Hi,

I have used Eigen for half a year, but only limited to dence matrixes.
However, I need to solve eigenvalue decompozation of large sparse matrix, is Eigen able to do this ?
I am very appreciate for your help.

My platform is VS2015, using C++.

Thank you

P.S. I have noted that Eigen has already include Arpack in to its Unsupported list: http://eigen.tuxfamily.org/dox/unsuppor ... odule.html
However, there is little infomation nor example for me to use it.

Try this lib: https://github.com/yixuan/spectra it's header only and built on top of Eigen.

ggael wrote:Try this lib: https://github.com/yixuan/spectra it's header only and built on top of Eigen.

Thank you so mush. Your answer is very helpful.

ggael wrote:Try this lib: https://github.com/yixuan/spectra it's header only and built on top of Eigen.

Hi Gael,

Thank you for your answer.
However, after playing around your suggested Spectra library, I found it is based on dense matrixes. That is, filling the sparse matrixes into a dense one before computing eigenvalues. This consums lots of time.

Do you have any other suggestsions? Or detail introduction/exmple of the unsupported ArpackSupport Module?

Thank you.

no no, it does support truly sparse matrices!! See the second example of the main page "Sparse matrix is supported via the SparseGenMatProd class." Also, the API doc does not seem to be uptodate, it does support all shift and invert mode for sparse matrices, see: https://github.com/yixuan/spectra/tree/ ... lude/MatOp

ggael wrote:no no, it does support truly sparse matrices!! See the second example of the main page "Sparse matrix is supported via the SparseGenMatProd class." Also, the API doc does not seem to be uptodate, it does support all shift and invert mode for sparse matrices, see: https://github.com/yixuan/spectra/tree/ ... lude/MatOp

Thank you very much.
Problem solved.