Spin Contamination "Rule of Thumb?"

[Stephen]

Hi All, 

I'm performing a few calculations of Ni(II) atom supported on a metal-organic framework. The Ni(II) atoms can exhibit different spin states, which means I must run UKS T while varying MULTI to account for all the configurations. 

Is there a specific rule for spin contamination when looking at the S**2? For example, 

  Ideal and single determinant S**2 :                   12.000000      12.019722

I've assumed that this particular structure has no spin contamination, I have some other calculations where the difference between the ideal and single determinant spin states is much larger (Delta_S**2 = ~2.00). 

Does anyone have an appropriate cutoff for spin contamination? 

With much appreciation, 

Stephen 

[Vladimir Rybkin]

Dear Stephen,

the "cutoff" is what you find reasonable. To me, 12.019722 is practically no contamination, whereas delta=2.0 is a large one and the results are likely unreasonable. At least, you will not likely be able to get the latter published.

Yours,

Vladimir

пятница, 22 января 2021 г. в 20:55:03 UTC+1, Stephen:

[Stephen Vicchio]

Hi  Vladimir, 

Thanks for the response! What would you say about the difference in ideal and single determinant spin states that are Delta_S**2 = ~0.1-0.2? I have a few calculations where the ideal S**2 is 6.00000 and the single determinant S**2 is 6.150000. 

I found a rule for Gaussian calculations where the difference should be less than 10% of the ideal S**2. However, is that a reasonable assertion for all UKS DFT calculations? 

With much appreciation, 

Stephen 

--
You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/l69VOiAKtEo/unsubscribe.
To unsubscribe from this group and all its topics, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/0afd748c-f222-4c46-847f-c0a3aa8f416dn%40googlegroups.com.

--

Stephen P. Vicchio

PhD Candidate | Getman Research Group

Dept. of Education GAANN Fellow

Director of Student Organization Finances, Graduate Student Government

Department of Chemical & Biomolecular Engineering

105 Earle Hall | Clemson, SC 29634-0909

svi...@g.clemson.edu

[Vladimir Rybkin]

Dear Stephen,

I would say S**2 is 6.150000 is fine. You should, nevertheless, whether your target properties look reasonable.

Yours,

Vladimir

понедельник, 25 января 2021 г. в 14:26:15 UTC+1, Stephen: