Dear all,
I use FEniCSx to solve the incompressible NS equations with low Reynolds number in parallel (8 cores) and save the solutions using xdmf functions.
I want to load the solutions as the initial condition for a simulation with a higher Reynolds number.
How to load the xdmf file in parallel and set them as the initial data?
Right now I don’t think it is possible to read functions in parallel from a XDMF file. The way to do it would be to use checkpointing io to restore the saved solutions but checkpointing is not implemented yet in dolifinx (see this issue).
A workaround would be to read your solutions from a single process with meshio, assuming you are using CG1 or DG0 elements, as I explained here, and then distribute your data manually with MPI. dolfinx might have helper functions to handle that kind of communications, but I doubt they are available from the python API.
Obviously, you could also simply skip writing solutions to a file, and run your other simulation right after the first one, so you still have you previous solution in memory.