Affected Repositories

• PyAutoArray (primary)

Branch Survey

Suggested branch: feature/files-experiment-scripts-cleanup

Implementation Steps

1. Worktree ~/Code/PyAutoLabs-wt/files-experiment-scripts-cleanup/PyAutoArray.
2. git rm -r files/; commit.
3. Verify witness grep; pytest test_autoarray -x -q sanity run.
4. PR → glance merge on green → /prm close-out.

Key Files

• files/cdf_audit.md, files/ghost_peak.png, files/ghost_peak_experiment.py, files/ghost_peak_findings.md, files/pca_rotation.png, files/pca_rotation_experiment.py — removed

files/*_experiment.py import a module that no longer exists — decide whether to keep them

Type: maintenance
Target: PyAutoArray
Repos:

• PyAutoArray
  Difficulty: small
  Autonomy: supervised
  Priority: low
  Status: formalised
  Consequence: glance
  Witness: every script under PyAutoArray/files/ either imports successfully against the installed stack, or is gone; git grep -c rectangular_spline returns 0.
  Review-minutes: 3
  Unattended: ready

The finding

PyAutoArray/files/ghost_peak_experiment.py:51 and files/pca_rotation_experiment.py:40 both do:

from autoarray.inversion.mesh.interpolator.rectangular_spline import (...)

That module does not exist. The live contents of
autoarray/inversion/mesh/interpolator/ are abstract.py, delaunay.py, knn.py,
rectangular.py, rectangular_uniform.py, sibson.py. So both scripts fail at import.

Independently confirmed by the workspace's own PyAuto API gate, which refused an import
attempt with: "STALE autoarray.inversion.mesh.interpolator.rectangular_spline — not in
installed stack (missing .rectangular_spline); closest live names: rectangular".

Provenance

Both were committed with the rectangular-adapt-cdf spike (f9aceea3, f9c6d218) and the
module has since been renamed or removed. Their docstrings still told readers to
source ~/Code/PyAutoLabs-wt/rectangular-adapt-cdf/activate.sh — a task bundle long gone —
which is corroborating evidence that nothing has run them since. Those docstring lines were
corrected in PyAutoBrain#393 (phase 2 of the workspace regroup); the dead import was left
alone as out of scope for a prose fix, and is filed here instead.

Nothing references either file except each other and files/ghost_peak_findings.md.

The decision, which is a human's

Not obviously a bug to fix — the question is whether these spike scripts are still wanted:

• Keep and repair — repoint the import at whichever live module replaced
  rectangular_spline (rectangular.py / rectangular_uniform.py), and check the symbols
  they import still exist under their new names. Only worth it if the experiments are still
  interesting.
• Retire — delete both scripts, and decide separately whether ghost_peak_findings.md is
  a result worth keeping without the script that produced it (it may well be: the finding
  outlives the spike).

Do not "fix" them by guessing an import that makes them parse. If they are kept they should
actually run.