SobolReporter total effect index results greater than one or large negative numbers #31905
Description
Bug Description
When running a sensitivity analysis using Sobol sampling and the Sobol reporter to output Sobol indices, some of the results were unexpected with total effect Sobol indices greater than one and some very low negative values (< -100). When compared to other codes, MOOSE results are very different for the same number of samples.
Steps to Reproduce
This was done on a 1D TRISO input file varying the layer thicknesses, buffer density, and some material properties and calculating the Sobol indices for IPyC/SiC stress and kernel and SiC temperature. LHS sampling was used with 1000 samples in both the sample and re-sample matrices.
Impact
The total effect Sobol indices are not reliable for my problem, so I have to rely on other codes for these results. It would be preferable to have everything calculated using MOOSE.
[Optional] Diagnostics
No response