Fix Hi L2 flux correction convergence logic

[tmplummer]

Change Summary

Fixes a bug in how convergence criteria was being calculated. The bug occurred when an input flux is zero causing the computed chi_n for that pixel to be NaN which will never be less than the convergence criteria. The easy fix is to just use a nanmean when computing the mean ratio. This is appropriate because zero fluxes will just stay zeros and thus are already converged.

Testing

I first added the test confirming that when a zero is in the input flux, the max iterations is hit. Once the bug was fixed, the test passes.

Closes: #3142

[Copilot]

Pull request overview

This PR updates the Hi ENA flux-correction convergence logic so zero-valued flux bins do not poison the per-pixel convergence metric with NaNs. It also adds a regression test in the ENA maps test suite for a zero-flux scenario.

Changes:

• Replaced np.mean with np.nanmean in predictor_corrector_iteration() when computing the convergence ratio.
• Added a Hi-specific regression test intended to verify convergence when one ESA flux is set to zero.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

File	Description
imap_processing/ena_maps/utils/corrections.py	Adjusts the predictor-corrector convergence calculation to ignore NaN ratio entries.
imap_processing/tests/ena_maps/test_corrections.py	Adds a regression test for zero-flux convergence behavior in the Hi correction path.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[bryan-harter]

Seems good to me!