Firefly and PC GAMESS-related discussion club

Learn how to ask questions correctly

Re^3: H_eff in XMCQDPT

Alex Granovsky

Dear Sanya,

>Thank you very much. Actually, I'm currently dealing with homo- and heteroleptic metal complexes, where several quasidegenerate excited state of the same nature are possible (say, excitations involving Ligand1, Ligand2, and Ligand3).

I think you simple math may work for supra-molecules containing
several quasi-degenerate ligands. Take care to include all quasi-
degenerated states into model space!

> The systems have up to 1000 basis functions, and each XMCQDPT2 calculation takes up to 3-4 days on a quad-core PC (and a lot of disk I/O). So, it is not easy to experiment with increasing dimension of Heff :)

You could opt for latest Firefly 8.0.0 beta which has improved XMCQDPT2 code.

> As for the check of the overlap matrix of non-orthogonal eigenvectors, it's a good idea, thanks. So, if off-diagonal values are small and all eigenvalues of non-symmetric Heff are real, it's no need to further increase Heff, isn't it?

No, the idea is that if they are large and/or some eigenvalues
are complex, your XMCQDPT2 job is definitely running into problems.  

>By the way, in the next FF version, is it possible to print the WARNING that off-diagonal values are large or some eigenvalues of non-symmetric Heff are complex?

Is it too difficult to check this manually in the output? :)


>On Tue May 31 '11 9:34pm, Alex Granovsky wrote
>>Dear Sanya,

>>I'm afraid XMCQDPT2 with dimension of model space equal
>>to ca. 30-50 as you suggest will be prone to intruder
>>state problem. I'd not use more than ca. 20 states for
>>organic pi-systems. Unfortunately there are no general
>>recipe here. You can try to perform several calculations
>>with increasing dimension of Heff and check
>>how stable the transition energies of interest are. Another
>>thing to check is the overlap matrix of non-orthogonal
>>eigenvectors corresponding to Heff before symmetrization.
>>There should not be large off-diagonal values here, and
>>all eigenvalues of non-symmetric Heff must be
>>real numbers.

>>On Sun May 29 '11 5:11pm, sanya wrote
>>>How to estimate the size of H_eff in XMCQDPT (nstate or kstate keys)? I know that for 1 excited state H_eff of 10-12 states is enough, but typically I need 3-4 excited states. Does this proportion keeps for higher-lying states?

[ Previous ] [ Next ] [ Index ]           Wed Jun 1 '11 7:00pm
[ Reply ] [ Edit ] [ Delete ]           This message read 733 times