I'm trying to figure out how to use the MPI FEAST eigensolver in MKL. In the C-MPI/3_sparse examples in FEAST, the matrices are created on all processors and not distributed. Is there any way to distribute sparse matrices over all processes to solve for eigenvalues?
It should be noted that in version 11.2, MKL team implemented the Parallel Direct Sparse Solver for Clusters. You can use this solver for Extended Eigensolver RCI interface for computing the distribute computations across many MPI processes. It seems that's all we can suggest to you in that case.