Fix JAX multivariate-normal PSD validation

[FlorianPfaff]

Summary

Fix the JAX random backend's eager covariance validation for multivariate_normal.

The existing JAX CI regression

tests/backend/test_jax_random_multivariate_normal_validation.py::test_multivariate_normal_rejects_invalid_covariance_geometry[bad_cov3-cov must be positive semidefinite]

fails on the current main-based JAX matrix because the indefinite covariance diag(1, -0.1) is not rejected.

Root cause / fix

_validate_multivariate_normal_cov used JAX allclose and eigvalsh for checks that are immediately converted to Python booleans. This is an eager validation path rather than a traced numerical kernel, and relying on JAX linear algebra here makes validation sensitive to backend/runtime behavior.

Convert the already validated finite covariance to a host NumPy array and perform the symmetry and eigenvalue checks with NumPy. Sampling and returned arrays remain JAX-backed.

The existing 1e-8 PSD tolerance and validation messages are unchanged.

Evidence

The failure is present in the existing Python 3.12 and 3.13 JAX CI artifacts from current-main-based PR #5379. The repository already contains the exact regression test, so no duplicate test was added.

Scope

• based directly on main at b7d3edc9ac05600de5345dc20fd6f3db33154d86;
• one production file changed;
• 4 additions / 3 deletions;
• no matching open PR found for this PSD-validation failure.

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Max errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no		no	7.24s
✅ JSON	prettier	7	0	0		0	0.99s
✅ JSON	v8r	7		0		0	5.27s
✅ MARKDOWN	markdownlint	68	0	0		0	1.98s
✅ MARKDOWN	markdown-table-formatter	68	0	0		0	0.44s
✅ PYTHON	black	1989	15	0		0	65.59s
✅ PYTHON	isort	1989	28	0		0	2.96s
✅ REPOSITORY	betterleaks	yes		no		no	1.23s
✅ REPOSITORY	checkov	yes		no		no	34.45s
✅ REPOSITORY	git_diff	yes		no		no	0.14s
✅ REPOSITORY	secretlint	yes		no		no	103.93s
✅ REPOSITORY	syft	yes		no		no	2.98s
✅ REPOSITORY	trivy-sbom	yes		no		no	3.44s
✅ YAML	prettier	11	0	0		0	0.61s
✅ YAML	v8r	11		0		0	17.63s
✅ YAML	yamllint	11		0		0	1.05s

Notices

⚠️ Your configuration references items that have been removed from MegaLinter and are ignored: MAKEFILE_CHECKMAKE. See Removed linters to find their replacements.

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@10.0.0 --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_BETTERLEAKS,REPOSITORY_SECRETLINT,REPOSITORY_SYFT,REPOSITORY_TRIVY_SBOM,YAML_PRETTIER,YAML_YAMLLINT,YAML_V8R

Show us your support by starring ⭐ the repository