Repository navigation
fix: avoid overflow in NumericDiff for finite inputs - #237
Open
Deepanshu Pal (deepanshupal) wants to merge 4 commits into
Open
Deepanshu Pal (deepanshupal) wants to merge 4 commits into
Deepanshu Pal (deepanshupal) wants to merge 4 commits into
Conversation
This branch has not been deployed
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
NumericDiff can overflow its intermediate difference or magnitude sum even when both inputs are finite. Comparing the largest float with half that value returns a perfect score instead of 2/3. Comparing it with its negative returns NaN. Both Python and TypeScript have the same problem.
Scale both inputs by their largest absolute value before computing the normalized difference. This keeps the intermediates bounded without changing the formula for finite inputs, and preserves the zero/zero result.
Matching tests in both languages cover large positive and negative values, identical values, subnormals, zeros, and ordinary values. Three cases fail on the unchanged implementation and all eight pass with the fix. All offline tests pass (Python: 79; TypeScript: 78), and the JS build succeeds. Live API tests require credentials and were not run successfully.