Stabilize Complex Watson fitting for extreme weights

[FlorianPfaff]

Summary

• rescale finite nonnegative weights before summing them or forming the weighted Complex Watson scatter matrix
• preserve the estimator's invariance to multiplying all weights by a common positive factor
• add a regression using finite weights near the float64 limit

Bug

ComplexWatsonDistribution.estimate_parameters previously evaluated both sum(weights) and Z * weights[:, None] at the caller's raw scale. Individually finite weights could therefore overflow in the reduction or matrix product. For example, weights proportional to [8, 4, 2, 1] but scaled so the largest value is float64.max produced an infinite total and a NaN scatter matrix, even though common weight scaling must not change the estimate.

Fix

Normalize weights by their largest value before computing the total and weighted scatter. The scaled weights lie in [0, 1]; their sum and weighted samples remain bounded, while the final normalization is algebraically identical to the original formula.

Validation

• reproduced the pre-fix overflow with finite extreme weights
• independently verified the stabilized scatter equals the ordinary-scale reference
• added tests/distributions/test_complex_watson_extreme_weights.py, checking phase-invariant mean direction and concentration parameter equality
• branch is two commits ahead of main, zero behind; changes one production file and adds one focused regression test

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no	no	15.05s
✅ JSON	prettier	7	0	0	0	0.76s
✅ JSON	v8r	7		0	0	2.71s
✅ MARKDOWN	markdownlint	68	0	0	0	1.03s
✅ MARKDOWN	markdown-table-formatter	68	0	0	0	0.46s
✅ PYTHON	black	1490	184	0	0	56.61s
✅ PYTHON	isort	1490	329	0	0	1.55s
✅ REPOSITORY	betterleaks	yes		no	no	1.54s
✅ REPOSITORY	checkov	yes		no	no	37.35s
✅ REPOSITORY	gitleaks	yes		no	no	10.41s
✅ REPOSITORY	git_diff	yes		no	no	0.11s
✅ REPOSITORY	secretlint	yes		no	no	32.79s
✅ REPOSITORY	syft	yes		no	no	3.35s
✅ REPOSITORY	trivy-sbom	yes		no	no	4.23s
✅ REPOSITORY	trufflehog	yes		no	no	23.17s
✅ YAML	prettier	11	0	0	0	0.42s
✅ YAML	v8r	11		0	0	7.37s
✅ YAML	yamllint	11		0	0	0.4s

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

Show us your support by starring ⭐ the repository