[CP2K-user] [CP2K:14431] Cell_OPT convergence is too slow

[jt...@gmail.com]

Hi Rui Zou

Your system is very big! So maybe it's normal to be slow. (though you can 
add cores and nodes to run it) 

What Matthias said is right, and I suggest LBFGS rather than BFGS as your 
minimizer, you can find description of minimizer in cp2k manual 
<https://manual.cp2k.org/trunk/CP2K_INPUT/MOTION/CELL_OPT.html>.
Also, DIIS sometimes faster than CG, though it may lead you to convergence 
problem. 

Best regards
JT Yang
On Saturday, December 26, 2020 at 10:37:11 AM UTC+8 91... at qq.com wrote:

> Hi, Matthias Krack
>
> Thank you so much for your advice. 
> In this calculation I used a supercell of 320 atoms. I suspect this also 
> caused the convergence rate to be too slow. When performing CELL_OPT in 
> CP2K, is it necessary to expand the cell first? 
>
> Best regards
> Rui Zou
>
> 在2020年12月25日星期五 UTC+8 下午7:02:58<Matthias Krack> 写道：
>
>> Hi Rui Zou
>>
>>  
>>
>> There are two issues which might hamper the convergence process of your 
>> cell_opt run:
>>
>>    1. The SCF convergence threshold value is with 5.0E-6 quite large 
>>    which will cause a low accuracy of the calculated forces. Noisy forces will 
>>    in turn “confuse” the optimizer.
>>    2. The applied constraints KEEP_ANGLES and KEEP_SYMMETRY might 
>>    prevent the optimiser from finding an energetically lower lying structure 
>>    with less symmetry indicated by residual forces.
>>
>>  
>>
>> HTH
>>
>>  
>>
>> Matthias
>>
>>  
>>
>> *Von:* c... at googlegroups.com <c... at googlegroups.com> *Im Auftrag von *rui 
>> zou
>> *Gesendet:* Freitag, 25. Dezember 2020 05:41
>> *An:* cp2k <c... at googlegroups.com>
>> *Betreff:* [CP2K:14431] Cell_OPT convergence is too slow
>>
>>  
>>
>> Hi everyone!
>>
>>  
>>
>> I am a new user of CP2K. I want to use CP2K to optimize the cell. 
>> However, the convergence of this process is too slow. Can you help me check 
>> if there is a problem with my input file? Thank  you so much!
>>
>>  
>>
>> Best regards,
>>
>> Rui Zou
>>
>> -- 
>> 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.
>> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/cp2k/2ddfaea7-2b59-4877-9d72-7c4bb462b4ddn%40googlegroups.com 
>> <https://groups.google.com/d/msgid/cp2k/2ddfaea7-2b59-4877-9d72-7c4bb462b4ddn%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20201226/4ec25be2/attachment.htm>