Skip to content

Avoid overflow when symmetrizing ellipse matrices - #5178

Closed
FlorianPfaff wants to merge 3 commits into
mainfrom
chatgpt/fix-ellipse-symmetrization-overflow-20260805
Closed

Avoid overflow when symmetrizing ellipse matrices#5178
FlorianPfaff wants to merge 3 commits into
mainfrom
chatgpt/fix-ellipse-symmetrization-overflow-20260805

Conversation

@FlorianPfaff

@FlorianPfaff FlorianPfaff commented Aug 5, 2026

Copy link
Copy Markdown
Owner

Summary

  • scale each operand before averaging an ellipse matrix with its transpose
  • avoid intermediate overflow for valid finite matrices near the floating-point limit
  • add a backend-portable strict-overflow regression test

Bug

0.5 * (matrix + matrix.T) can overflow while forming the intermediate sum even when the mathematically averaged result is finite. This affects the public pyrecest.tracking.symmetrize helper and all ellipse covariance/extent paths that use it.

Fix

Compute 0.5 * matrix + 0.5 * matrix.T instead. The result is algebraically equivalent, backend-friendly, and does not double finite entries before scaling.

Validation

The regression uses np.finfo(np.float32).max entries and runs under strict NumPy overflow handling. The previous implementation produces an overflow before scaling; the revised implementation preserves the finite matrix exactly across the supported backends.

@FlorianPfaff
FlorianPfaff marked this pull request as draft August 5, 2026 09:20
@github-actions

github-actions Bot commented Aug 5, 2026

Copy link
Copy Markdown
Contributor

MegaLinter analysis: Success

Descriptor Linter Files Fixed Errors Warnings Elapsed time
✅ COPYPASTE jscpd yes no no 26.93s
✅ JSON prettier 7 0 0 0 1.21s
✅ JSON v8r 7 0 0 3.36s
✅ MARKDOWN markdownlint 68 0 0 0 1.89s
✅ MARKDOWN markdown-table-formatter 68 0 0 0 0.73s
✅ PYTHON black 1866 87 0 0 92.41s
✅ PYTHON isort 1866 140 0 0 2.76s
✅ REPOSITORY betterleaks yes no no 2.44s
✅ REPOSITORY checkov yes no no 54.32s
✅ REPOSITORY gitleaks yes no no 17.41s
✅ REPOSITORY git_diff yes no no 0.3s
✅ REPOSITORY secretlint yes no no 85.65s
✅ REPOSITORY syft yes no no 6.22s
✅ REPOSITORY trivy-sbom yes no no 6.44s
✅ REPOSITORY trufflehog yes no no 32.4s
✅ YAML prettier 11 0 0 0 0.77s
✅ YAML v8r 11 0 0 11.89s
✅ YAML yamllint 11 0 0 0.62s

Notices

📣 MegaLinter 9.5.0 is out! Discover the new features and security recommendations in the release announcement. (Skip this info by defining SECURITY_SUGGESTIONS: false)

See detailed reports in MegaLinter artifacts

Your project could benefit from a custom flavor, which would allow you to run only the linters you need, and thus improve runtime performances. (Skip this info by defining FLAVOR_SUGGESTIONS: false)

  • Documentation: Custom Flavors
  • Command: npx mega-linter-runner@9.6.0 --custom-flavor-setup --custom-flavor-linters PYTHON_BLACK,PYTHON_ISORT,COPYPASTE_JSCPD,JSON_V8R,JSON_PRETTIER,MARKDOWN_MARKDOWNLINT,MARKDOWN_MARKDOWN_TABLE_FORMATTER,REPOSITORY_CHECKOV,REPOSITORY_GIT_DIFF,REPOSITORY_GITLEAKS,REPOSITORY_BETTERLEAKS,REPOSITORY_SECRETLINT,REPOSITORY_SYFT,REPOSITORY_TRIVY_SBOM,REPOSITORY_TRUFFLEHOG,YAML_PRETTIER,YAML_YAMLLINT,YAML_V8R

MegaLinter is graciously provided by OX Security
Show us your support by starring ⭐ the repository

Copy link
Copy Markdown
Owner Author

Closing as superseded by #5239. Both PRs make the identical production change in tracking/ellipse_geometry.py; #5239 is the newer replacement and additionally exercises integer-dtype overflow through the public covariance projection path.

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