[CP2K-user] Convergence of hybrid functional calculations

[Xiaoming Wang]

Hi,

Using EPS_SCHWARZ=1.0e-8 solved the problem.

Best,
Xiaoming

On Wednesday, August 21, 2019 at 7:31:14 PM UTC-4, Xiaoming Wang wrote:
>
> Hello,
>
> Thanks very much for all the above suggestions. I played with the 
> parameter ENERGY_GAP
> and the OUTER_LOOP, but the problem persisted. For EPS_SCHWARZ, I reduced 
> it 
> from 1.0e-6 to 1.0e-7, and I didn't see any improvement on the 
> convergence. For further
> smaller value, say 1.0e-8, the calculation seems to consume a lot of 
> memory and I have
> to run my job on much more nodes.  I will keep on trying that to see if 
> there will be any
> difference. Meanwhile, any other comments or suggestions are appreciated.
>
> Best,
> Xiaoming
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190822/d34466fd/attachment.htm>