Hello -

I'm interested in working with low-rank matrices, specifically projection operators which look like outer products p * p^T . My system is sparse, so what I really need is a matrix object to represent a rank-1 (or sum of rank-1) matrices. Does such a thing exist in Eigen? Is there another strategy to project a Vector onto some subspace? Any experience other users have had would be helpful!

I think that there is already a thread on that topic.

In short, you can simply explicitely write 8p * p.transpose()) * v to apply it. To make it more convenient, you might emcapsulate it into a wrapper structure storing the matrix p, with the appropriate operator* overloads.

Thanks - that is what I ended up doing, and sorry I didn't search the archives hard enough. I figured if there was a built in capability it might be faster than my homebrew operator!