[CP2K-user] Cores used in geometric optimisation of Benzene

Matt W mattwa... at gmail.com
Wed Jul 3 15:02:01 CET 2019


Hi, 

will it seems it is working ... your settings are not optimal but there is 
a steady decrease in energy / maximum force:

> grep 'Max. grad' Benzene3.out 
  Max. gradient              =         0.2139968883
  Max. gradient              =         0.0746403531
  Max. gradient              =         0.0894084145
  Max. gradient              =         0.0854218746
  Max. gradient              =         0.0687801374
  Max. gradient              =         0.0394113794
  Max. gradient              =         0.0488575204
  Max. gradient              =         0.0204708321
  Max. gradient              =         0.0201705598
  Max. gradient              =         0.0203362535
  Max. gradient              =         0.0209995999
  Max. gradient              =         0.0174069893
  Max. gradient              =         0.0105286161
  Max. gradient              =         0.0065007496
  Max. gradient              =         0.0038402128
  Max. gradient              =         0.0035055641
  Max. gradient              =         0.0039487260
  Max. gradient              =         0.0020366718
  Max. gradient              =         0.0014921412

I would use the default BFGS geometry optimizer rather than CG, I find it 
much more efficient.

Change the preconditioner to FULL_ALL and set energy_gap to 0.005 or so to 
be more efficient in the SCF loops.

Matt

On Wednesday, July 3, 2019 at 5:25:07 AM UTC+1, Stuti Vora wrote:
>
>
> The code ran for two hours, I terminated it then.
>
> The input file is the same as above, only I have increased the box size to 
> 20x20x20 and upped the max iterations to 400
>
> On Tuesday, 2 July 2019 16:59:12 UTC+5:30, Matt W wrote:
>>
>> Better post the output so we can see what happened.
>>
>> On Tuesday, July 2, 2019 at 7:32:47 AM UTC+1, Stuti Vora wrote:
>>>
>>> I ran the code again with a 20x20x20 box but that didn't work. I 
>>> terminated the program after an hour, none of the convergence criteria had 
>>> been met till then.
>>>
>>> Any other suggestions as to what could be the problem? I ultimately plan 
>>> to use the same code to optimize a IL-water system. 
>>> Also, how can I check if cp2k is installed correctly? Though I don't 
>>> think installation is the problem, I was able to optimize a water molecule 
>>> without any problems.
>>>
>>> Stuti
>>>
>>> On Saturday, 29 June 2019 00:27:49 UTC+5:30, Matt W wrote:
>>>>
>>>> Ah, well you need to put it in a much bigger box then. 10x10x10 A 
>>>> minimum. There are more things to do to get a truly isolated molecule, but 
>>>> that would give you 95%. I suspect that might fix some of the problems you 
>>>> have.
>>>>
>>>> Matt
>>>>
>>>> On Friday, June 28, 2019 at 5:35:10 PM UTC+1, Stuti Vora wrote:
>>>>>
>>>>> Hi Matt, 
>>>>>
>>>>> Well, I think its running currently, is there anyway I can check if 
>>>>> its not?
>>>>> Also, its not a crystal, just a simple molecule.
>>>>>
>>>>> Stuti
>>>>>
>>>>>
>>>>>>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190703/15901dbc/attachment.html>


More information about the CP2K-user mailing list