I thought that I remembered a thread either in this forum or on the mailing list regarding the Moore-Penrose inverse (also called the pseudoinverse). I can't seem to locate such a topic, however. Before I recreate such code, can anyone tell me if it has already been implemented and, if so, where I would find it? I would have thought the SVD module but I don't see it there.

See http://listengine.tuxfamily.org/lists.t ... reads.html, the whole thread "How about pseuodinverse?" has several different implementations.

You probably already know this but using pseudo inverse in numerical computation is really bad. It is one of few cases of: you will NEVER get close to the correct answer.