Kohn-Sham Matrix and Overlap Matrix when SCF is performed with OT

Matt W mattwa... at gmail.com
Tue Apr 11 18:16:05 UTC 2017


well I see agreement to lots of decimal places

-0.87736367 - -0.87736368

between the lowest eigenvalues, so I'd think it is only the lack of 
precision in printing out the matrices that leads to a difference.

The values you gave in the text file correspond to the beta spin 
eigenvalues in the cp2k output, just in case you were checking the alpha 


On Tuesday, April 11, 2017 at 6:59:17 PM UTC+1, Soumya Ghosh wrote:
> Hi Matt,
> Thanks for your comment. I have attached the inputfile (*.inp), logfile 
> (output file: *.log) and the eigenvalues (*.txt) that I obtain by symmetric 
> diagonalization.
> On Tuesday, April 11, 2017 at 12:41:47 PM UTC-5, Matt W wrote:
>> Hi,
>> what do you mean by 'not exactly'? Slight difference, or off by a large 
>> amount?
>> What are your accuracy settings - EPS_DEFAULT and EPS_SCF?
>> Matt
>> On Tuesday, April 11, 2017 at 5:58:40 PM UTC+1, Soumya Ghosh wrote:
>>> Hi,
>>> I printed out KS matrix and Overlap matrix for a system where the SCF is 
>>> performed with OT method. I performed symmetric orthogonalization of the KS 
>>> matrix using the Overlap matrix (S^(-1/2)*KS*S^(-1/2)) and then 
>>> diagonalized the matrix to obtain the eigenvalues. These eigenvalues, 
>>> however, do not exactly match with the MO energies printed by CP2K. Am I 
>>> missing something? Please comment   
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170411/ebef0b04/attachment.htm>

More information about the CP2K-user mailing list