Skip to content

GAMG + processorAgglomeration + interpolateCorrection fails

Summary

Divide by zero when

  • GAMG
  • processorAgglomerator masterCoarsests;
  • interpolateCorrection true;

Steps to reproduce

Any parallel case using GAMG. E.g. pitzDaily tutorial.

What is the current bug behaviour?

Sigfpe since coarseLevel correction is not sent across when processor agglomerating. Instead the coarse-level correction stays zero for the missing/remote elements. This causes a divide by zero when normalising.

What is the expected correct behavior?

No failure.

Relevant logs and/or images

Environment information

  • OpenFOAM version : v2212

Possible fixes

Avoid accessing coarse-level when interpolating correction. Or send across coarse-level correction.