Eigenvalue Probem

Question

I'm trying to diagonalize a matrix in julia, but the matrix is itself too large to calculate. But I can efficiently calculate the matrix vector product. I'm looking for a solution so that I can use the matrix vector product into eigs and solve using lanzcos method. Is there a package that gives me an output in the form of matrix vector product?

Can someone help here?

Answer 1

You can use the ARPACK based eigs function which is included in Julia 0.6.2 or you can use a pure Julia implementation of a related method in https://github.com/JuliaMath/IterativeSolvers.jl. The latter isn't Lanczos based but I guess it doesn't really matter as long as it gives you the eigenvalues.