[CP2K-user] Spin Contamination "Rule of Thumb?"

[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: 

> 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 
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20210125/96f293a8/attachment.htm>