1
100%
develop: 100%

Ran 15 Aug 2025 09:03PM UTC

Files 18

Run time 0s

Badge

Embed ▾

Committed 15 Aug 2025 08:58PM UTC coverage: 99.582%. Remained the same

Job # 16999260755.1

Build Type

push

github

Committed by

web-flow

Commit Message

Fixing test suite after regression of scipy.stats.multivariate_normal in scipy-1.16.0 (#1102)

Closes #902

We used to use scipy.stats.multivariate_normal to test fitting in 2D.
For binned fits, this requires the cdf of the model, and there is no
closed form. In the past, that was okay, because the numerical
integration routine in scipy was apparently good enough. But that
changed in 1.16.0 when they switched to a worse-performing
implementation - in regard to numerical accuracy and speed, see
https://github.com/scipy/scipy/issues/23469.

The workaround is to replace scipy.stats.multivariate_normal with a
factorized version which can be implemented based on the pdfs and cdfs
of two 1D normal distributions.

As a drive-by fix, I should have finally fixed
test_Template_with_model_2D. I don't know why, but I used an extreme
coarse binning, which must have caused the numerical issues on some
platforms.

---------

Co-authored-by: pre-commit-ci[bot] <66853113+pre-commit-ci[bot]@users.noreply.github.com>

Run Details

3816 of 3832 relevant lines covered (99.58%)

1.0 hits per line

scikit-hep / iminuit / 16999260755 / 1
100%
develop: 100%

README BADGES
x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 16999260755.1

scikit-hep / iminuit / 16999260755 / 1 100% develop: 100%

README BADGES x

Markdown

Textile

RDoc

HTML

Rst

Source Files on job 16999260755.1

scikit-hep / iminuit / 16999260755 / 1
100%
develop: 100%

README BADGES
x