Loosen kap/mle regression gate to hardware-reproducible tolerance - #28
Merged
Conversation
scipy's kappa4 MLE converges to the same optimum with ~1e-7 floating-point drift across CPU types: the identical commit passed the 1e-9 gate on the PR runner and failed it on the master push runner (max abs diff 2.6e-7, kap/mle only). Give that one combination an abs=1e-5 tolerance - still far below the 1e-3+ shift any real numerical change produces - and keep every other combination at rel=1e-9.
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes the
test (3.13)failure on the master CI run after #27 merged.Not a Python 3.13 issue: the identical commit passed 3.13 on the PR run. The
kap-mleregression test compares against golden values at rel=1e-9, but scipy's kappa4 MLE optimizer lands on the same optimum with hardware-dependent floating-point drift of ~1e-7 across runner CPU types (max abs diff on the failing run: 2.6e-7).This gives that one combination an abs=1e-5 tolerance — still far below the 1e-3+ shift any genuine numerical change produces — and keeps every other (distribution, fit) combination at rel=1e-9. Documented inline in the test module.