Repository navigation
fix(inversion): regularization_weights_mapper_dict uses correct linear_obj_list index - #304
Merged
Conversation
…r_obj_list index `regularization_weights_mapper_dict` iterated `enumerate(cls_list_from(cls=Mapper))` and passed the resulting Mapper-only index into `regularization_weights_from`, which looks up `self.linear_obj_list[index]`. When a non-Mapper linear object (e.g. a `LightProfileLinearObjFuncList` for an MGE bulge) precedes the source mapper in `linear_obj_list`, the dict ended up keyed with the wrong linear obj's regularization weights — e.g. the 40-element lens MGE weights stored under a 530-pixel source mapper. matplotlib's `ax.tripcolor` then refused to render the source-plane regularization-weights subplot with "The length of c must match either the number of points or the number of triangles". Replace the enumerate-based index with `self.linear_obj_list.index(mapper)` so each mapper receives weights derived from its own entry. New regression tests verify the dict slot for a Mapper preceded by a linear obj (and for two mappers separated by a linear obj) holds that mapper's weights. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
Fixes
AbstractInversion.regularization_weights_mapper_dictmis-indexing when a non-Mapper linear object precedes a sourceMapperinlinear_obj_list. The mapper's dict slot was being populated with the regularization weights of whicheverlinear_obj_list[i]happened to share its enumerated position in the Mapper-only sublist — for a typical MGE-lens-light + pixelized-source layout, that meant the 40-element lens MGE weights ended up keyed under a 530-pixel source mapper. matplotlib'sax.tripcolorthen refused to render the source-plane regularization-weights subplot ininversion_0_0.png, loggingCould not plot the source-plane via the Mapper: The length of c must match either the number of points or the number of triangles. The numerical reconstruction itself was unaffected.API Changes
None — internal indexing fix in
AbstractInversion.regularization_weights_mapper_dict. The property's signature and return type are unchanged; the dict now correctly maps eachMapperto its own regularization weights when other linear objects sit ahead of it inlinear_obj_list. See full details below.Test Plan
python -m pytest test_autoarray/inversion/inversion/test_abstract.py -k regularization_weights_mapper_dict -v— two new regression tests pass.python -m pytest test_autoarray/inversion/— full inversion suite (164 tests) green.vis_pixNautilus fit no longer emitsCould not plot the source-plane via the Mapperlog lines.Full API Changes (for automation & release notes)
Removed
Added
Renamed
Changed Signature
Changed Behaviour
autoarray.inversion.inversion.abstract.AbstractInversion.regularization_weights_mapper_dict— whenlinear_obj_listcontains non-Mapper entries (e.g.LightProfileLinearObjFuncList) ahead of aMapper, the dict slot for that mapper now contains the mapper's own regularization weights rather than the weights of whichever linear obj happened to share its Mapper-only enumerate index. Inversions whoselinear_obj_liststarts with aMapperproduced correct output before and after this change.Migration
🤖 Generated with Claude Code