Skip to content

Fix cluster moments computation

Pavol Strizenec requested to merge pavol/athena:fix_ClusterMoments into master

It was found, that differences in cluster moments, seen between rel. 21 and rel. 22, are coming from change in Eigen computation of vector eigenvalues, where rel. 22 version have small non-zero values for the same input as in rel. 21 where values were exactly zero. This MR is changing a test == 0.0 to < epsilon.

MR will of course fail the Tier0 tests, but that's expected and should bring results back to values in rel. 21.

Merge request reports