PyAutoInsight/sampler_candidates.json at main · PyAutoLabs/PyAutoInsight · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
[
  {
    "id": "dynesty",
    "name": "dynesty",
    "family": "dynamic nested sampling",
    "paper_url": "https://arxiv.org/abs/1904.02180",
    "code_url": "https://github.com/joshspeagle/dynesty",
    "expected_uses": [
      "Evidence and posterior baseline with adjustable effort allocation."
    ],
    "limitations": [
      "Prior transforms and likelihood-contour exploration need validation; no universal dimensional scaling advantage."
    ],
    "integration_status": "integrated: DynestyStatic and DynestyDynamic",
    "benchmark_status": "not_assessed: this literature review establishes no benchmark result",
    "reviewed_at": "2026-10-08",
    "investigation_prompt": "Investigate dynesty on a normalized Gaussian and a separated-mode target using repeated seeds. Check posterior coverage and evidence accuracy. Record package version, initialization, total wall time including setup, evaluations and uncertainty before proposing a science campaign."
  },
  {
    "id": "nautilus",
    "name": "Nautilus",
    "family": "neural-network importance nested sampling",
    "paper_url": "https://arxiv.org/abs/2306.16923",
    "code_url": "https://github.com/johannesulf/nautilus",
    "expected_uses": [
      "Evidence and posterior estimation for expensive gradient-free likelihoods."
    ],
    "limitations": [
      "Learned proposal coverage and importance-weight stability need checking on unseen modes."
    ],
    "integration_status": "integrated: Nautilus",
    "benchmark_status": "not_assessed: this literature review establishes no benchmark result",
    "reviewed_at": "2026-10-08",
    "investigation_prompt": "Investigate Nautilus on a normalized Gaussian and a separated-mode target using repeated seeds. Check posterior coverage and evidence accuracy. Record package version, initialization, total wall time including setup, evaluations and uncertainty before proposing a science campaign."
  },
  {
    "id": "blackjax-nuts",
    "name": "BlackJAX NUTS",
    "family": "gradient Hamiltonian Monte Carlo",
    "paper_url": "https://arxiv.org/abs/2402.10797",
    "code_url": "https://github.com/blackjax-devs/blackjax",
    "expected_uses": [
      "Posterior refinement for differentiable JAX targets."
    ],
    "limitations": [
      "Requires gradients, warmup and convergence diagnostics; NUTS alone supplies no evidence and can miss isolated modes."
    ],
    "integration_status": "integrated: BlackJAXNUTS",
    "benchmark_status": "not_assessed: this literature review establishes no benchmark result",
    "reviewed_at": "2026-10-08",
    "investigation_prompt": "Investigate BlackJAX NUTS on a normalized Gaussian and a separated-mode target using repeated seeds. Check posterior coverage, divergences and effective sample size. Record package version, initialization, total wall time including setup, evaluations and uncertainty before proposing a science campaign."
  },
  {
    "id": "ultranest",
    "name": "UltraNest",
    "family": "reactive nested sampling",
    "paper_url": "https://arxiv.org/abs/2101.09604",
    "code_url": "https://github.com/JohannesBuchner/UltraNest",
    "expected_uses": [
      "Independent evidence cross-check for non-Gaussian and multimodal targets."
    ],
    "limitations": [
      "Contour coverage and stopping settings still require repeated-seed checks; robustness can cost evaluations."
    ],
    "integration_status": "prospective: no PyAutoFit search wrapper found",
    "benchmark_status": "not_assessed: this literature review establishes no benchmark result",
    "reviewed_at": "2026-10-08",
    "investigation_prompt": "Investigate UltraNest on a normalized Gaussian and a separated-mode target using repeated seeds. Check posterior coverage and evidence accuracy. Record package version, initialization, total wall time including setup, evaluations and uncertainty before proposing a science campaign."
  },
  {
    "id": "pocomc",
    "name": "pocoMC",
    "family": "flow-preconditioned Monte Carlo",
    "paper_url": "https://arxiv.org/abs/2207.05660",
    "code_url": "https://github.com/minaskar/pocomc",
    "expected_uses": [
      "Posterior and evidence investigation for strongly correlated nonlinear targets."
    ],
    "limitations": [
      "Flow training, population coverage and evidence uncertainty need validation; check the pinned implementation against the 2022 method."
    ],
    "integration_status": "prospective: no PyAutoFit search wrapper found",
    "benchmark_status": "not_assessed: this literature review establishes no benchmark result",
    "reviewed_at": "2026-10-08",
    "investigation_prompt": "Investigate pocoMC on a normalized Gaussian and a separated-mode target using repeated seeds. Check posterior coverage and evidence accuracy. Record package version, initialization, total wall time including setup, evaluations and uncertainty before proposing a science campaign."
  }
]