I am trying to find all eigenvectors (or at least a couple sorted by eigenvalues) of a sparse matrix. However, looking at cuSolver, the closest thing is the cusolverSpScsreigvsi solver - this however only tries to find one eigenvalue and eigenvector around some guess. Since my eigenvalues will be multiply degenerate, I’m not convinced it will work for me.
Is there any other way to do it?
There seem to be functions in cuSolverDN (dense) that do exactly that, however my matrices will be too large to store in memory directly.