[molpro-user] Questiion about stable=opt condition in gaussian

Kirk Peterson kipeters at wsu.edu
Sun Sep 18 17:22:53 BST 2011

Dear Anirban,

it sounds to me like in the end Gaussian is optimizing to a symmetry-broken solution.  In Molpro you could try reducing the symmetry from D2h to C2v and giving it the occupations that Gaussian ends up with.  Probably though you really don't want this since it's presumably just an artifact of using a UHF wavefunction in this case.

best regards,


On Sep 17, 2011, at 11:47 AM, anirban mandal wrote:

> Hi all,
> I have a quick question about the calculations in Gaussian (98 and 03) using the condition "stable=opt". I did an UHF calculation for a Ru2 triplet using both Molpro 2010 and Gaussian (same basis set). First the energies didn't match, because they didn't converge to the same state. So I did dome orbital reordering in Gaussian to get the same state as obtained by Molpro, and the energies matched.
> Now the problem is, there is an option in Gaussian called "stable=opt". If specified, it checks for internal and external instabilities of the wavefunction and if instabilities found, it again minimizes the energy. If I use that option, I do get lower energy than Molpro or Gamess everytime, but the orbitals now look really really strange. It looks like the orbitals get localized. For example, if I have two Ru atoms, the mo's have contributions from the s, p , and d orbitals of both Ru atoms. One the calculations are with stable=opt, some orbitals have contribution from one Ru atom, others have contribution from the other Ru atom. I don't know how to reproduce those orbitals in Molpro. I was wondering if someone had the same problem before, and how to actually handle it. Thank you guys.
> Best regards,
> Anirban.
> _______________________________________________
> Molpro-user mailing list
> Molpro-user at molpro.net
> http://www.molpro.net/mailman/listinfo/molpro-user

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.molpro.net/pipermail/molpro-user/attachments/20110918/07b9a617/attachment.html>

More information about the Molpro-user mailing list