OT eigensolver problem

Matt W MattWa... at gmail.com
Mon Oct 26 16:39:09 CET 2015

Just to be a bit clearer - your ground state density is converged - which 
means that your occupied orbitals are converged.

But, the OT method only works with the occupied orbitals. To get unoccupied 
orbitals an extra computation needs to be done. 

You ask for extra unoccupied states in the PDOS section (maybe elsewhere 
too, I've only looked very quickly).

CP2K then uses some algorithm, maybe Gram Schmitt or similar to build 
unoccupied orbitals that are orthogonal to the occupied ones.

This is the bit of the calculation that is not converging and where Samuel 
suggests you allow CP2K to carry out extra iterations.



On Monday, October 26, 2015 at 9:28:46 AM UTC, Samuel Andermatt wrote:
> Well ot might not be converged, but your scf is converged.
> *** SCF run converged in     1 steps ***
> and
>   outer SCF iter =    1 RMS gradient =   0.80E-07 energy =       
> -784.0104898082
>   outer SCF loop converged in   1 iterations or    1 steps
> Therefore restarting with the same density doesn't help.
> You need to change max_:iter_lumos in the scf section.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20151026/b676d3c0/attachment.html>

More information about the CP2K-user mailing list