[CP2K-user] [CP2K:18151] Re: Temperature drift in polaron study

Marcella Iannuzzi marci.akira at gmail.com
Mon Dec 5 12:43:44 UTC 2022

Dear Vinayak

Are you still running with hybrid functionals?
If yes, I would suggest to switch to a GGA functional until the structure 
is  equilibrated at the desired conditions.
Finer calculation of the electronic properties by means of higher level 
theory, e.g., hybrids, are significantly more expensive and should be used 
only when they really needed.
The thermostat might need substantially more than 500 fs to bring the 
system to the desired temperature. 
Playing with the parameter of the thermostat might help to speed this 
process up. 
During the equilibration part of the simulation, you can also apply a 
temperature tolerance, which will force the temperature to remain within a 
certain range by rescaling the velocities.
The part of the simulation where such a velocity rescaling is active cannot 
not be considered a proper sampling of the ensemble.


On Friday, December 2, 2022 at 7:14:31 PM UTC+1 vina... at seas.upenn.edu 

> Hi Marcella,
> Thanks a lot for your advice! I have been trying to follow it but still 
> facing some issues.
> 1. The simulation has run for 500 fs and seems to be settling around 150K 
> rather than 300K (plot attached). I dont see how that may increase later.
> 2. For some reason it is taking us a lot of time to run even 500fs of 
> calculations. It has taken a week to run just 500fs. 
> I am using 6 nodes, with 32 cores on each using mpirun. The jobs script is 
> attached as well. Is this the speed one should expect or is there a faster 
> way to proceed?
> We have 192 atoms in the system, so we have tried to optimize for that as 
> well. 
> It would be extremely helpful if you could help us with the above. 
> Thanks a lot,
> Vinayak
> On Monday, November 14, 2022 at 4:38:09 AM UTC-5 Marcella Iannuzzi wrote:
>> Dear Vinayak
>> Starting from an optimised geometry it is expected that the just 
>> initialised finite temperature decreases. 
>> Even by setting a thermostat (NVT), it takes quite some time  (some ps) 
>> before the thermostat succeeds in equilibrating the system to the desired 
>> temperature. 
>> Regards
>> Marcella
>> On Friday, November 11, 2022 at 6:10:44 PM UTC+1 vina... at seas.upenn.edu 
>> wrote:
>>> Hi All,
>>> Thanks for creating this wonderful forum for cp2k. 
>>> I am new to AIMD simulations but have been doing MD and DFT separately 
>>> for some time. Currently I am trying to do a polaron evolution study in a 
>>> pervoskite at a finite temperature. The 0K results are great and as 
>>> expected but with MD at 300K and Nose thermostat I see a large temperature 
>>> drift in the first few fs. I have performed geometrical optimisation on the 
>>> supercell. Please find all the files which I use attached. You can also 
>>> find the .ener file where you can see the quick drift. 
>>> Any help would be extremely appreciated.
>>> Thanks,
>>> Vinayak

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/0ab4a1b5-13aa-4220-9f03-b9e95aeee343n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20221205/f3477f8e/attachment.htm>

More information about the CP2K-user mailing list