[CP2K-user] PBE0-D3 MD problem with HF module

[Vladimir Rybkin]

Not really, as I it's all system-dependent. You would likely need hundreds 
of CPU cores.

Yours,

Vladimir

вторник, 2 апреля 2019 г., 15:55:24 UTC+2 пользователь Brandon Meza 
Gonzalez написал:
>
> Dear Vladimir,
>
> Thanks for your time and help.
>
> Do you have a specific memory or architectural recommendation for this 
> calculations?
>
> Brandon,
>
> El martes, 2 de abril de 2019, 3:26:09 (UTC-6), Vladimir Rybkin escribió:
>>
>> Dear Brandon,
>>
>> first you should add the following line to your &INTERACTION_POTENTIAL 
>> section:
>> T_C_G_DATA ./t_c_g.dat
>>
>> The file t_c_g.dat is in the /data.
>>
>> Second, and the main, HF exchange calculations are expensive and memory 
>> consuming. It is very likely that you run out of memory. I see you are 
>> using only 24 MPI processes and one thread, which is likely not enough to 
>> run HFX even with ADMM.
>>
>> Yours,
>>
>> Vladimir
>>
>> воскресенье, 31 марта 2019 г., 23:27:34 UTC+2 пользователь Brandon Meza 
>> Gonzalez написал:
>>>
>>> Hi, cp2k users:
>>>
>>> I was suffering problems in HF module in the input that I show you, 
>>> using the Poisson resolutor (xyz). 
>>>
>>> I was able to run the MD simulations without the HF module and the job 
>>> continues running, but when I implement HF in order to describe the system 
>>> through PBE0 the calculation crashes. 
>>>
>>> Is there a solution or recommendation for this problem?
>>>
>>> Best regards,
>>>
>>> Brandon
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190403/ddc21a66/attachment.htm>