Dear everyone, I am relatively new to GSL, but GSL seems perfect for my application in spectral clustering. In these types of problems I typically need to examine the eigenvectors corresponding to either the largest or the smallest eigenvalues of a big symmetric matrix A.
I've experimented with GSL and it seems straightforward in the API to calculate a full decomposition, however since A is usually large (~2000x2000) this seems wastefull in terms of computational expense. Is it possible to calculate only the (say the 5) largest eigenvalues and corresponding eigenvectors in a manner similar to e.g. "eigs" (rather than "eig") in Matlab? --- Ørjan _______________________________________________ Help-gsl mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-gsl
