I have some very large static data which I use to solve my PDEs with FEniCS. When running my program in parallel via MPI, the data are being loaded multiple times, for each processor. This is very wasteful and sometimes results in maxing out my memory.
Is there a way to load these static data only once and share between processes?