Geometry optimization - low settings optimized structure and high settings optimized structure.

[ganta.pra... at gmail.com]

Dear Prof. Matthias Krack, 

Thank you very much for your suggestion. Just need another opinion,  I am 
using a reasonable values for low settings for instance (cutoff=400, 
rel_cutoff=40, ngrids=4, EPS_DEFAULT=1E-8) compared to the high-settings 
(cutoff=600, rel_cutoff=60, ngrids=4, EPS_DEFAULT=1E-12). 

Is there any chance? Please let me know. 

Thank you.




On Tuesday, June 20, 2017 at 3:04:14 PM UTC+2, Matthias Krack wrote:
>
> Hi
>
> due to my experience, such an approach does not pay off, since sloppy 
> settings for CUTOFF, EPS_DEFAULT, or EPS_SCF result in noisy forces and a 
> poor Hessian matrix which "confuses" the geometry optimizer and thus delays 
> or even disables any convergence.
>
> Matthias
>
> On Tuesday, 20 June 2017 14:30:57 UTC+2, ganta.... at gmail.com wrote:
>>
>> Hi all,
>>
>> I am doing geometry optimization using CP2K and my structure is 
>> crystalline (FeOOH). It has a total of 400 atoms and it is taking a lot of 
>> time to do geometry optimization. 
>>
>> My question is that, will it help, if I do a quick geometry optimization 
>> with very low settings (reducing CUT_OFF, REL_CUTOFF, EPS_DEFAULT, EPS_SCF) 
>> and use the resultant optimized structure to do geometry optimization with 
>> high settings.
>>
>> I am quite new to CP2K and I would like to know your opinions. Attached 
>> is my input file. Feel free to give any advice regarding my input file to 
>> speed up geometry optimization.  
>>
>> Thank you. Have a nice day.
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20170620/0735e27a/attachment.htm>