[CP2K-user] [CP2K:15319] Many SCF cycles in each geometry optimization step

Thomas Kühne tku... at gmail.com
Mon Jun 28 21:27:44 UTC 2021


Dear Hongyang, 

you have given the answer to your own question, the observed behavior is a consequence 
of the employed geometry optimizer. LBFGS, is typically faster anyhow, which is not only 
due to the algorithm, but also due to the particular implementation. 

Best, 
Thomas

> Am 10.05.2021 um 07:39 schrieb ma... at gmail.com <ma455... at gmail.com>:
> 
> Hi,
> 
> I'm a new user of cp2k 8.1. I'm calculating a P doped Si system (output file attached). The system contains 1000 atoms. The calculation goes well and the SCF converges fairly smoothly. However, different from my previous calculations, I found in this calculation there are many SCF cycles (SCF converged again and again) for each geometry optimization step. In my previous calculations, the basic steps are: SCF ->  OPTIMIZATION STEP 1 -> SCF -> OPTIMIZATION STEP 2 -> SCF -> OPTIMIZATION STEP 3. But in this calculation, the steps are: SCF -> OPTIMIZATION STEP 1 -> SCF (converged) -> SCF (converged) -> SCF (converged) -> SCF (converged) ...... -> OPTIMIZATION STEP 2. I suspect this maybe caused by the different OPTIMISER. In my previous calculations, I used BFGS. But in this calculation I used LBFGS. However, based on the tutorial, LBFGS should be faster in large system calculation but the results are not what I expected. Could someone please provide some suggestions on improving this issue? Thanks
> 
> Regards,
> Hongyang
> 
> -- 
> You received this message because you are subscribed to the Google Groups "cp2k" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to cp... at googlegroups.com <mailto:cp... at googlegroups.com>.
> To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/e713484b-b0f4-4886-b689-6459ab67cd5dn%40googlegroups.com <https://groups.google.com/d/msgid/cp2k/e713484b-b0f4-4886-b689-6459ab67cd5dn%40googlegroups.com?utm_medium=email&utm_source=footer>.
> <s5-hse-01-out.txt>



==============================
Thomas D. Kühne
Dynamics of Condensed Matter
Chair of Theoretical Chemistry
University of Paderborn
Warburger Str. 100
D-33098 Paderborn
Germany
thomas... at upb.de
+49/(0)5251/60-5726

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20210628/e4f0143a/attachment.htm>


More information about the CP2K-user mailing list