Fix some Maxwell couplings with MPI
Description
Several issues were caused by improper use of mesh_nearest_point.
This did not fix the similar GPU problem (#793 (closed))
Checklist
-
I have checked that my code follows the Octopus coding standards -
I have added tests for all the new features added in this request.
Merge request reports
Activity
changed milestone to %14.0
added Bugfix label
requested review from @nicolastd
requested review from @martin.lueders
Codecov Report
All modified and coverable lines are covered by tests
Comparison is base (
88dcf7d) 68.28% compared to head (7b2e5a5) 68.28%.Additional details and impacted files
@@ Coverage Diff @@ ## main #2249 +/- ## ======================================= Coverage 68.28% 68.28% ======================================= Files 597 597 Lines 99447 99455 +8 ======================================= + Hits 67905 67914 +9 + Misses 31542 31541 -1
Edited by Octopus Buildbot- Resolved by Nicolas Tancogne-Dejean
@fbonafe There is no MPI-nonMPI splitting in the test maxwell/14, see https://www.octopus-code.org/testsuite/compare/commits/1525/input/14-quadrupole-pot.01-hydrogen-gs.inp
Could you reduce the tolerances of this test to avoid re-introducing a problem. 1e-12 tolerances should be good for this test.
Please check the other tests too, to make them more reliable.
- Last reply by Nicolas Tancogne-Dejean
added 1 commit
- f1f92697 - Adress review comments and fix ispin index problem
added 1 commit
- fa25cd42 - Fix deadlock by moving if below the derivatives_grad call
-
@nicolastd @martin.lueders I addressed your comments, this should be ready to merge.
@fbonafe As asked above, could you reduce the tolerances to 1e-12 for the test 14, as the values now all match and agree among the bots.
