MPI/SEQ repro issue in dev_2022_biolink
In gls_mixing.F the following lines need to be removed to guarantee the same results between MPI and sequential runs:
!--------------------------------------------------------------------
! Initialization
!--------------------------------------------------------------------
!
#if !defined EXACT_RESTART
IF (FIRST_TIME_STEP) THEN
DO k=0,N
DO j=jmin,jmax
DO i=imin,imax
trb( i, j, k, nstp, itke ) = trb_min(itke)
trb( i, j, k, nstp, igls ) = trb_min(igls)
ENDDO
ENDDO
ENDDO
ENDIF
#endif
This has been done in the dev_2022_biolink_water2sedFluxes branch: 81e63899
Those lines are not in the MASTER branch. If those lines are really needed (why?), they could as well be corrected. Imin,imax,jmin,jmax are probably not the right loop boundaries within the MPI framework.