Hello,

In the context of an ANN implementation we'd like to find the optimal weights also using conjugate gradient CG or gradient descend GD. I found the Eigen documentation corresponding to the CG implementation but it only works for sparse matrices? is there a way to efficiently do it for dense matrices as well? do you have any work in progress in this area or would I have to adapt good known implementations such as Octave's fmincg.m?

TIA,
Best regards,
Giovanni

ConjugateGradient does work with dense matrices.