Jammy2211 · Oct 7, 2026 · Oct 7, 2026 · Oct 7, 2026
diff --git a/campaigns.yaml b/campaigns.yaml
diff --git a/dashboard.html b/dashboard.html
diff --git a/dashboard.md b/dashboard.md
diff --git a/tasks/autofit_profiling_bootstrap.md b/tasks/autofit_profiling_bootstrap.md
@@ -8,6 +8,11 @@ The original task below is retained verbatim. Current scheduling state lives in
 - VERIFIED: `gh repo view PyAutoLabs/autofit_profiling` → does not exist; no local checkout. The Mind source draft was removed in Mind `dbbe5b22` (2026-10-03, move to Pulse); it is not in Mind complete/ or active.md, so this file is the live record.
 - Human decision still pending: the repo-creation question (Witness). No producer registration until the repo exists with ported baselines.
 
+## Check-in 2026-10-07
+
+- https://github.com/PyAutoLabs/autofit_profiling created 2026-10-07T11:38Z: one Initial commit with LICENSE only, no PRs. The repo-creation question is answered.
+- Status → `ready`: porting/bootstrap unstarted; no producer registration until baselines are ported.
+
 ---
 
 # autofit_profiling: bootstrap the repo + general PyAutoFit profiling epic

diff --git a/tasks/gradient_cost_probe.md b/tasks/gradient_cost_probe.md
@@ -8,6 +8,10 @@ The original task below is retained verbatim. Current scheduling state lives in
 - VERIFIED: [autolens_workspace_developer#117](https://github.com/PyAutoLabs/autolens_workspace_developer/issues/117) closed 2026-07-28; no change since the pin. The Mind source draft was removed in Mind `dbbe5b22` (2026-10-03, move to Pulse); it is not in Mind complete/ or active.md, so this file is the live record.
 - Next: select one bounded step when prioritised.
 
+## Check-in 2026-10-07
+
+- Context: Heart unit-test timing (PyAutoLens py3.13, 2026-10-07) flagged test__forward_gradient_matches_reverse_through_fitness 5.03→12.60 s and test__multi_start_gradient_point_source_fit_runs_in_forward_mode 15.51→30.38 s. https://github.com/PyAutoLabs/PyAutoLens/pull/768 rewrote that test file, so this is not a controlled library-cost comparison; this probe is the right instrument. Unissued.
+
 ---
 
 # A gradient-cost probe: forward vs `value_and_grad` ms/eval and a strict FD check, on any registry cell

diff --git a/tasks/interferometer_decision_matrix_last_cell.md b/tasks/interferometer_decision_matrix_last_cell.md
@@ -18,6 +18,12 @@ The original task below is retained verbatim. Current scheduling state lives in
 - Caveat (executing agent): the A100 row's own A/B shows certified solver = PDIP to 0.0 nat, so the gap may come from F/D on older revisions rather than the solver. Not verified.
 - Correction: PyAutoArray#582 was first released in 2026.9.27.2.
 
+## Check-in 2026-10-07
+
+- https://github.com/PyAutoLabs/autolens_profiling/pull/372 (issue #369) merged 2026-10-04T20:40Z: CPU rect 39² alma_high r5.0 from RAL 375978_3; the human accepted the 1.59e-3 nat CPU/A100 gap on 2026-10-04.
+- Wiki Next line confirms every decision-matrix cell is measured, blocked or not measured with a citation (https://github.com/PyAutoLabs/autolens_profiling/blob/719459fefb40c7e5be157f652818dbf923830557/wiki/campaigns/interferometer_likelihood.md).
+- Status → `complete`.
+
 ---
 
 # Interferometer decision matrix — fill the last cell (CPU rect 39² at alma_high r5.0) and flip the wiki row to shipped

diff --git a/tasks/interferometer_fixed_mapper_curvature_preload.md b/tasks/interferometer_fixed_mapper_curvature_preload.md
@@ -18,6 +18,10 @@ The original task below is retained verbatim. Current scheduling state lives in
 - Finding for the human: no production stage qualifies as configured - interferometer SLaM `source_pix_2` fixes the mass but leaves `weight_power`/`weight_floor` free, which change the mapping. Phase 1 measures a regularization-only configuration; adopting such a stage is a science decision.
 - Later phases: search-side opt-in (`fixed_mapper=True`), JAX jit/vmap + compile size, A100/jvla rows, data-vector preload, SLaM regularization-only stage, mixed MGE+mapper, dense path.
 
+## Check-in 2026-10-07
+
+- Mind draft `draft/feature/autoarray/interferometer_curvature_preload_phase1.md` (2026-10-04) still unstarted.
+
 ---
 
 # Interferometer fixed-mapper searches: reuse the W~ curvature matrix across likelihood calls via `preloads.curvature_matrix`

diff --git a/tasks/interferometer_nnls_memo_scattered_stream_guard.md b/tasks/interferometer_nnls_memo_scattered_stream_guard.md
@@ -8,6 +8,12 @@ The original task below is retained verbatim. Current scheduling state lives in
 - VERIFIED 2026-10-04: no PR, issue, branch or results on GitHub; unstarted. No Mind active claim.
 - Next: select one bounded step when the interferometer campaign prioritises it.
 
+## Check-in 2026-10-07
+
+- Correction: the 2026-10-04 "unstarted" was superseded. https://github.com/PyAutoLabs/PyAutoArray/pull/615 (closes PyAutoArray#613) "back off the fnnls warm-start memo on scattered evaluation streams" merged 2026-10-06 (58bdda0a), released in 2026.10.7.1. Mind record complete/2026/10/nnls-memo-scattered-backoff.md.
+- Local witness (n=576, solver-only, min of 3, 3 seeds, 64 solves): iid memo on/off 1.49x → 1.17x; walk 0.18x unchanged.
+- Status → `active`: the autolens_profiling#332 ALMA Delaunay 2.17x after-measurement was not re-run and needs compute authorization.
+
 ---
 
 # fnnls warm-start memo: stop it slowing scattered evaluation streams (interferometer CPU, autolens_profiling#332)

diff --git a/tasks/interferometer_streaming_scaling.md b/tasks/interferometer_streaming_scaling.md
@@ -19,6 +19,11 @@ The original task below is retained verbatim. Current scheduling state lives in
 - Parity 1.2e-9 nats at 5e5. The 1e8 row was skipped (budget).
 - Go/no-go overtaken; recorded as release evidence for 2026.10.4.1.
 
+## Check-in 2026-10-07
+
+- https://github.com/PyAutoLabs/autolens_profiling/pull/375 (issue #368) merged 2026-10-04T20:49Z: phase 1 CPU cells + results/streaming_scaling rows; https://github.com/PyAutoLabs/autolens_profiling/blob/719459fefb40c7e5be157f652818dbf923830557/wiki/campaigns/interferometer_streaming.md.
+- Optional next: 1e8 CPU row and A100 rows (compute authorization). Library candidate still unfiled: chunk transformer.image_from in apply_sparse_operator.
+
 ---
 
 # Campaign: interferometer streaming (array-free) vs in-memory — memory and time scaling to 2e8 visibilities

diff --git a/tasks/mge_nnls_fix_pyautoarray_571_slam_60.md b/tasks/mge_nnls_fix_pyautoarray_571_slam_60.md
@@ -17,6 +17,12 @@ The original task below is retained verbatim. Current scheduling state lives in
 - HPCPullPyAuto pulls MAINS, and Nerves main excludes jax 0.10.2 (installed on RAL and locally), so a sync also needs a jax reinstall.
 - Next: sync when `squeue -u jnightin -t R` shows no euclid_dr1 jobs on the shared base, or the human accepts the risk.
 
+## Check-in 2026-10-07
+
+- RAL 2026-10-07 ~18:40Z: `squeue -u jnightin` showed 0 RUNNING, 0 PENDING (one probe). The 2026-10-04 blocker condition (no euclid_dr1 jobs on the shared base) is met.
+- Three later probes timed out: shared mirror /mnt/ral/jnightin/PyAuto HEAD/tag UNVERIFIED.
+- Human decision (same day, in chat): **never** HPCPullPyAuto the shared mirror while euclid_dr1 depends on it. Phase 3 runs from a private RAL checkout at tag 2026.10.7.1 (contains PyAutoArray#595) with its own environment; Nerves at that tag excludes jax 0.10.* and RAL has jax 0.10.2, so the private env needs a compliant jax. Status → `ready`; compute for the A100 rows still needs separate authorization.
+
 ---
 
 # Linear-solver programme phase 3: GPU/vmap/A100 timing and parity rows for the solver corpus

diff --git a/tasks/pair_jax_xla_env_vars_with_measured.md b/tasks/pair_jax_xla_env_vars_with_measured.md
@@ -9,6 +9,10 @@ Original task retained verbatim below. Current scheduling state: `campaigns.yaml
 - Context: jax-version policy changes (Mind `complete/2026/10/jax-lapack-compatibility-repair.md`) merged to library mains 2026-10-04, unreleased.
 - Next: select one bounded step when prioritised.
 
+## Check-in 2026-10-07
+
+- Context: the jax-version policy is now released; all 2026.10.7.1 libraries require Nerves with the JAX deadlock exclusions. Unissued.
+
 ---
 
 # Pair JAX/XLA env vars with measured compile and run times, per backend

diff --git a/tasks/point_source_image_plane_gpu_breakdown.md b/tasks/point_source_image_plane_gpu_breakdown.md
@@ -16,6 +16,11 @@ The original task below is retained verbatim. Current scheduling state lives in
 - Forward-mode NaN bug issued: [PyAutoLens#767](https://github.com/PyAutoLabs/PyAutoLens/issues/767). Reproduced on CPU fp64: forward `[nan]x5`, reverse finite; stack 2026.10.4.1+3 `b695e57b6`. Mind registry commit `0112ccba`.
 - The autolens_inference image-plane fit measurement still awaits human authorisation.
 
+## Check-in 2026-10-07
+
+- https://github.com/PyAutoLabs/PyAutoLens/issues/767 closed; fixed by https://github.com/PyAutoLabs/PyAutoLens/pull/768 (merged 2026-10-04T20:41Z, 0659903d4), in tag 2026.10.7.1 (merge-base verified). Correctness only; no timing claim.
+- Stays `needs-decision`: the autolens_inference image-plane fit measurement still awaits human authorisation.
+
 ---
 
 # Point-source A100 speed-up campaign: profile and optimize with the shared breakdown

diff --git a/tasks/pointsolver_cpu_speed_campaign_remainder.md b/tasks/pointsolver_cpu_speed_campaign_remainder.md
@@ -15,6 +15,12 @@ The original task below is retained verbatim. Current scheduling state lives in
 - Issue: [autolens_profiling#370](https://github.com/PyAutoLabs/autolens_profiling/issues/370). Branch `feature/point-source-wiki-reconcile` @`0560284`, parked at Heart RED.
 - Still owed: RAL cleanup, register_model grad-zero prompt, test move, CI smoke cells, `nopad` deletion.
 
+## Check-in 2026-10-07
+
+- https://github.com/PyAutoLabs/autolens_profiling/pull/373 (issue #370) merged 2026-10-04T20:49Z: wiki reconcile + completion evidence (https://github.com/PyAutoLabs/autolens_profiling/blob/719459fefb40c7e5be157f652818dbf923830557/results/notes/point_source_cpu_campaign.md#campaign-completion-evidence-2026-10-04).
+- Still owed (unchanged), now tracked by Mind draft `draft/maintenance/autolens/point_source_cpu_campaign_owed_leftovers.md` (2026-10-04): RAL cleanup, test move, register_model grad-zero prompt, CI smoke cells, constant_folding A/B, nopad deletion, quiet re-runs.
+- RAL leftover folders not re-listed 2026-10-07 (unverified).
+
 ---
 
 # Point-source (single-source) CPU campaign — carried leftovers and completion evidence

diff --git a/tasks/runtime_cell_single_jit_gpu_warmup.md b/tasks/runtime_cell_single_jit_gpu_warmup.md
@@ -15,6 +15,12 @@ The original task below is retained verbatim. Current scheduling state lives in
 - Local CPU witness: `single_jit` 0.417 ms vs median 0.388 ms (p10 0.320, p90 0.545).
 - Imaging cells not yet wired.
 
+## Check-in 2026-10-07
+
+- Option (a) merged: https://github.com/PyAutoLabs/autolens_profiling/pull/374 (issue #371, 2026-10-04). `full_pipeline_single_jit_median_ms` (p10/p90) sits beside the unchanged `single_jit`; the GPU headline is labelled "first block after compile" (https://github.com/PyAutoLabs/autolens_profiling/blob/719459fefb40c7e5be157f652818dbf923830557/scripts/point_source_source/source_plane/likelihood_runtime_solved.py).
+- Imaging release-sweep cells are not wired. `git log --since=2026-10-04 origin/main -- results/` shows only #375 streaming rows and a README layout refactor: no release sweep on main, so the new field has no sweep rows yet.
+- Stays `active`: wire imaging cells; the field appears at the next release sweep.
+
 ---
 
 # Runtime cells' A100 `single_jit` includes the post-compile warm-up — source-plane 0.642 ms vs a steady 0.267 ms on the same node

diff --git a/tasks/setup_baseline_campaign.md b/tasks/setup_baseline_campaign.md
@@ -69,3 +69,9 @@ keeps its existing receipts and snapshots; this task creates no new measurements
 Until those steps are complete, retain `needs-decision`: no campaign is reported
 as running or accepted, and specification implementation remains distinct from
 the later collection and review deliverables.
+
+## Check-in 2026-10-07
+
+- Specification merged: https://github.com/PyAutoLabs/autolens_profiling/pull/385 (2026-10-06; issue #384 closed 2026-10-06; PyAutoPulse#17 closed 2026-10-06). `baseline/campaign.json` is status draft, version 1, with `baseline/README.md` readiness contract (https://github.com/PyAutoLabs/autolens_profiling/blob/719459fefb40c7e5be157f652818dbf923830557/baseline/campaign.json, https://github.com/PyAutoLabs/autolens_profiling/blob/719459fefb40c7e5be157f652818dbf923830557/wiki/campaigns/setup_baseline.md).
+- Six unknowns recorded verbatim: no frozen revisions/env lock; no collection authorization/window; hardware/affinity identity unselected; each cell needs complete settings, input hashes, witness + tolerances; compile per-cell builder unavailable; proposed fp64/mixed coverage is not evidence of support.
+- Status stays `needs-decision`: no collection or acceptance authorized; the check-in authorizes none.
diff --git a/tasks/timing_noise_audit.md b/tasks/timing_noise_audit.md
@@ -9,6 +9,11 @@ The original task below is retained verbatim. Current scheduling state lives in
 - Adjacent but distinct: PyAutoHeart#276 / PR #277 (unit-timing distinct baseline) completed 2026-10-04; that is Heart unit-test timing, not profiling gates.
 - Next: start the audit (read-and-classify) phase via start_dev, reusing #362.
 
+## Check-in 2026-10-07
+
+- https://github.com/PyAutoLabs/autolens_profiling/issues/362 still open, 0 comments, unassigned, unchanged since 2026-10-02.
+- Related new Mind draft: `draft/bug/autolens_profiling/call_accounting_ci_timing_threshold.md` (2026-10-06).
+
 ---
 
 # Audit timing tests and profiling gates for measurement noise