fractional number of electrons with HF

Clicked A Few Times

1:44:15 AM PDT - Fri, Jul 11th 2014
Dear All, Is it possible to run NWCem Hartree-Fock calculation with specified fractional occupation numbers? For a DFT calculation, this can be done with 'fon' keyword, but it seems not to work for HF. Thank you very much for a suggestion! Best regards, Vladimir.

Forum Vet

9:19:48 AM PDT - Fri, Jul 11th 2014
You can run HF calculations in the DFT module by using the "xc hfexch" line. dft xc hfexch end

Clicked A Few Times

9:49:30 PM PDT - Fri, Jul 11th 2014
Yes, this is the exactly the next thing I'd like to better understand. What is xc hfexch ? Is it DFT with Optimized Effective Potential or, equivalently, EXX ? Then it is not HF (HF is not DFT), which leads me to the original question of how to run HF with fractional occupations.

Forum Vet

12:28:26 PM PDT - Mon, Jul 14th 2014
Quote:Nazarov Jul 11th 8:49 pm Yes, this is the exactly the next thing I'd like to better understand. What is xc hfexch ? Is it DFT with Optimized Effective Potential or, equivalently, EXX ? Then it is not HF (HF is not DFT), which leads me to the original question of how to run HF with fractional occupations. Setting xc hfexch gives you the same exact Hamiltonian as in a Hartree-Fock method. Using fractional occupation in HF has not the same theoretical footing as in DFT, of course.

Clicked A Few Times

5:02:59 PM PDT - Mon, Jul 14th 2014
Very good. Then does NWChem have optimized effective potential (OEP) functionality?

Forum Vet

9:57:09 AM PDT - Tue, Jul 15th 2014
Quote:Nazarov Jul 14th 4:02 pm Very good. Then does NWChem have optimized effective potential (OEP) functionality? You might want to have a look at the following section of the NWChem documentation http://nwchemgit.github.io/index.php/Release62:Density_Functional_Theory_for_Molecules#SIC_-...

Clicked A Few Times

5:25:56 PM PDT - Tue, Jul 15th 2014
This seems to be exactly what I've been looking for. Thank you very much! So if I write dft sic oep end task dft I have a DFT calculation with OEP within KLI approximation. No other xc potential is mixed up to it. Is it correct?

Clicked A Few Times

7:54:55 PM PDT - Wed, Jul 16th 2014
I am sorry for possibly not clearly stating my problem. My problem IS NOT to correct some not self-interaction free xc functional with the use of OEP. It rather is to carry out calculations with OEP, which is known to be self-interaction free. Does NWChem allow this? Thank you very much. Vladimir.

Forum Vet

1:57:27 PM PDT - Thu, Jul 17th 2014
Quote:Nazarov Jul 16th 6:54 pm I am sorry for possibly not clearly stating my problem. My problem IS NOT to correct some not self-interaction free xc functional with the use of OEP. It rather is to carry out calculations with OEP, which is known to be self-interaction free. Does NWChem allow this? Thank you very much. Vladimir. No, We do not have OEP implemented in NWChem

Clicked A Few Times

7:56:19 PM PDT - Thu, Jul 17th 2014
Thank you for an exhaustive answer. I still am curious: Since you eliminate self-interaction from an arbitrary functional with the use of OEP, presumably, in some form you calculate OEP. Cannot this be used for a pure OEP calculation regardless of SIC? More technically (and probably naive): Is not a combination of xc SOME_X WEIGHT sic oep equivalent to just using OEP? I, however, have tried xc slater 0.0 sic oep and it doesn't work for this purpose. Many thanks for elucidating this! Vladimir.

Forum >> NWChem's corner >> NWChem functionality