> I am wondering if it is appropriate to perform a structural relaxation > in a non-spinpolarized setting and then continue with a spinpolarized > calculations? Or should one always start with a spinpolarized setting > from the very beginning?
I've encountered several cases where quite a different relaxed structure was found for a non-spinpolarized and spinpolarized case (although 'quite different' depends of course on the kind of feature you are interested in). Hence, whenever you are interest in more than just a rough idea about the relaxed structure, I would say you'd better start with the kind of polarization you want to have in the end. Stefaan
