[CP2K-user] choice of preconditioner and accuracy

[Daniele Fontanari]

Hello everyone,

I noticed, when testing MD and GEO_OPT runs on a system with around ~1000 
atoms, that each SCF step can take considerably less if I switch from 
FULL_ALL to FULL_KINETIC inside FORCE_EVAL / DFT / SCF / OT / 
PRECONDITIONER.
I was wondering if the choice might impact also accuracy of results for the 
convergence of the SCF, leading to possibly wrong results (expecially for 
md) or if the convergence is essentially the same with all method.

cheers
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20181024/d5c00b2f/attachment.htm>