fix: unpark ellipse example scripts after JAX refactor

[Jammy2211]

Overview

The five ellipse example scripts under autogalaxy_workspace/scripts/ellipse/ (simulator, fit, modeling, multipoles, database) were parked on 2026-04-24 in config/build/no_run.yaml pending a JAX refactor of the ellipse code. That refactor has now landed in PyAutoGalaxy via three merged PRs (#408 ellipse-xp, #410 fit-ellipse-jax, #412 analysis-ellipse-jax, all completed 2026-05-14). This task verifies the scripts pass under PYAUTO_TEST_MODE=2, fixes any remaining API drift, and removes the NEEDS_FIX markers.

Plan

• Try each of the five ellipse example scripts under PYAUTO_TEST_MODE=2 first — some may simply pass now thanks to the stronger sampler bypass and the refactor.
• For scripts that still fail, investigate and fix the underlying bug (most likely candidate is modeling.py with a KeyError on ellipses.0.centre_0 kwargs after API drift, and the previous multipoles.py timeout).
• Remove each script's entry from autogalaxy_workspace/config/build/no_run.yaml once it passes.
• If ellipse/database passes, also remove the fallback entry from PyAutoBuild/autobuild/config/no_run.yaml (autogalaxy section) so the mega-run no longer skips it.
• Re-run the workspace mega-run via run_all_scripts.sh and confirm no NEEDS_FIX 2026-04-24 banner remains for any ellipse entry.

Detailed implementation plan

Affected Repositories

• autogalaxy_workspace (primary)
• PyAutoBuild (secondary — one-line removal in config/no_run.yaml, only if ellipse/database passes)

Work Classification

Workspace

Branch Survey

Repository	Current Branch	Dirty?
./autogalaxy_workspace	main	dirty in dataset/imaging/* and dataset/interferometer/* — looks like leftover simulator regen artifacts (FITS/JSON/PNG), unrelated to this task. Will be sidestepped by working in a worktree off origin/main.
./PyAutoBuild	main	clean

Suggested branch: feature/unpark-ellipse-scripts
Worktree root: ~/Code/PyAutoLabs-wt/unpark-ellipse-scripts/ (created later by /start_workspace)

Implementation Steps

1. Worktree both repos to feature/unpark-ellipse-scripts off origin/main under ~/Code/PyAutoLabs-wt/unpark-ellipse-scripts/.
2. For each of the five ellipse scripts, run from the worktreed workspace:

   cd ~/Code/PyAutoLabs-wt/unpark-ellipse-scripts/autogalaxy_workspace
   PYAUTO_TEST_MODE=2 python scripts/ellipse/<name>.py

   in this order: simulator.py, fit.py, modeling.py, multipoles.py, database.py.
3. For any script that still fails, diagnose and fix. Known candidates flagged in the original prompt:
   • modeling.py: KeyError on ellipses.0.centre_0 kwargs — likely model-component path drift after the JAX refactor; check Model composition vs. the post-refactor Ellipse class layout.
   • modeling.py and multipoles.py: timed out under TEST_MODE=1 in the mega-run — TEST_MODE=2 bypasses the sampler more aggressively and may resolve this.
4. As each script passes, remove its - ellipse/<name> # NEEDS_FIX 2026-04-24 … line from autogalaxy_workspace/config/build/no_run.yaml.
5. If ellipse/database passes, also remove the line - ellipse/database # Ellipse model needs refactor and JAX support from the autogalaxy: section of PyAutoBuild/autobuild/config/no_run.yaml.
6. Re-run the mega-run via autogalaxy_workspace/run_all_scripts.sh (or the relevant /smoke_test invocation) and confirm:
   • No ellipse entry shows a NEEDS_FIX banner.
   • All five ellipse scripts complete cleanly.
7. Ship via /ship_workspace with cross-reference comment back to this issue.

Key Files

• autogalaxy_workspace/config/build/no_run.yaml — remove 5 ellipse/* lines
• autogalaxy_workspace/scripts/ellipse/{simulator,fit,modeling,multipoles,database}.py — fix any remaining API drift / xp call-site bugs
• PyAutoBuild/autobuild/config/no_run.yaml — possibly remove - ellipse/database line in autogalaxy: section

Original Prompt

Click to expand starting prompt

• All ellipse example scripts under autogalaxy_workspace/scripts/ellipse/ are currently in
  autogalaxy_workspace/config/build/no_run.yaml with a NEEDS_FIX 2026-04-24 marker.

  The five entries are:

  • ellipse/simulator
  • ellipse/fit
  • ellipse/modeling
  • ellipse/multipoles
  • ellipse/database

  They were parked because the ellipse model needs a refactor and JAX support (tracked separately in
  PyAutoPrompt/autogalaxy/ellipse_fitting_jax.md). In particular, ellipse/modeling and
  ellipse/multipoles time out under PYAUTO_TEST_MODE=1 in the mega-run, and
  ellipse/modeling additionally raises a KeyError on ellipses.0.centre_0 kwargs after API drift.

  When the JAX refactor lands:

  1. Try running each ellipse script with PYAUTO_TEST_MODE=2 first — some may just need the
     stronger sampler bypass.
  2. Remove the five ellipse/* lines from autogalaxy_workspace/config/build/no_run.yaml.
  3. If the refactor also unlocks aggregator-style usage, the ellipse/database entry in
     PyAutoBuild/autobuild/config/no_run.yaml (the fallback list) can be removed too.
  4. Re-run the mega-run (run_all_script_fix_failures skill in autogalaxy_workspace) to confirm
     every ellipse script passes.

[Jammy2211]

Library PR Created — 2026-05-15

PR

• fix: dedupe number_of_cores in Drawer for from_dict round-trip PyAutoFit#1270 — fix: dedupe number_of_cores in Drawer for from_dict round-trip

Status

Library shipped. Workspace changes are next (no_run.yaml cleanup + database.py pathlib patch), already prepared in the same worktree. Next: /ship_workspace.

API Changes

None — internal bugfix only. Direct af.Drawer(...) construction is unaffected; the fix restores the previously-broken from_dict round-trip path that the aggregator hits when scraping saved Drawer fits. Surfaced via autogalaxy_workspace/scripts/ellipse/database.py.

[Jammy2211]

Shipped — 2026-05-15

PRs

• fix: unpark ellipse example scripts after JAX refactor #73 — fix: unpark ellipse example scripts after JAX refactor (workspace)
• config: remove ellipse/database from autogalaxy no_run fallback PyAutoBuild#89 — config: remove ellipse/database from autogalaxy no_run fallback (build config)

Upstream PR

• fix: dedupe number_of_cores in Drawer for from_dict round-trip PyAutoFit#1270 — fix: dedupe number_of_cores in Drawer for from_dict round-trip (library)

Summary

All five ellipse example scripts (simulator, fit, modeling, multipoles, database) now run cleanly under PYAUTO_TEST_MODE=2 and are removed from autogalaxy_workspace/config/build/no_run.yaml and the fallback in PyAutoBuild/autobuild/config/no_run.yaml. The previously-flagged modeling.py KeyError and multipoles.py timeout were resolved by the upstream ellipse JAX refactor (PyAutoGalaxy #408 / #410 / #412). database.py exposed two unrelated bugs that were also fixed: a leftover path.exists(...) from the prior os.path → pathlib refactor (workspace-side), and a Drawer.__init__ from_dict round-trip crash on duplicate number_of_cores (library-side, PyAutoFit #1270).

Session Notes

• Scope expanded mid-task to include a PyAutoFit library PR after database.py's aggregator path surfaced the Drawer deserialization bug. The fix (kwargs.pop("number_of_cores", None) before super().__init__) is one line plus a to_dict/from_dict regression test; full PyAutoFit suite passes (1243/1243).
• autogalaxy_workspace/smoke_tests.txt still has # ellipse/modeling.py disabled with a comment citing rhayes777/PyAutoFit#1179. That comment may be stale now that the JAX refactor + Drawer fix are in — worth a follow-up check to re-enable, but kept out of this PR to avoid mass-promotion (per the "smoke tests are a small curated subset" guideline).
• The smoke (3.13) CI job failed on this PR but is also failing on main (pre-existing setup_notebook workspace-root error in interferometer/start_here.py and interferometer/simulator.ipynb — unrelated to ellipse). smoke (3.12) passes on both. Worth filing as a separate workspace issue.