Validate Gaussian mixture covariance shapes

[FlorianPfaff]

Bug

GaussianMixture.mixture_parameters_to_gaussian_parameters(...) accepted a plain covariance matrix with shape (dim, dim) without checking how many mixture components were supplied.

For one two-dimensional component, NumPy broadcasting changed a covariance such as

[[4.0, 1.5],
 [1.5, 9.0]]

into a three-axis intermediate with the wrong orientation. Summing over axis 2 produced [5.5, 10.5], which was then broadcast into the returned covariance

[[5.5, 10.5],
 [5.5, 10.5]]

instead of preserving the component covariance. For multiple components, the same unvalidated layout could likewise produce shape-dependent broadcasting or failures unrelated to the actual input contract.

Fix

Normalize and validate the mixture parameter layout before moment conversion:

• normalize scalar and one-dimensional mean inputs to the established (n_components, dim) representation;
• require covariance tensors to have shape (dim, dim, n_components);
• accept a natural (dim, dim) matrix only for a single component and reshape it explicitly;
• preserve scalar/vector variance shorthand for one-dimensional mixtures;
• reject ambiguous multi-component covariance matrices with a clear ValueError instead of allowing broadcasting.

Regression coverage

The constructor test module now verifies that:

• a single two-dimensional covariance matrix is preserved exactly;
• a single (dim, dim) matrix is rejected when two component means are supplied;
• the existing two-component one-dimensional conversion still returns mean 1.5 and covariance 3.25.

Validation

• independently reproduced the pre-fix corrupted covariance [[5.5, 10.5], [5.5, 10.5]];
• exercised the patched shape normalization and recovered the exact input covariance [[4.0, 1.5], [1.5, 9.0]];
• verified the existing one-dimensional mixture calculation remains unchanged;
• verified the ambiguous multi-component matrix is rejected;
• branch comparison: 2 commits ahead, 0 behind main;
• final diff is limited to the Gaussian-mixture converter and its focused constructor tests.

GitHub Actions provides the authoritative full backend, lint, packaging, documentation, and integration validation.

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no	no	26.44s
✅ JSON	prettier	7	0	0	0	1.14s
✅ JSON	v8r	7		0	0	4.73s
✅ MARKDOWN	markdownlint	68	0	0	0	1.82s
✅ MARKDOWN	markdown-table-formatter	68	0	0	0	0.69s
✅ PYTHON	black	1865	89	0	0	91.25s
✅ PYTHON	isort	1865	141	0	0	2.68s
✅ REPOSITORY	betterleaks	yes		no	no	2.26s
✅ REPOSITORY	checkov	yes		no	no	53.15s
✅ REPOSITORY	gitleaks	yes		no	no	17.26s
✅ REPOSITORY	git_diff	yes		no	no	0.33s
✅ REPOSITORY	secretlint	yes		no	no	81.56s
✅ REPOSITORY	syft	yes		no	no	6.16s
✅ REPOSITORY	trivy-sbom	yes		no	no	6.13s
✅ REPOSITORY	trufflehog	yes		no	no	31.38s
✅ YAML	prettier	11	0	0	0	0.79s
✅ YAML	v8r	11		0	0	12.23s
✅ YAML	yamllint	11		0	0	0.5s

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