs_adapt_data: + regularization = al.reg.AdaptSplit( + inner_coefficient=0.1, outer_coefficient=10.0, signal_scale=0.1 + ) + regularization_label = "adapt_split" + else: + regularization = al.reg.ConstantSplit(coefficient=1.0) + regularization_label = "constant_split" + +pixelization = al.Pixelization(mesh=mesh, regularization=regularization) + +galaxy_models["source"] = af.Model(al.Galaxy, redshift=1.0, pixelization=pixelization) + +model = af.Collection(galaxies=af.Collection(**galaxy_models)) + +# The one Settings deviation a variant may make; every cell otherwise shares these. +SETTINGS_KWARGS: dict = {"use_border_relocator": True} +if "use_edge_zeroed_pixels" in VARIANT: + SETTINGS_KWARGS["use_edge_zeroed_pixels"] = VARIANT["use_edge_zeroed_pixels"] + +# The relative fallback guard the memo-ON cells run under: a seeded solve whose error +# fraction exceeds TOLERANCE x its entry's dense-sign reference has its entry dropped, +# so the next solve for that key restarts dense. Read from the settings actually used +# rather than hard-coded, so the matrix records the value it measured under. +TOLERANCE = float(al.Settings(**SETTINGS_KWARGS).nnls_warm_start_error_tolerance) +TOLERANCE_LABEL = f"{TOLERANCE:g}" + +# =================================================================== +# Instance sequences +# =================================================================== + + +def prior_step_sigmas() -> np.ndarray: + """ + Per-parameter random-walk step sigma: ``RANDOM_WALK_STEP_FRACTION`` of each + prior's width, where "width" is a Gaussian prior's sigma and a uniform prior's + ``upper - lower``. + """ + sigmas = [] + for prior in model.priors_ordered_by_id: + sigma = getattr(prior, "sigma", None) + if sigma is None: + sigma = float(prior.upper_limit) - float(prior.lower_limit) + sigmas.append(RANDOM_WALK_STEP_FRACTION * float(sigma)) + return np.asarray(sigmas) + + +def prior_support() -> tuple[np.ndarray, np.ndarray]: + """Per-parameter (lower, upper) physical limits, used to keep the walk in support.""" + lower = np.asarray([float(p.lower_limit) for p in model.priors_ordered_by_id]) + upper = np.asarray([float(p.upper_limit) for p in model.priors_ordered_by_id]) + # Inset finite limits so a step that lands exactly on a bound is still interior. + span = np.where(np.isfinite(upper - lower), (upper - lower), 0.0) + return lower + 1.0e-6 * span, upper - 1.0e-6 * span + + +def random_walk_instances(n: int) -> list: + """ + ``n`` instances along a seeded Gaussian random walk from the prior medians. + + This is the regime the memo targets: successive sampler evaluations sit close + together, so their passive sets overlap almost entirely. + """ + rng = np.random.default_rng(SEED) + step_sigmas = prior_step_sigmas() + lower, upper = prior_support() + + vector = np.asarray(model.physical_values_from_prior_medians, dtype=float) + + instances = [] + for _ in range(n): + vector = np.clip(vector + rng.normal(0.0, step_sigmas), lower, upper) + instances.append(model.instance_from_vector(vector=list(vector))) + return instances + + +def iid_instances(n: int) -> list: + """ + ``n`` independent instances with unit values uniform in the central 20% of every + prior — the pessimistic case, where neighbouring evaluations are unrelated. + """ + rng = np.random.default_rng(SEED + 1) + return [ + model.instance_from_unit_vector( + unit_vector=list(rng.uniform(IID_UNIT_LOW, IID_UNIT_HIGH, size=model.prior_count)) + ) + for _ in range(n) + ] + + +def adapt_images_for(instance) -> al.AdaptImages: + """ + Rebuild the adapt images for ``instance``: the dicts are keyed on the instance's + own source galaxy object, so they cannot be shared between instances. + + A rectangular mesh needs no image-plane mesh grid; an adaptive regularization + additionally needs the source's adapt image. + """ + kwargs: dict = {} + if image_plane_mesh_grid is not None: + kwargs["galaxy_image_plane_mesh_grid_dict"] = { + instance.galaxies.source: image_plane_mesh_grid + } + kwargs["galaxy_name_image_plane_mesh_grid_dict"] = { + "('galaxies', 'source')": image_plane_mesh_grid + } + if needs_adapt_data: + kwargs["galaxy_image_dict"] = {instance.galaxies.source: adapt_image} + kwargs["galaxy_name_image_dict"] = {"('galaxies', 'source')": adapt_image} + return al.AdaptImages(**kwargs) + + +# =================================================================== +# Solver-stat capture +# =================================================================== + +import autoarray.util.fnnls as fnnls_mod # noqa: E402 +from autoarray.inversion.inversion import nnls_memo # noqa: E402 + +_orig_fnnls_cholesky = fnnls_mod.fnnls_cholesky + +_solve_records: list[dict] = [] +_recording = [False] + + +def _wrapped_fnnls_cholesky(ZTZ, ZTx, P_initial=np.zeros(0, dtype=int), stats=None): + """ + Record every fnnls solve's diagnostics. ``reconstruction_positive_only_from`` + resolves ``fnnls_cholesky`` by ``from ... import`` at call time, so replacing the + module attribute is enough to see solves from inside a real fit. + + A raising solve is recorded too — that is exactly the memo-seed retry the library + catches, and it must be counted, not silently dropped. + + The ``stats`` dict is held BY REFERENCE rather than copied out here, because + ``reconstruction_positive_only_from`` writes ``seed_source`` and + ``warm_start_fallback`` into it *after* ``fnnls_cholesky`` returns. Reading them at + this point would always see them missing; they are read once the evaluation is over. + """ + stats = {} if stats is None else stats + start = time.perf_counter() + try: + reconstruction = _orig_fnnls_cholesky(ZTZ, ZTx, P_initial, stats=stats) + except BaseException as e: + if _recording[0]: + _solve_records.append( + { + "n": int(np.shape(ZTZ)[0]), + "solve_s": time.perf_counter() - start, + "failed": type(e).__name__, + } + ) + raise + elapsed = time.perf_counter() - start + if _recording[0]: + _solve_records.append( + { + "n": int(np.shape(ZTZ)[0]), + "solve_s": elapsed, + "failed": None, + "outer_iterations": int(stats["outer_iterations"]), + "inner_iterations": int(stats["inner_iterations"]), + "total_iterations": int(stats["outer_iterations"] + stats["inner_iterations"]), + "n_passive": int(stats["n_passive"]), + "warm_start_errors": int(stats["warm_start_errors"]), + "stats": stats, + } + ) + return reconstruction + + +fnnls_mod.fnnls_cholesky = _wrapped_fnnls_cholesky + + +def one_evaluation(instance, memo_on: bool) -> dict: + """One full likelihood evaluation, with its solve diagnostics and timings.""" + _solve_records.clear() + _recording[0] = True + + start = time.perf_counter() + fit = al.FitImaging( + dataset=dataset, + tracer=al.Tracer(galaxies=list(instance.galaxies)), + adapt_images=adapt_images_for(instance), + settings=al.Settings(**SETTINGS_KWARGS, nnls_warm_start_memo=memo_on), + xp=np, + ) + recon_start = time.perf_counter() + reconstruction = np.asarray(fit.inversion.reconstruction) + reconstruction_s = time.perf_counter() - recon_start + figure_of_merit = float(fit.figure_of_merit) + eval_s = time.perf_counter() - start + + _recording[0] = False + solves = list(_solve_records) + + successful = [s for s in solves if s["failed"] is None] + if not successful: + raise RuntimeError("No fnnls solve was recorded for this evaluation.") + + # The final successful solve is the one that produced the reconstruction; any + # earlier one is a memo-seed attempt the library retried. + primary = successful[-1] + + # `seed_source` / `warm_start_fallback` are set by + # `reconstruction_positive_only_from` after `fnnls_cholesky` returns, so they are + # only readable now. A seeded attempt and its dense retry SHARE one stats dict, so + # distinct dicts are counted -- one per positive-only solve the evaluation ran. + unique_stats = {id(s["stats"]): s["stats"] for s in solves if "stats" in s} + n_fallbacks = sum(1 for st in unique_stats.values() if st.get("warm_start_fallback", False)) + seed_source = primary["stats"].get("seed_source", "unknown") + + return { + # How many of this evaluation's solves discarded their memo seed as worse than + # `nnls_warm_start_error_tolerance` x the entry's dense-sign reference. + "n_fallbacks": n_fallbacks, + "seed_source": seed_source, + # errors / solve size — the quantity a fallback tolerance would threshold on. + "warm_start_error_fraction": primary["warm_start_errors"] / max(primary["n"], 1), + "reconstruction_s": reconstruction_s, + "eval_s": eval_s, + "log_likelihood": figure_of_merit, + "n_solves": len(solves), + "retries": len(solves) - 1, + "solve_s_total": float(sum(s["solve_s"] for s in solves)), + "reconstruction": reconstruction, + **{ + k: primary[k] + for k in ( + "n", + "solve_s", + "outer_iterations", + "inner_iterations", + "total_iterations", + "n_passive", + "warm_start_errors", + ) + }, + } + + +def run_sequence(instances: list, memo_on: bool) -> list[dict]: + """Run one instance sequence end-to-end with the memo in one state.""" + nnls_memo._nnls_passive_set_memo.clear() + rows = [] + for index, instance in enumerate(instances): + row = one_evaluation(instance, memo_on=memo_on) + row["index"] = index + rows.append(row) + nnls_memo._nnls_passive_set_memo.clear() + return rows + + +# =================================================================== +# Warm-up + configuration report +# =================================================================== + +from autoarray.inversion.inversion.imaging_numba.sparse import ( # noqa: E402 + InversionImagingSparseNumba, +) + +_median_instance = model.instance_from_vector(vector=model.physical_values_from_prior_medians) + +n_image_pixels = dataset.data.shape[0] +n_over_sampled_pixels = dataset.grids.lp.over_sampled.shape[0] + +print("\n--- Configuration (determines run time) ---") +print(f" Model variant: {MODEL}") +print(f" Instrument: {instrument}") +print(f" Dataset: {dataset_name} (adapt image: {adapt_image_source})") +print(f" Pixel scale: {pixel_scale} arcsec/pixel") +print(f" Mask radius: {mask_radius} arcsec") +print(f" Image pixels (masked): {n_image_pixels}") +print(f" Over-sampled pixels: {n_over_sampled_pixels}") +print(f" Mesh: {mesh_label}") +print(f" Regularization: {regularization_label}") +print(f" Lens light: {lens_light_kind or 'none'}") +print(f" Settings overrides: {SETTINGS_KWARGS}") +print(f" Warm-start tolerance: {TOLERANCE_LABEL}x the dense-sign reference") +print(f" Free parameters: {model.prior_count}") +print(f" Instances per sequence: {N_INSTANCES}") +print(f" OMP_NUM_THREADS: {os.environ.get('OMP_NUM_THREADS', '(unset)')}") +print(f" Operated-matrix memo: {os.environ.get('AUTOARRAY_NUMBA_OPERATED_MEMO', '(unset)')}") + +print("\n--- Warm-up evaluation (numba compile) ---") +_warm_start = time.perf_counter() +_warm = one_evaluation(_median_instance, memo_on=False) +warmup_s = time.perf_counter() - _warm_start +nnls_memo._nnls_passive_set_memo.clear() +print(f" warm-up total: {warmup_s:.4f} s (log_likelihood = {_warm['log_likelihood']})") +print( + f" solve: n = {_warm['n']}, outer = {_warm['outer_iterations']}, " + f"inner = {_warm['inner_iterations']}, n_passive = {_warm['n_passive']}, " + f"warm_start_errors = {_warm['warm_start_errors']}" +) + +_check_fit = al.FitImaging( + dataset=dataset, + tracer=al.Tracer(galaxies=list(_median_instance.galaxies)), + adapt_images=adapt_images_for(_median_instance), + settings=al.Settings(**SETTINGS_KWARGS), + xp=np, +) +assert isinstance(_check_fit.inversion, InversionImagingSparseNumba), ( + f"Expected InversionImagingSparseNumba, got {type(_check_fit.inversion).__name__}" +) +del _check_fit + +# =================================================================== +# The A/B runs +# =================================================================== + +SEQUENCES = { + "random_walk": random_walk_instances(N_INSTANCES), + "iid": iid_instances(N_INSTANCES), +} + +results: dict[str, dict] = {} +parity_violations: list[str] = [] + +for sequence_name, instances in SEQUENCES.items(): + print(f"\n--- Sequence '{sequence_name}' ({N_INSTANCES} instances) ---") + + cells = {} + for memo_on in (False, True): + label = "on" if memo_on else "off" + start = time.perf_counter() + cells[label] = run_sequence(instances, memo_on=memo_on) + print(f" memo {label.upper():<3} done in {time.perf_counter() - start:.1f} s") + + # --- parity: the NNLS optimum is unique, so the memo must change nothing --- + max_rel_log_likelihood = 0.0 + max_abs_reconstruction = 0.0 + for row_off, row_on in zip(cells["off"], cells["on"]): + denominator = max(abs(row_off["log_likelihood"]), 1e-300) + max_rel_log_likelihood = max( + max_rel_log_likelihood, + abs(row_on["log_likelihood"] - row_off["log_likelihood"]) / denominator, + ) + max_abs_reconstruction = max( + max_abs_reconstruction, + float(np.max(np.abs(row_on["reconstruction"] - row_off["reconstruction"]))), + ) + if not np.isclose( + row_on["log_likelihood"], row_off["log_likelihood"], rtol=PARITY_RTOL, atol=0.0 + ): + # Recorded rather than asserted: a parity break is the headline finding of + # this matrix, and aborting here would throw away the cell that found it. + # The script still exits non-zero at the end (see the guard after the chart). + message = ( + f"[{MODEL}/{instrument}/{sequence_name}] memo parity FAILED at instance " + f"{row_off['index']}: off = {row_off['log_likelihood']!r}, " + f"on = {row_on['log_likelihood']!r}" + ) + print(f" WARNING: {message}") + parity_violations.append(message) + + results[sequence_name] = { + "cells": cells, + "parity": { + "rtol": PARITY_RTOL, + "max_rel_log_likelihood_deviation": max_rel_log_likelihood, + "max_abs_reconstruction_deviation": max_abs_reconstruction, + }, + } + print( + f" parity: max rel Δlog_likelihood = {max_rel_log_likelihood:.3e}, " + f"max |Δreconstruction| = {max_abs_reconstruction:.3e}" + ) + +# =================================================================== +# Summary +# =================================================================== + +METRICS = ( + "warm_start_error_fraction", + "outer_iterations", + "inner_iterations", + "total_iterations", + "n_passive", + "warm_start_errors", + "solve_s", + "reconstruction_s", + "eval_s", +) + + +def summarize(rows: list[dict]) -> dict: + summary = { + "n_evaluations": len(rows), + "retries": int(sum(r["retries"] for r in rows)), + "n_fallbacks": int(sum(r["n_fallbacks"] for r in rows)), + } + for metric in METRICS: + values = np.asarray([r[metric] for r in rows], dtype=float) + summary[f"median_{metric}"] = float(np.median(values)) + summary[f"mean_{metric}"] = float(np.mean(values)) + # A median hides the pathological evaluation the matrix is hunting for. + summary[f"p90_{metric}"] = float(np.percentile(values, 90)) + summary[f"max_{metric}"] = float(np.max(values)) + return summary + + +for sequence_name, block in results.items(): + block["summary"] = {label: summarize(rows) for label, rows in block["cells"].items()} + median_off = block["summary"]["off"]["median_total_iterations"] + median_on = block["summary"]["on"]["median_total_iterations"] + block["iteration_reduction_ratio"] = ( + float(median_off / median_on) if median_on else float("inf") + ) + block["solve_speedup_ratio"] = float( + block["summary"]["off"]["median_solve_s"] / block["summary"]["on"]["median_solve_s"] + ) + block["eval_speedup_ratio"] = float( + block["summary"]["off"]["median_eval_s"] / block["summary"]["on"]["median_eval_s"] + ) + +al_version = al.__version__ +total_retries = int( + sum(b["summary"][k]["retries"] for b in results.values() for k in ("off", "on")) +) +total_fallbacks = int( + sum(b["summary"][k]["n_fallbacks"] for b in results.values() for k in ("off", "on")) +) + +print("\n" + "=" * 96) +print(f"NNLS WARM-START MEMO A/B — {instrument.upper()} — model '{MODEL}' — v{al_version}") +print("=" * 96) +header = ( + f" {'sequence':<12} {'memo':<5} {'outer':>7} {'inner':>7} {'total':>7} " + f"{'ws_err':>7} {'n_pass':>7} {'solve s':>10} {'recon s':>10} {'eval s':>10} " + f"{'retry':>6} {'fallbk':>7}" +) +print(header) +print(" " + "-" * (len(header) - 2)) +for sequence_name, block in results.items(): + for label in ("off", "on"): + s = block["summary"][label] + print( + f" {sequence_name:<12} {label.upper():<5} " + f"{s['median_outer_iterations']:>7.0f} {s['median_inner_iterations']:>7.0f} " + f"{s['median_total_iterations']:>7.0f} {s['median_warm_start_errors']:>7.0f} " + f"{s['median_n_passive']:>7.0f} {s['median_solve_s']:>10.4f} " + f"{s['median_reconstruction_s']:>10.4f} {s['median_eval_s']:>10.4f} " + f"{s['retries']:>6d} {s['n_fallbacks']:>7d}" + ) + print( + f" {'':<12} {'->':<5} iteration reduction " + f"{block['iteration_reduction_ratio']:.2f}x | solve " + f"{block['solve_speedup_ratio']:.2f}x | eval " + f"{block['eval_speedup_ratio']:.2f}x | parity max rel Δlnl " + f"{block['parity']['max_rel_log_likelihood_deviation']:.2e}" + ) +print("=" * 96) +print( + f" Median values over {N_INSTANCES} evaluations per cell. Total memo-seed retries: " + f"{total_retries}. Total tolerance fallbacks (seed dropped as worse than " + f"{TOLERANCE_LABEL} x the dense-sign reference): {total_fallbacks}" +) + +summary_dict = { + "autolens_version": al_version, + "device": device_info_dict(), + "instrument": instrument, + "model": MODEL, + "configuration": { + "model": MODEL, + "dataset": dataset_name, + "adapt_image_source": adapt_image_source, + "settings_overrides": {k: v for k, v in SETTINGS_KWARGS.items()}, + "pixel_scale_arcsec": pixel_scale, + "mask_radius_arcsec": mask_radius, + "image_pixels_masked": int(n_image_pixels), + "over_sampled_pixels": int(n_over_sampled_pixels), + "source_pixels": int(n_mesh_vertices), + "free_parameters": int(model.prior_count), + "inversion_path": "sparse_numba", + "use_jax": False, + "mesh": mesh_label, + "regularization": regularization_label, + "lens_light": lens_light_kind or "none", + "n_instances_per_sequence": N_INSTANCES, + "random_walk_step_fraction_of_prior_width": RANDOM_WALK_STEP_FRACTION, + "iid_unit_range": [IID_UNIT_LOW, IID_UNIT_HIGH], + "seed": SEED, + "omp_num_threads": os.environ.get("OMP_NUM_THREADS", None), + "autoarray_numba_operated_memo": os.environ.get("AUTOARRAY_NUMBA_OPERATED_MEMO", None), + }, + "warmup_incl_numba_compile_s": warmup_s, + "total_memo_seed_retries": total_retries, + "total_warm_start_fallbacks": total_fallbacks, + "nnls_warm_start_error_tolerance": TOLERANCE, + "parity_violations": parity_violations, + "sequences": { + sequence_name: { + "summary": block["summary"], + "parity": block["parity"], + "iteration_reduction_ratio": block["iteration_reduction_ratio"], + "solve_speedup_ratio": block["solve_speedup_ratio"], + "eval_speedup_ratio": block["eval_speedup_ratio"], + "rows": { + label: [{k: v for k, v in row.items() if k != "reconstruction"} for row in rows] + for label, rows in block["cells"].items() + }, + } + for sequence_name, block in results.items() + }, +} + +# `fiducial` keeps the original unsuffixed filename so the recorded results stay valid. +_model_suffix = "" if MODEL == "fiducial" else f"_{MODEL}" + +dict_path, chart_path = resolve_output_paths( + _cli, + default_dir=_workspace_root / "results" / "breakdown" / "imaging", + default_basename=(f"delaunay_numba_nnls_iterations_{instrument}{_model_suffix}_v{al_version}"), +) +dict_path.write_text(json.dumps(summary_dict, indent=2)) +print(f"\n Results dict saved to: {dict_path}") + +# =================================================================== +# Chart +# =================================================================== + +import matplotlib # noqa: E402 + +matplotlib.use("Agg") +import matplotlib.pyplot as plt # noqa: E402 + +sequence_names = list(results.keys()) +fig, axes = plt.subplots(2, len(sequence_names), figsize=(12, 7), sharex="col") + +COLORS = {"off": "#C44E52", "on": "#4C72B0"} + +for column, sequence_name in enumerate(sequence_names): + block = results[sequence_name] + for row_index, (metric, ylabel) in enumerate( + (("total_iterations", "Active-set iterations"), ("solve_s", "fnnls solve (s)")) + ): + ax = axes[row_index, column] + for label in ("off", "on"): + rows = block["cells"][label] + ax.plot( + [r["index"] for r in rows], + [r[metric] for r in rows], + marker="o", + markersize=3, + linewidth=1.3, + color=COLORS[label], + label=f"memo {label.upper()}", + ) + ax.set_ylabel(ylabel, fontsize=10) + ax.grid(alpha=0.3) + if row_index == 0: + ax.set_title( + f"{sequence_name} — median iteration ratio OFF/ON = " + f"{block['iteration_reduction_ratio']:.2f}x", + fontsize=10, + ) + ax.legend(fontsize=9) + else: + ax.set_xlabel("Evaluation index", fontsize=10) + +fig.suptitle( + f"NNLS Warm-Start Memo A/B — numba CPU — {instrument.upper()} — " + f"model '{MODEL}' (v{al_version})", + fontsize=12, + fontweight="bold", +) +fig.tight_layout() +fig.savefig(chart_path, dpi=150) +plt.close(fig) +print(f" Chart saved to: {chart_path}") + +if parity_violations: + print(f"\n PARITY FAILED in {len(parity_violations)} evaluation(s):") + for violation in parity_violations: + print(f" {violation}") + _sys.exit(1) diff --git a/scripts/imaging/likelihood_breakdown/nnls_iterations_matrix.py b/scripts/imaging/likelihood_breakdown/nnls_iterations_matrix.py new file mode 100644 index 000000000..3be401ced --- /dev/null +++ b/scripts/imaging/likelihood_breakdown/nnls_iterations_matrix.py @@ -0,0 +1,499 @@ +""" +Aggregator: NNLS Warm-Start Memo Robustness Matrix +================================================== + +Reads every ``delaunay_numba_nnls_iterations_*_v.json`` written by +``delaunay_numba_nnls_iterations.py --model `` and renders the one table the +robustness question needs: for each instrument x model x sequence, memo OFF vs ON on +the *same* instance sequence. + +Everything is recomputed from the per-evaluation ``rows`` each result JSON already +carries, never from its pre-baked ``summary`` block — so a result recorded before the +matrix existed (the 30-instance fiducial) is aggregated on exactly the same footing as +a 20-instance matrix cell. + +Why a matrix at all: an NNLS warm start is a *discrete guess at the passive set*, so its +quality is a property of the solution's structure. A different mass model, lens light, +mesh, regularization or source morphology is a different structure, and the fiducial +alone cannot tell you whether the memo's seed is ever *worse* than the dense-sign start +it replaces. + +Usage +----- +``python nnls_iterations_matrix.py [--version 2026.8.17.1] [--results-dir ...] [--out ...]`` + +Writes ``results/notes/nnls_warm_start_memo_matrix.md`` and prints the same table. +""" + +import argparse +import json +import os +import re +import sys +from pathlib import Path + +import numpy as np + +if os.environ.get("AUTOLENS_PROFILING_SMOKE") == "1": + print(f"[smoke] {__file__}: imports + module setup OK; exiting.") + sys.exit(0) + + +def _profiling_root() -> Path: + for parent in Path(__file__).resolve().parents: + if (parent / "ruff.toml").exists(): + return parent + raise RuntimeError("autolens_profiling root (ruff.toml) not found") + + +ROOT = _profiling_root() + +# Display order: fiducial first, then the variants in the order they escalate away +# from it (mass, structure, light, mesh, regularization, index space, data, size). +MODEL_ORDER = [ + "fiducial", + "powerlaw", + "subhalo", + "no_lens_light", + "sersic_light", + "rectangular", + "adapt_reg", + "no_edge_zeroing", + "source_complex", + "mesh_600", + "mesh_2000", +] +INSTRUMENT_ORDER = ["euclid", "hst"] +SEQUENCE_ORDER = ["random_walk", "iid"] +SEQUENCE_LABEL = {"random_walk": "rw", "iid": "iid"} + +FILENAME_RE = re.compile(r"^delaunay_numba_nnls_iterations_.*_v(?P[0-9][0-9.]*)\.json$") + +# A memo-ON cell whose seed is this much worse than the dense-sign start is the thing +# the matrix is hunting for; a parity deviation above this is a correctness break. +PARITY_ALERT = 1.0e-8 + + +def cell_stats(rows: list[dict]) -> dict: + """Per-cell statistics, recomputed from the per-evaluation rows.""" + + def column(key: str) -> np.ndarray: + return np.asarray([row[key] for row in rows], dtype=float) + + iterations = column("total_iterations") + errors = column("warm_start_errors") + solve_s = column("solve_s") + eval_s = column("eval_s") + # Recomputed rather than read: the pre-matrix fiducial JSONs predate the field. + error_fraction = errors / np.maximum(column("n"), 1.0) + + return { + "n_evaluations": len(rows), + # Index 0 of a memo-ON sequence has an empty memo and therefore falls back to the + # dense-sign start, so a max that sits there is the cold start, not a pathology. + "argmax_iterations": int(np.argmax(iterations)), + "median_iterations": float(np.median(iterations)), + "max_iterations": float(np.max(iterations)), + "median_errors": float(np.median(errors)), + "max_errors": float(np.max(errors)), + "median_solve_s": float(np.median(solve_s)), + "max_solve_s": float(np.max(solve_s)), + "median_eval_s": float(np.median(eval_s)), + "median_error_fraction": float(np.median(error_fraction)), + "p90_error_fraction": float(np.percentile(error_fraction, 90)), + "max_error_fraction": float(np.max(error_fraction)), + "retries": int(sum(row["retries"] for row in rows)), + # None (rendered as an en dash) for results recorded before the relative + # fallback guard existed, which is not the same statement as "zero fallbacks". + "n_fallbacks": ( + int(sum(row["n_fallbacks"] for row in rows)) + if all("n_fallbacks" in row for row in rows) + else None + ), + } + + +def load_cells(results_dir: Path, version: str | None) -> tuple[list[dict], str]: + """Every (instrument, model, sequence) cell found on disk, plus the version used.""" + candidates = {} + for path in sorted(results_dir.glob("delaunay_numba_nnls_iterations_*.json")): + match = FILENAME_RE.match(path.name) + if match: + candidates.setdefault(match.group("version"), []).append(path) + + if not candidates: + raise FileNotFoundError(f"No delaunay_numba_nnls_iterations_*_v*.json under {results_dir}") + + if version is None: + # Newest version present, compared as a release tuple rather than as a string + # (so 2026.8.9 does not sort above 2026.8.17). + version = max(candidates, key=lambda v: tuple(int(part) for part in v.split("."))) + + if version not in candidates: + raise FileNotFoundError( + f"No results for v{version} under {results_dir}; found {sorted(candidates)}" + ) + + cells = [] + for path in candidates[version]: + data = json.loads(path.read_text()) + model = data.get("model") or data.get("configuration", {}).get("model") or "fiducial" + for sequence_name, block in data["sequences"].items(): + cells.append( + { + "instrument": data["instrument"], + "model": model, + "sequence": sequence_name, + "parity_log_likelihood": block["parity"]["max_rel_log_likelihood_deviation"], + "parity_reconstruction": block["parity"]["max_abs_reconstruction_deviation"], + "off": cell_stats(block["rows"]["off"]), + "on": cell_stats(block["rows"]["on"]), + "path": path, + } + ) + + def sort_key(cell: dict) -> tuple: + return ( + INSTRUMENT_ORDER.index(cell["instrument"]) + if cell["instrument"] in INSTRUMENT_ORDER + else len(INSTRUMENT_ORDER), + MODEL_ORDER.index(cell["model"]) if cell["model"] in MODEL_ORDER else len(MODEL_ORDER), + SEQUENCE_ORDER.index(cell["sequence"]) + if cell["sequence"] in SEQUENCE_ORDER + else len(SEQUENCE_ORDER), + ) + + return sorted(cells, key=sort_key), version + + +def ratio(cell: dict) -> float: + """Median iteration reduction OFF/ON. >1 means the memo saved iterations.""" + median_on = cell["on"]["median_iterations"] + return cell["off"]["median_iterations"] / median_on if median_on else float("inf") + + +COLUMNS = [ + "instrument", + "model", + "seq", + "n", + "iters med OFF→ON", + "iters max OFF→ON", + "errs med OFF→ON", + "errs max ON", + "solve s med OFF→ON", + "solve s max ON", + "eval s med OFF→ON", + "ratio", + "err frac ON med/p90/max", + "err frac OFF med", + "Δlnl", + "retries", + "fallbacks", +] + + +def fallback_label(cell: dict) -> str: + """Tolerance fallbacks in the cell, or an en dash for a pre-guard result JSON.""" + counts = [cell[label]["n_fallbacks"] for label in ("off", "on")] + if any(count is None for count in counts): + return "–" + return str(sum(counts)) + + +def table_rows(cells: list[dict]) -> list[list[str]]: + rows = [] + for cell in cells: + off, on = cell["off"], cell["on"] + rows.append( + [ + cell["instrument"], + cell["model"], + SEQUENCE_LABEL.get(cell["sequence"], cell["sequence"]), + str(off["n_evaluations"]), + f"{off['median_iterations']:.0f}→{on['median_iterations']:.0f}", + f"{off['max_iterations']:.0f}→{on['max_iterations']:.0f}", + f"{off['median_errors']:.0f}→{on['median_errors']:.0f}", + f"{on['max_errors']:.0f}", + f"{off['median_solve_s']:.3f}→{on['median_solve_s']:.3f}", + f"{on['max_solve_s']:.3f}", + f"{off['median_eval_s']:.2f}→{on['median_eval_s']:.2f}", + f"{ratio(cell):.2f}×", + f"{on['median_error_fraction']:.3f}/{on['p90_error_fraction']:.3f}/" + f"{on['max_error_fraction']:.3f}", + f"{off['median_error_fraction']:.3f}", + f"{cell['parity_log_likelihood']:.1e}", + str(off["retries"] + on["retries"]), + fallback_label(cell), + ] + ) + return rows + + +def markdown_table(cells: list[dict]) -> list[str]: + lines = ["| " + " | ".join(COLUMNS) + " |", "|" + "---|" * len(COLUMNS)] + lines += ["| " + " | ".join(row) + " |" for row in table_rows(cells)] + return lines + + +def name(cell: dict) -> str: + return f"{cell['instrument']}/{cell['model']}/{SEQUENCE_LABEL.get(cell['sequence'])}" + + +def findings(cells: list[dict]) -> list[str]: + """The five failure modes the matrix exists to detect, plus the fallback calibration.""" + lines = ["## Findings", ""] + + worst_ratio = min(cells, key=ratio) + lines.append( + f"- **Worst cell by iteration ratio:** `{name(worst_ratio)}` at " + f"{ratio(worst_ratio):.2f}× " + f"({worst_ratio['off']['median_iterations']:.0f} → " + f"{worst_ratio['on']['median_iterations']:.0f} median iterations)." + ) + + worst_max = max(cells, key=lambda c: c["on"]["max_iterations"]) + cold = [c for c in cells if c["on"]["argmax_iterations"] == 0] + lines.append( + f"- **Worst cell by MAX memo-ON iterations:** `{name(worst_max)}` at " + f"{worst_max['on']['max_iterations']:.0f} " + f"(memo OFF max on the same sequence: {worst_max['off']['max_iterations']:.0f}), " + f"and its worst evaluation is index {worst_max['on']['argmax_iterations']}. In " + f"{len(cold)}/{len(cells)} cells the memo-ON maximum IS evaluation 0, whose memo is " + f"empty and which therefore ran the dense-sign start — a cold start, not a " + f"pathology. The cells where it is not: " + + ", ".join( + f"`{name(c)}` (idx {c['on']['argmax_iterations']})" + for c in cells + if c["on"]["argmax_iterations"] != 0 + ) + + "." + ) + + slower = [c for c in cells if c["on"]["median_solve_s"] > c["off"]["median_solve_s"]] + if slower: + lines.append( + "- **Cells where median memo-ON solve seconds EXCEED memo-OFF " + f"({len(slower)}/{len(cells)}):** " + + ", ".join( + f"`{name(c)}` ({c['off']['median_solve_s']:.3f} → " + f"{c['on']['median_solve_s']:.3f} s)" + for c in sorted( + slower, + key=lambda c: c["on"]["median_solve_s"] - c["off"]["median_solve_s"], + reverse=True, + )[:6] + ) + + "." + ) + else: + lines.append( + "- **Median memo-ON solve seconds never exceed memo-OFF** in any cell — the " + "skipped dense unconstrained solve covers every extra active-set move." + ) + + retried = [c for c in cells if c["off"]["retries"] + c["on"]["retries"]] + if retried: + detail = ( + ", ".join(f"`{name(c)}` ({c['off']['retries'] + c['on']['retries']})" for c in retried) + + "." + ) + else: + detail = "none in any cell." + lines.append( + "- **Retries** (a memo-seeded solve raised and the library fell back to the " + f"dense-sign start): {detail}" + ) + + guarded = [c for c in cells if fallback_label(c) != "–"] + fired = [c for c in guarded if fallback_label(c) != "0"] + if not guarded: + detail = "not recorded — every result JSON here predates the guard." + elif fired: + detail = ( + ", ".join(f"`{name(c)}` ({fallback_label(c)})" for c in fired) + + f" — of {len(guarded)}/{len(cells)} cells recorded with the guard." + ) + else: + detail = ( + f"never fired in any of the {len(guarded)}/{len(cells)} cells recorded with " + "the guard, which is the intent: the default tolerance sits above the worst " + "seed/dense ratio the matrix measured, so it is protective rather than " + "flapping." + ) + lines.append( + "- **Tolerance fallbacks** (a memo seed's error fraction exceeded " + "`nnls_warm_start_error_tolerance` × its entry's dense-sign reference, so the " + f"entry was dropped): {detail}" + ) + + broken = [c for c in cells if c["parity_log_likelihood"] > PARITY_ALERT] + max_parity = max(cells, key=lambda c: c["parity_log_likelihood"]) + if broken: + detail = ( + "BROKEN in " + + ", ".join(f"`{name(c)}` ({c['parity_log_likelihood']:.1e})" for c in broken) + + "." + ) + else: + detail = ( + f"no cell exceeds {PARITY_ALERT:g}; worst is `{name(max_parity)}` at max rel " + f"Δlnl {max_parity['parity_log_likelihood']:.1e}, max |Δreconstruction| " + f"{max_parity['parity_reconstruction']:.1e}." + ) + lines.append(f"- **Parity:** {detail}") + + lines += ["", "### Fallback-tolerance calibration (memo-seed error fraction)", ""] + + for sequence_name in SEQUENCE_ORDER: + group = [c for c in cells if c["sequence"] == sequence_name] + if not group: + continue + on_median = np.asarray([c["on"]["median_error_fraction"] for c in group]) + on_max = np.asarray([c["on"]["max_error_fraction"] for c in group]) + off_median = np.asarray([c["off"]["median_error_fraction"] for c in group]) + lines.append( + f"- **{sequence_name}** ({len(group)} cells): memo-ON error fraction median " + f"{on_median.min():.3f}–{on_median.max():.3f} (across-cell median " + f"{np.median(on_median):.3f}), per-cell max up to {on_max.max():.3f}; " + f"dense-sign (OFF) median {off_median.min():.3f}–{off_median.max():.3f}." + ) + + helpful = [c for c in cells if ratio(c) >= 1.0] + harmful = [c for c in cells if ratio(c) < 1.0] + if helpful and harmful: + helpful_ceiling = max(c["on"]["median_error_fraction"] for c in helpful) + harmful_floor = min(c["on"]["median_error_fraction"] for c in harmful) + if helpful_ceiling < harmful_floor: + threshold = 0.5 * (helpful_ceiling + harmful_floor) + verdict = ( + f"The two populations SEPARATE: every cell where the seed still saves " + f"iterations has a median memo-ON error fraction ≤ {helpful_ceiling:.3f}, " + f"and every cell where it does not has ≥ {harmful_floor:.3f}. A fallback " + f"tolerance of **X ≈ {threshold:.2f}** ({100 * threshold:.0f}% of entries " + f"wrong) separates them across this matrix." + ) + else: + threshold = helpful_ceiling + verdict = ( + f"The two populations OVERLAP: helpful cells reach a median memo-ON error " + f"fraction of {helpful_ceiling:.3f} while the least-wrong non-helpful cell " + f"sits at {harmful_floor:.3f}. **No X separates them**, so a fallback " + f"tolerance cannot be calibrated from the error fraction alone. The highest " + f"cut that never discards a seed which was still saving iterations is " + f"X ≈ {threshold:.2f}, but it fires on none of the non-helpful cells below " + f"it, i.e. the fallback would be inert where it is wanted." + ) + lines += [ + "", + f"- **Separating X.** Helpful cells (ratio ≥ 1×): {len(helpful)}; " + f"non-helpful (< 1×): {len(harmful)}. {verdict}", + ] + else: + lines += [ + "", + "- **Separating X.** Every cell falls on one side of ratio = 1×, so this " + "matrix cannot calibrate a fallback tolerance from a contrast.", + ] + + # The absolute fraction cannot separate the populations, but the seed's quality + # *relative to the start it replaces* very nearly can — and "no better than the + # dense-sign start" is a relative statement to begin with. + def seed_quality(cell: dict) -> float: + off_fraction = cell["off"]["median_error_fraction"] + return cell["on"]["median_error_fraction"] / off_fraction if off_fraction else float("inf") + + if helpful and harmful: + helpful_quality = max(seed_quality(c) for c in helpful) + harmful_quality = min(seed_quality(c) for c in harmful) + worst_relative = max(cells, key=seed_quality) + lines += [ + "", + f"- **Relative seed quality (ON / OFF median error fraction).** The absolute " + f"fraction cannot separate the populations, but the ratio nearly does: helpful " + f"cells top out at {helpful_quality:.2f}, non-helpful cells start at " + f"{harmful_quality:.2f}, with the worst seed in the matrix at " + f"{seed_quality(worst_relative):.2f} (`{name(worst_relative)}`). A fallback " + f'phrased as *"the memo seed is no better than the dense-sign start"* — i.e. ' + f"**ON/OFF error fraction ≳ 0.9** — is therefore the discriminator the raw " + f"fraction is not. Its cost is that computing the OFF fraction means computing " + f"the dense-sign start, which is the work the memo exists to skip.", + ] + + lines += [ + "", + "- **Observability caveat.** The error count is only known once the solve has run, " + "so a fallback cannot threshold on the current evaluation's fraction. Any X would " + "have to be applied to the *previous* evaluation's fraction as a running proxy — " + "which is exactly the quantity the random-walk/iid contrast above shows is stable " + "within a sequence and unstable across regimes.", + ] + + return lines + + +def main() -> int: + parser = argparse.ArgumentParser(description=__doc__) + parser.add_argument("--version", default=None, help="Library version to aggregate.") + parser.add_argument("--results-dir", default=None) + parser.add_argument("--out", default=None) + args = parser.parse_args() + + results_dir = ( + Path(args.results_dir) if args.results_dir else ROOT / "results" / "breakdown" / "imaging" + ) + out_path = ( + Path(args.out) + if args.out + else ROOT / "results" / "notes" / "nnls_warm_start_memo_matrix.md" + ) + + cells, version = load_cells(results_dir, args.version) + + models = sorted( + {c["model"] for c in cells}, key=lambda m: MODEL_ORDER.index(m) if m in MODEL_ORDER else 99 + ) + missing = [m for m in MODEL_ORDER if m not in models] + + header = [ + "# NNLS warm-start memo: model robustness matrix", + "", + f"Aggregated from `results/breakdown/imaging/delaunay_numba_nnls_iterations_*_v{version}.json`", + "by `scripts/imaging/likelihood_breakdown/nnls_iterations_matrix.py`. Companion to", + "[`nnls_warm_start_memo.md`](./nnls_warm_start_memo.md), which measured the single fiducial.", + "", + "Each model variant changes exactly ONE thing about that fiducial (Delaunay Hilbert-1250 +", + "ConstantSplit + MGE-60 linear lens light + Isothermal + shear), memo OFF then ON on the", + "*same* seeded instance sequences: `rw` = random walk (N(0, 1% of each prior's width) per", + "step, the sampler-like regime), `iid` = independent draws from the central 20% of every", + "prior (uncorrelated, pessimistic). Laptop CPU fp64, `OMP_NUM_THREADS=1`,", + "`AUTOARRAY_NUMBA_OPERATED_MEMO=0`. `ratio` = median iterations OFF / ON (>1 = memo saved", + "iterations). `err frac` = warm-start errors / solve size — the quantity a fallback", + "tolerance would threshold on. `Δlnl` = max relative log-likelihood deviation OFF vs ON.", + "", + "Three variants could not be a pure one-thing change, and say so here rather than in a", + "footnote: `rectangular` must drop `ConstantSplit` for `Constant` (split regularization is", + "structurally incompatible with the rectangular interpolator); `source_complex` ships as", + "data/noise/PSF with no truth tracer, so its adapt image is the positive-clipped data, and", + 'it is sampled at 0.05"/px, i.e. hst geometry; `no_edge_zeroing` flips the solve onto the', + "full-system branch but does NOT change the index set — `Delaunay(zeroed_pixels=0)` already", + "keeps every index, and the only mesh here that zeroes any (`rectangular`, its shape-derived", + "136-pixel border) zeroes a *static* set. So the memo key never churns in this matrix.", + "", + ] + if missing: + header += [f"Not measured: {', '.join(f'`{m}`' for m in missing)}.", ""] + + lines = header + markdown_table(cells) + [""] + findings(cells) + [""] + + out_path.parent.mkdir(parents=True, exist_ok=True) + out_path.write_text("\n".join(lines)) + + print("\n".join(lines)) + print(f"\n {len(cells)} cells from {len({c['path'] for c in cells})} result files.") + print(f" Matrix written to: {out_path} ({len(lines)} lines)") + return 0 + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/scripts/misc/likelihood_breakdown/README.md b/scripts/misc/likelihood_breakdown/README.md index 816571df0..c67793231 100644 --- a/scripts/misc/likelihood_breakdown/README.md +++ b/scripts/misc/likelihood_breakdown/README.md @@ -25,8 +25,8 @@ For *how long the likelihood actually takes* on production hardware — i.e. a s | `imaging/delaunay` | hst | hpc_a100_fp64 | sparse (w-tilde) | 98.0 ms | v2026.7.6.649 | | `imaging/delaunay` | hst | hpc_a100_mp | dense (mapping) | 96.8 ms | v2026.7.6.649 | | `imaging/delaunay` | hst | hpc_a100_mp | sparse (w-tilde) | 95.5 ms | v2026.7.6.649 | -| `imaging/delaunay_numba` | euclid | local_cpu_fp64 | sparse (numba) | 4.84 s | v2026.8.17.1 | -| `imaging/delaunay_numba` | hst | local_cpu_fp64 | sparse (numba) | 4.48 s | v2026.8.17.1 | +| `imaging/delaunay_numba` | euclid | local_cpu_fp64 | sparse (numba) | 1.19 s | v2026.8.17.1 | +| `imaging/delaunay_numba` | hst | local_cpu_fp64 | sparse (numba) | 3.22 s | v2026.8.17.1 | | `imaging/mge` | hst | local_cpu_fp64 | dense (mapping) | 179.5 ms | v2026.7.6.649 | | `imaging/mge` | hst | hpc_a100_fp64 | dense (mapping) | 7.8 ms | v2026.7.6.649 | | `imaging/pixelization` | hst | local_cpu_fp64 | dense (mapping) | 8.65 s | v2026.7.6.649 | @@ -95,6 +95,8 @@ If, on the other hand, the two numbers agree closely, the per-step bars are a fa | `imaging/delaunay.py` | Imaging | DelaunayBrightnessImage | 13-step pipeline; Hilbert-curve mesh. | | `interferometer/delaunay.py` | Interferometer | DelaunayBrightnessImage + sparse-DFT | 11-step pipeline. The transform-mapping-matrix step is the interferometer-specific replacement for imaging's PSF convolution. | | `datacube/delaunay.py` | Datacube | DelaunayBrightnessImage × N channels | 8-step pipeline. Channel-invariant steps profiled once; channel-variant steps profiled on channel 0 and multiplied by `N_channels` for the cube cost. | +| `imaging/delaunay_numba.py` | Imaging | DelaunayBrightnessImage (numba CPU, `use_jax=False`) | 18-step pipeline; sparse-operator CPU path, MGE-60 linear lens light. | +| `imaging/delaunay_numba_nnls_iterations.py` | Imaging | as `delaunay_numba.py` | **Diagnostic, not a breakdown.** A/B of the cross-evaluation NNLS warm-start memo (`aa.Settings(nnls_warm_start_memo=...)`, PyAutoArray#498) over two 30-instance sequences (random walk / i.i.d.), reporting active-set iterations, solve and evaluation time, and memo-vs-no-memo parity. Findings: [`results/notes/nnls_warm_start_memo.md`](../../../results/notes/nnls_warm_start_memo.md). | Four cells are intentionally absent from this package: - `interferometer/mge` — full-pipeline-by-design, no per-step decomposition (see runtime). diff --git a/scripts/misc/simulators/README.md b/scripts/misc/simulators/README.md index a97698d52..282d01a80 100644 --- a/scripts/misc/simulators/README.md +++ b/scripts/misc/simulators/README.md @@ -50,7 +50,7 @@ Auto-generated by `scripts/build_readme.py` from the latest `*_summary_v