question on reading ASCII orbitals for casscf calculation
seth at spawn.scs.uiuc.edu
Mon Oct 21 16:59:14 BST 2002
Yes, it runs like a champion. The key is invoke the mcscf engine with either MCSCF or CASSCF rather than MULTI, which MATROP will interpret as a matrix multiplication command. I haven't yet tried to convert the orbitals to GAMESS or G98 format, but I'm quite sure it would not be hard (although there may be an issue re: differences between MolPro GAMESS contraction schemes: check the molpro-user archive). The orbitals from MOLPRO come out in five f15.8 fields/line separated each by commas. I don't think
the commas are necessary to read them back in (the manual says they're free format, but I haven't really tested things too aggressively).
I'm not sure what version of Molden we have here, but it does work to view the orbitals. Recently, I've been using Molekel, as the pictures it produces look a little nicer. Molekel will read the molden-style files that MolPro puts out, though you have to reorder the atoms in the appropriate section of the file in order to make it work. Basically, you want the atoms ordered in the same way that MolPro lists them in the 'ATOMIC COORDINATES' section of the output. It's would not be to hard to write a code
to do this for you. If you use Molekel to view CAS orbitals, be aware that that program lists orbitals by ascending orbital energy, while MolPro lists them by occupation number (the more reasonable way for such a calculation). Thus, the orbitals in your active space may not be the highest-lying orbitals that Molekel lists, though they will often be so.
Olga Dmitrenko wrote:
> Dear Seth,
> I liked your question so much. So, what did you do finally to get your job
> running? I have similar problem with my calculations - I need to do first HF
> or DFT, have a look at the orbirals, maybe rotate them and only then I need
> to go to casscf and rspt2.
> Second question. How do you look at your orbitals. Molden? Which version?
> I found that old versions of molpro and molden work well together, but I
> have troubles with newer molpro and same old molden.
> Maybe you know the answer and it'll save us a lot of time.
> And the last question. From the fact that you are able to save orbitals into
> the text file, I came to the conclusion that it is quite possible to
> convert orbitals from Gamess or G98 to molpro format. Maybe you did it
> already? It seems to me that in Gamess or G98 it's much easier to operate
> with the orbitals.
> I would highly appreciate if you answer my questions. But I would not mind
> if not.
> Olga Dmitrenko
> Postdoc fellow
> Univ of Delaware
> Dept of Chem&Biochem
More information about the Molpro-user