Skip to content

Only share AWH correlation tensor integral over main ranks.

Magnus Lundborg requested to merge ml_fix_awh_shared_friction_freeze into release-2023

The shared correlation tensors are only updated when doing energy output, and then only from the main ranks. The correlation tensor updates would hang if sharing biases over multiple walkers and running more than one MPI rank per walker.

Fixes: #4925 (closed)

Edited by Mark Abraham

Merge request reports