Correct the stale sparse-operator incompatibility note - #263
Merged
Conversation
The comment said TransformerNUFFT could not be used with `apply_sparse_operator`. That was true only for the legacy pynufft backend's kernel-deconvolved adjoint scale, and PyAutoArray#329 (bd18a769, 2026-05-22) removed the NotImplementedError guard for nufftax. The note had been stale for three months; my previous pass hedged it as "not re-verified" rather than checking. Re-verified 2026-08-22: `apply_sparse_operator` runs with TransformerNUFFT at every scale tried, and its dirty image matches the TransformerDFT one to ~3e-13 relative. Records which backend is actually faster (governed by N_vis x N_pix, crossover ~1e7) so the note carries the useful fact rather than a retracted one. The script still uses the plain TransformerNUFFT + direct forward NUFFT route, now for a stated reason — it exercises a different code path from Path B rather than duplicating it — instead of implying it has no choice.
Merged
2 tasks
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
scripts/interferometer/jax_likelihood/rectangular_sparse.pysaidTransformerNUFFTcould not be used withapply_sparse_operator. That was true only for the legacy pynufft backend's kernel-deconvolved adjoint scale, and PyAutoLabs/PyAutoArray#329 (bd18a769, 2026-05-22) removed theNotImplementedErrorguard for nufftax. The note had been stale for three months.My previous pass (#261) hedged it as "not re-verified" rather than checking. This checks.
Re-verified 2026-08-22:
apply_sparse_operatorruns withTransformerNUFFTat every scale tried, and its dirty image matches theTransformerDFTone to ~3e-13 relative. The note now carries the useful fact instead of a retracted one, including which backend is actually faster (governed byN_vis × N_pix, crossover ~1e7).The script still uses the plain
TransformerNUFFT+ direct forward NUFFT route — now for a stated reason, that it exercises a different code path from Path B rather than duplicating it, instead of implying it has no choice.Companion to PyAutoLabs/PyAutoArray#479 and PyAutoLabs/autolens_workspace#498.
Scripts Changed
scripts/interferometer/jax_likelihood/rectangular_sparse.py— comment only; no executable changeTest Plan
python -m py_compilecleanrectangular_sparse.pyis not insmoke_tests.txt, so this gate does not execute it either wayGenerated by Claude Code