Hi,
I'm looking for a c++ library for dense and sparse matrices.
I really like the implementation of dense matrices in Eigen3 and I found the syntax really nice. However I found that the support to sparse matrices is not at the same level (to compile some little test codes I defined the token EIGEN_YES_I_KNOW_SPARSE_MODULE_IS_NOT_STABLE_YET ).
I would like to know what are the development plans for the sparse matrices in Eigen3. Will it take long before reaching a good level? Or do you think it is better to use uBlas (badly documented but more mature)?
Thanks
Carlo

hi,

indeed, the sparse module is not as mature as I'd like, but it is still quite usable for some specific tasks. Actually, compared to uBlas I'd still recommend Eigen, I really had some bad experiences with it.

And yes the goal is to have a support for sparse matrices which is as good as for dense one asap.

Am I correct that there is no sparse solver available out of the box - that is, one must install either UMFPACK, SuperLU, TAUCS, etc in order to solve a sparse system in Eigen? If so, which one of these is the easiest to build? I tried to build TAUCS for hours with no success.

Also, do any of these support sparse least squares (where A in Ax=b is non-square)?

Thanks,

David