Stabilize piecewise-constant extreme-weight normalization

[FlorianPfaff]

Bug

PiecewiseConstantDistribution accepts arbitrary nonnegative finite interval weights and normalizes them internally. The constructor currently computes mean(w) before normalization. For valid weights near the backend dtype maximum, that reduction can overflow to inf; the subsequent division then collapses the represented density toward zero instead of preserving the input weight ratios.

For example, [max_float, max_float / 2, 0] has a perfectly well-defined normalized density, but the direct mean can overflow even though every input is finite.

Fix

Normalize after scaling by the largest input weight. The scaling division is split through sqrt(max_weight) so backends that lower division through reciprocals do not underflow the reciprocal of a near-maximum finite value.

This preserves the existing normalization formula and all ordinary-scale behavior while avoiding the unnecessary large intermediate reduction.

Regression coverage

Adds tests/distributions/test_piecewise_constant_extreme_weights.py, which constructs the distribution from backend-dtype maximum finite weights and verifies:

• the normalized density heights preserve the expected 2:1 ratio;
• the exact piecewise integral remains one;
• construction succeeds under NumPy overflow/invalid/divide traps.

Scope

• based directly on main at baac3b1736bb8e81eafc1e3f69e10104636103b9;
• 3 small commits ahead / 0 behind at the current head;
• one production file changed (16 additions / 4 deletions);
• one focused 32-line regression test added;
• no overlap found with the currently open PyRecEst bug-fix PRs.

[github-actions]

✅MegaLinter analysis: Success

Descriptor	Linter	Files	Fixed	Errors	Max errors	Warnings	Elapsed time
✅ COPYPASTE	jscpd	yes		no		no	7.41s
✅ JSON	prettier	7	0	0		0	0.78s
✅ JSON	v8r	7		0		0	5.79s
✅ MARKDOWN	markdownlint	68	0	0		0	2.01s
✅ MARKDOWN	markdown-table-formatter	68	0	0		0	0.4s
✅ PYTHON	black	1989	15	0		0	58.01s
✅ PYTHON	isort	1989	28	0		0	2.79s
✅ REPOSITORY	betterleaks	yes		no		no	1.04s
✅ REPOSITORY	checkov	yes		no		no	34.07s
✅ REPOSITORY	git_diff	yes		no		no	0.15s
✅ REPOSITORY	secretlint	yes		no		no	98.38s
✅ REPOSITORY	syft	yes		no		no	2.72s
✅ REPOSITORY	trivy-sbom	yes		no		no	2.61s
✅ YAML	prettier	11	0	0		0	0.61s
✅ YAML	v8r	11		0		0	16.33s
✅ YAML	yamllint	11		0		0	0.83s

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