docs: record the autogalaxy-leg shipping - #107
Merged
Merged
Conversation
…rrors, New User Guide) Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_012Eacj9naBWSKrevsqnVJWR
Jammy2211
pushed a commit
that referenced
this pull request
Aug 13, 2026
…delaunay-nn worktree Two-tier architecture: tier 1 (components, no data) in scripts/misc/hazards/; tier 2 (likelihood: linear algebra, residuals, chi-squared) in scripts/<dataset>/hazards/. Linear algebra is lumped with the data-specific tier because the conditioning floors are absolute constants judged against the data's flux/noise scale. Registers in planned.md (blocked by delaunay-nn-laptop-gpu-profile, PR #106 open) and files the phase-2 tier-2 prompt. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0132HpisYts9aDHWUE4e7Zr2
Jammy2211
pushed a commit
that referenced
this pull request
Aug 13, 2026
…rial review Codex review returned RETHINK; all 7 factual claims confirmed. Adopted: - typed measurements replace the universal predicate (prior_mass / epsilon_neighbourhood / reachability / error_curve) — MC prior volume is meaningless for measure-zero gradient sites and for the ~5e-6-wide 0.99999 clamp, so revision 1's acceptance test contradicted its own design - reachability is first-class: EllProfile.__init__ validates at construction, so the ell_comps clamp is unreachable on the public numpy path while JAX tracing reaches it - phase 1 re-cut as a vertical slice, one case per subject shape - a consumer loop (--check regression mode + hazards_index.json) - anchors key on a semantic finding ID, re-verified by re-running the reproducer Tier becomes metadata on a finding rather than a directory: detectors all live in scripts/misc/hazards/, dataset cells in scripts/<dataset>/hazards/. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_0132HpisYts9aDHWUE4e7Zr2
Jammy2211
pushed a commit
that referenced
this pull request
Aug 13, 2026
Jammy2211
pushed a commit
that referenced
this pull request
Aug 13, 2026
Jammy2211
added a commit
that referenced
this pull request
Aug 13, 2026
Jammy2211
added a commit
that referenced
this pull request
Aug 13, 2026
Jammy2211
pushed a commit
that referenced
this pull request
Aug 23, 2026
…re-run once The cancelled job's log localises the hang precisely: after `resolved jax 0.11.1` and 8 passing scripts, `imaging/jax_likelihood/mge.py` opened its group and the job produced no further output for 59 minutes, leaving orphan python processes at teardown. Two candidate causes, and the evidence favours the first. This repo has a documented intermittent hang (four prior runs at ~6h against an 11-14min norm; a sibling JAX likelihood script is already parked for it). Against the alternative that the jax 0.10.2 -> 0.11.1 move caused it: merged autolens#268 ran the same imaging/jax_likelihood/mge.py under 0.11.1 and passed in 29.7s, 23/23 overall, so the tfp-nightly Matern-kernel path is not broken by the bump. Re-running the same commit to discriminate. #107 stays unmerged meanwhile. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01KMknTjH347FBZUE82nNEby
Jammy2211
pushed a commit
that referenced
this pull request
Aug 23, 2026
…oarray guard) Surfaced by the phase-3 pynufft PR (@PyAutoCTI#107), whose diff is one deleted .rst line — the failure is main's, not the PR's. lib-tests clones PyAutoArray main, and #440's Mask2D zero-length shape_native guard (merged 2026-08-09, after CTI's last main run on 2026-08-07) now rejects a (3, 0) region that a serial-EPER test built from an out-of-range pixels window. The assertion it fed was passing vacuously — an empty-array comparison is True for any expected value. Also records the PR trio's CI state in the active.md entry. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Q7kFdoYcD6wnTifNXV5T98
Jammy2211
pushed a commit
that referenced
this pull request
Aug 23, 2026
…3 doc scope @PyAutoCTI#108 takes option 1 on the serial-EPER zero-width region: test-only, pins today's raise so main goes green, leaves the clip-vs-raise semantics question open. @PyAutoCTI#107 merges after it. pylops is defunct (human-authorized), so #107 now removes the whole "optional requirements" stanza rather than just the pynufft line — the same green-CI evidence covers both. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Q7kFdoYcD6wnTifNXV5T98
Jammy2211
pushed a commit
that referenced
this pull request
Aug 23, 2026
… recorded All three legs (3.12, 3.13, no-jax) pass with the vacuous serial-EPER assertion replaced. #107 merges after it, once its branch picks up the fixed main. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Q7kFdoYcD6wnTifNXV5T98
Jammy2211
pushed a commit
that referenced
this pull request
Aug 24, 2026
Merged 2026-08-24: @PyAutoHands#259, @PyAutoHeart#162, @PyAutoCTI#107, plus @PyAutoCTI#108 — an unplanned test-only fix for a red CTI main this task walked into (autoarray's #440 shape_native guard exposing a vacuous serial-EPER assertion). All four green on every leg at merge; all four branches proven merged with 0 unmerged commits. Closes the residue sweep: phases 1-2 shipped 2026-08-23. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Q7kFdoYcD6wnTifNXV5T98
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Status bookkeeping for the autogalaxy leg of the regime reorganization: the
autogalaxy_multi_galaxy_package,autogalaxy_cluster_packageandautogalaxy_docs_regime_guideschild prompts shipped (autogalaxy_workspace#168, autogalaxy_workspace_test#97, PyAutoGalaxy#526 — all merged), and the epic's execution-order index marks items 6–8 shipped.lifecycle checkpasses.🤖 Generated with Claude Code
https://claude.ai/code/session_012Eacj9naBWSKrevsqnVJWR
Generated by Claude Code