Why simulation with PBE functional is too slow?

Matt W MattWa... at gmail.com
Mon Oct 7 19:43:31 CEST 2013


Hi Huan,

it is not normal for one GGA functional to be massively more expensive than 
another - typically PBE is one of the most robust in fact.

You have a few unusual keywords in your input, so it is hard to comment 
without seeing the output: 

Is each SCF cycle slower with PBE, or do you need many more SCF cycles?
How are you estimating memory usage? If you are monitoring it with, say, 
"top" linux command, is the memory usage consistent or does it peak in some 
section that you can identify (localization maybe?)?

Also, and maybe most importantly, if you strip your input down to a very 
simple calculation do you see these differences? You need to debug your 
results like you would some computer code your wrote. And without a sample 
of the output it is hard to help.

Matt

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20131007/30532efb/attachment.html>


More information about the CP2K-user mailing list