[CP2K-user] [CP2K:14337] Re: Spin Contamination

[ma...@gmail.com]

For anyone who reads this thread! It seems that it is possible to calculate 
the spin contamination value manually. One needs to include the 
eigenvectors into the CP2K output by adding the PRINT statement to the DFT 
block and print out the eigenvectors 
(https://manual.cp2k.org/cp2k-6_1-branch/CP2K_INPUT/FORCE_EVAL/DFT/PRINT/MO.html#EIGENVECTORS). 
I also specified the multiplicity of the calculation explicitly in the 
input.

The equation for the calculation of spin contamination is described here: 
https://en.wikipedia.org/wiki/Spin_contamination . And the eigenvectors are 
used to calculate the last term in the equation - I believe it is the sum 
of the overlap between the occupied alpha and beta orbitals. The overlap 
can be found by multiplying each alpha eigenvector with the corresponding 
beta eigenvector and then adding the results together.

I added an example of how to do it as an attachment.

Best Regards,
Mario
On Thursday, 3 December 2020 at 16:38:38 UTC ma... at gmail.com wrote:

> It seems that Juerg was right, it is not implemented for semi-empirical 
> methods. The information about spin contamination became available right 
> after I changed AM1 to the XC functional PBE.
>
> Thank you both for your input. Would you think this is something the 
> developers would be interested in adding in?
>
> Best Regards,
> Mario
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20201204/60cd947a/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: spin_contamination.xlsx
Type: application/vnd.openxmlformats-officedocument.spreadsheetml.sheet
Size: 17552 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20201204/60cd947a/attachment.xlsx>