Actually the problem is:

Calculating all eigenvalues and eigenvectors for a huge real symmetric sparse matrix. (it might be of the order of millions, and about 5% of the elements will be non-zero)

I want a C or C++ library to do this on linux.

You might ask "why ALL eigenvalues", but I really do need all of them.

I have tried ARPACK++, but it takes way too much time to calculate all eigenvalues. Intel mkl doesnt have subroutines to do this for sparse matrices. (on a matrix in CSR,CSC or some other sparse representation)

I have read about SLEPc, but it seems to have too many dependencies (PETSc,etc.). It will be great if you could suggest other libraries that can do this. But still if in your opinion, SLEPc is the most efficient for this problem, please do tell.


Thank you for your help in advance.