How can I perform truncated SVD using Eigen? I need only the top 5-10 eigenvalues/ eigenvectors for a large dense matrix. Would like to avoid the full JacobiSVD. Thanks

there is no such algorithm in Eigen yet.

On further digging, redsvd https://code.google.com/p/redsvd/ is doing the trick for my application. It is based on Eigen so very easy to use with the rest of my project without involving other dependencies. Thanks.