Skip to content

feat(interferometer): from_fits accepts raise_error_dft_visibilities_limit - #311

Merged
Jammy2211 merged 1 commit into
mainfrom
feature/datacube-hannah-preset
May 15, 2026
Merged

Jammy2211 merged 1 commit into
mainfrom
feature/datacube-hannah-preset

Conversation

@Jammy2211

Copy link
Copy Markdown
Collaborator

Summary

Thread raise_error_dft_visibilities_limit through Interferometer.from_fits to its constructor. Previously the kwarg was only reachable via the constructor — callers loading FITS data had no way to opt out of the >10,000-visibility DFT safety check.

This unblocks the JAX-profiling workflow against Hannah Stacey's ALMA cube (16,984 visibilities × 34 channels) where the JIT path intentionally uses TransformerDFT because pynufft is not JAX-traceable. With this kwarg, the profiler can load the dataset via the standard from_fits API instead of replicating its internals.

API Changes

Added one optional keyword argument (raise_error_dft_visibilities_limit: bool = True) to autoarray.Interferometer.from_fits. The default matches the existing constructor behaviour — no migration needed for existing callers. See full details below.

Test Plan

  • pytest test_autoarray/dataset/interferometer/test_dataset.py::test__from_fits__raise_error_dft_visibilities_limit__threads_kwarg -xvs passes.
  • Existing test_autoarray/dataset/interferometer/test_dataset.py suite still passes.
  • Verify CI green.
Full API Changes (for automation & release notes)

Added

  • autoarray.Interferometer.from_fits(..., raise_error_dft_visibilities_limit: bool = True) — new optional kwarg that forwards to the Interferometer constructor. Set to False to bypass the 10,000-visibility safety guard when loading a DFT-based interferometer dataset from FITS (use case: profiling the JAX-traceable DFT path at ALMA-scale before a JIT-friendly NUFFT is available).

Migration

  • None — default behaviour unchanged.

🤖 Generated with Claude Code

…limit

Thread the existing constructor kwarg through `Interferometer.from_fits` so
callers loading >10,000-visibility datasets with TransformerDFT (e.g. when
profiling the JAX-traceable DFT path at ALMA-scale before nufftax is wired
in) can opt out of the safety guard without reconstructing the dataset by
hand. Default behaviour unchanged.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
@Jammy2211

Copy link
Copy Markdown
Collaborator Author

Workspace PR: PyAutoLabs/autolens_workspace_developer#63

@Jammy2211
Jammy2211 merged commit 3ddf368 into main May 15, 2026
6 checks passed
@Jammy2211
Jammy2211 deleted the feature/datacube-hannah-preset branch May 15, 2026 14:20
@Jammy2211 Jammy2211 removed the pending-release PR queued for the next release build label Sep 4, 2026
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant