[CP2K-user] [CP2K:20569] Re: SETUP for AIMD

[Michela Benazzi]

 Hi Matteo,
I am so happy that helped. I am not sure what kind of cluster you are 
working with, but for my HPC cluster you have to apply to a long vs short 
partition, where the short partition allows you to run jobs up to 24 h, 
while the long partition allows to run jobs for longer. You can ask your 
administrator about it, perhaps!

As far as the script goes, I see that your RESTART PRINT is OFF. That means 
you are not checkpointing: change it to ON, that should help. That way, you 
can use the generated restart file (.restart) as a checkpoint instead of 
the input file.

Michela



On Wednesday, August 14, 2024 at 4:46:42 AM UTC-4 Matteo wrote:

> Thank you. I tried and it is working very well.
> The only problem is that my cluster will kill after 24 hours.
> Is there a way to restart the MD run from the same point ?
>
> Il giorno martedì 13 agosto 2024 alle 19:53:28 UTC+2 Michela Benazzi ha 
> scritto:
>
>> Perfect! I'd love an update, let me know if things work! Helping others 
>> troubleshoot is a good exercise for me also :)
>>
>> Michela
>>
>> On Tuesday, August 13, 2024 at 1:46:19 PM UTC-4 Matteo wrote:
>>
>>> Thank you very much for your suggestions:
>>> 1)I will try with DVZP basis set
>>> 2)Yes Nose-Hoover is always better
>>> 3)I will try Meta GG for example M06_2X was good in my previous 
>>> calculations.
>>> 4)This is just a mistake GLOBAL is ok for me
>>> 5)Ok also this should be good with 50 fs.
>>>
>>> Il giorno martedì 13 agosto 2024 alle 18:23:57 UTC+2 Michela Benazzi ha 
>>> scritto:
>>>
>>>>
>>>> Hi Matteo, I can brainstorm with you :)
>>>>
>>>> 1) Using double zeta basis sets is good 
>>>> 2) Try Nose-Hoover thermostat as well. CSVR can be a bit rough. I work 
>>>> with liquid metals (in some ways, similar to ILs) and my simulations 
>>>> benefited a lot from switching to N-H. 
>>>> 3) Read some literature on MGGA (meta GGA) functionals for XC potential 
>>>> and see what you think. I also noticed a vast improvement when I switched 
>>>> to that! 
>>>> 4) Is there a reason why you are using a MASSIVE thermostat? I am not 
>>>> sure what your purpose is, but it means that you are adding a thermal bath 
>>>> for every degree of freedom in your system, as opposed to just one for the 
>>>> whole system. Do you need to control the temperature/kinetic properties 
>>>> separately for subsets of your system? If not, switching to GLOBAL will 
>>>> speed up sampling.
>>>> 5) The default TIME_CON is 1000 fs. Why don't you test lower values for 
>>>> thermo/barostat? e.g. 50 fs.
>>>>
>>>> Thank you!
>>>>
>>>> Michela
>>>> On Monday, August 12, 2024 at 6:09:14 AM UTC-4 Matteo wrote:
>>>>
>>>>>  have other two questions about ab-initio MD.
>>>>> 1)I have a box with around 200 atoms of a ionic liquid (a cluster of 
>>>>> three neutral couples). Which combination of DFT Functional and basis set 
>>>>> do you suggest for achiving a fast Ab initio MD to arrive at around 30 ps ?
>>>>> 2)How parallelize efficiently CP2K. It is better OpenMP or a 
>>>>> combination with MPI ?
>>>>> I attach also my setup and I am open to all suggestions.
>>>>>
>>>>>

-- 
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 cp2k+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/0820baef-7750-4ef2-9cc6-c30fe954b103n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240814/9b339923/attachment-0001.htm>