Once in a while, the test QS/regtest-gapw-4/CO_xastpxfh_150Ry.inp was failing on the local machines due to some numerical inconsistency:
"Difference too large: 4.04e-12 > 4e-12".

It was consistently calculating -92.8473936558286 while the reference value is -92.84739365620344. Should we maybe reduce the threshold slightly?

It was consistently calculating -92.8473936558286 while the reference value is -92.84739365620344. Should we maybe reduce the threshold slightly?

It is possible to adjust the Gold value or (alternatively) the margin. Calibrating the Gold value is more ambitious because the "correct" value aka Gold is usually taken from some SDBG run and then the margin is widened based on PDBG and others. If you adjust the Gold value, you get risk the "error" returns.

psmp test passes with --mpiranks 1
i haven't found the source of the error