STRESS_TENSOR Convergence Issue

Sebastian Hütter sebastia... at
Mon Oct 17 15:09:21 UTC 2016

Hello Matt,

I believe I took this epsilon from one of the static energy examples, but 
it appears that I cannot find the source anymore. Hm.

New calculations with NREP 3, EPS_DEFAULT 1E-10 are running currently and 
while I now have a very asymmetric behaviour between compression and 
tension (I calculate both sides and average), results appear sane for now.

Is there a rule of thumb how many cells I will need for good results? In 
theory it's xyz-periodic anyway, but I understand in order to get fine 
enough grids I need a somewhat larger volume. In the mean time, I have also 
added CELL_REF statements with the unmodified cell to be sure there is no 
change between calculations there.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the CP2K-user mailing list