I/O from XDMF/HDF5 files in dolfin-x

dolfinx
23

For anyone out there looking for a may to save and write files in parallel I use PetscBinaryIO now. Syntax goes like :

import PetscBinaryIO
from mpi4py.MPI import COMM_WORLD as comm

io = PetscBinaryIO.PetscBinaryIO(complexscalars=True)
solnAsPetscBiIOVec = function.vector.array_w.view(PetscBinaryIO.Vec)
io.writeBinaryFile(f"function_proc{comm.rank}of{comm.size}-C.dat",
						[solnAsPetscBiIOVec])

function.vector[...] = io.readBinaryFile(f"file_proc{comm.rank}of{comm.size}-C.dat")

Be aware that in docker you need to run the command export PYTHONPATH=/usr/local/petsc/lib/petsc/bin/:$PYTHONPATH first or the above code will fail.

Obviously the files produced are discretisation and proc-dependant. In my experience though stuff goes well as long as every program reads them once.

3 Likes