Closed psakievich closed 10 months ago
I think I fixed it.
@psakievich I'm having trouble figuring out if it's parallel consistent. You want to call provide_output on every mpi rank, but it looks like impl_negative_jacobian_check only throws on ranks where it finds a negative volume. I'm having trouble seeing where there's an allreduce to make the result of impl_negative_jacobian_check global. If that's in there, then it's all good.
@alanw0 ah you are right. Still need the all-reduce
@alanw0 will this be parallel consistent? Looking at it now, I think not.