>I think it really depends on what you use the structure for. > >If it's an initial guess to, say, and ab initio optimization then Geoff's proposal makes a lot of sense. > >If the structure and energy terms are to used as such, such as in Henry Rzepa's case, then you really need the "right" parameters. > >I think both are needed. > >Perhaps is Henry could supply the necessary parameters, or even snippets of code, for his systems of particular interest this would speed their implementation up? UFF looks like an excellent general purpose solution to this problem (today I might want Si, tomorrow it would be eg Se). Indeed, this seems an excellent thing to implement for eg the Blue Obelisk concept (and for Peter Murray-Rust's CLAX service architecture proposal). I would find it useful for "approximate 3D refinements". The strength of a mechanics treatment of course is that it cannot break a bond once its defined. QM methods are notoriously weak at this; they tend to rearrange the electrons unless the starting guess is very close to what you want to get. A UFF-CLAX implementation would be a fantastic way of getting a really good 3D coordinate set based purely on connectivity and a reasonable force field (there are other ways of course, but these are based on fragment libraries, and heuristics). -- Henry Rzepa. +44 (020) 7594 5774 (Voice); +44 (0870) 132 3747 (eFax); rzepahs at mac.com (iChat) http://www.ch.ic.ac.uk/rzepa/ Dept. Chemistry, Imperial College London, SW7 2AZ, UK. (Voracious anti-spam filter in operation for received email. If expected reply not received, please phone/fax).