[CP2K-user] [CP2K:18469] Re: COM drifting

[Simone Ritarossi]

Dear Marcella,

Before SGCPMD, I equilibrated the system with an NVT.
MAX_SCF_HIST i commented out in this file, but in the run it was not 
commented on (was set to 6). 
In generale, you recommend a preconditioner such as FULL_ALL? 
"EPS_SCF  is rather tight to be a SGCPMD", but I stop the scf after 6 OT 
steps.

Many thanks in advance,
Simone

Il giorno venerdì 17 febbraio 2023 alle 15:46:24 UTC+1 Marcella Iannuzzi ha 
scritto:

>
>
> Dear Simone, 
>
> Did you run BOMD with this system? 
> It is always better to start sampling with SGCPMD from a stable BOMD.
>
>
> I don't know how you have determined the cutoff. One way to do it would be 
> to calculate the BO forces with increasing cutoff to check the deviation. 
> Since the parameters have been tested for the "almost identical system", I 
> understand that for that system under the same conditions the COM did not 
> drift. Is it so?
>
> Concerning the setting:
> The time step is very large. 0.5 would be safer.
> EPS_SCF  is rather tight to be a SGCPMD
> The preconditions is poor
> MAX_SCF_HIST is commented out
>
> However it is difficult to judge without knowing how the wfn's convergence 
> is behaving.
>
> Regards
> Marcella
>
> On Friday, February 17, 2023 at 2:58:55 PM UTC+1 Simone Ritarossi wrote:
>
>> Dear Marcella,
>>
>> I chose the cutoff as the minimum possible. Without COMVEL_TOL there is 
>> drifting (along x) that must not be there. I would like to solve the 
>> drifting without the forcing of COMVEL_TOL (e.g. by increasing the cutoff 
>> so that the forces can be calculated better? )
>> The noisygamma was chosen from previous simulations on almost identical 
>> systems, as advised by the supervisor.
>> Why do you say that " Your settings for SGCPMD look strange." ?
>>
>> Many thanks in advance,
>> Simone
>> Il giorno venerdì 17 febbraio 2023 alle 10:12:09 UTC+1 Marcella Iannuzzi 
>> ha scritto:
>>
>>> Dear Simone,
>>>
>>> The cutoff looks indeed small. 
>>> Does the COM drift by BOMD? Have you tried without COMVEL_TOL? 
>>> Your settings for SGCPMD look strange. 
>>> Have you run the needed tests for the nosy-gamma to find the right 
>>> parameters?
>>>
>>> Regards
>>> Marcella
>>>
>>> On Friday, February 17, 2023 at 2:25:00 AM UTC+1 Simone Ritarossi wrote:
>>>
>>>> Dear cp2k-users,
>>>>
>>>> I have a problem with COM (center of mass) drifting (along x). I'm 
>>>> running a LANGEVIN  (SGCPMD) on my  heterostructure, consisting of a GST + 
>>>> TiTe2 (633 atoms),  where I consider van der Waals interactions (with 
>>>> DFT-D3).  The cutoff is 320 Ry: maybe is it too small? Maybe at a 
>>>> temperature of 600K a gamma of 1.9e-3 is too big?
>>>>
>>>> I noticed that by using COMVEL_TOL=0 I have the following behaviour: 
>>>> the temperature of the whole system does not correspond to the average 
>>>> (weighted according to the number of atoms) of the temperatures of the 
>>>> individual atomic species. 
>>>>
>>>>
>>>> Many thanks in advance,
>>>> Simone
>>>>
>>>

-- 
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/9ba34af5-66f0-40c9-aff5-dcfe339fbe90n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20230217/2b3b555a/attachment.htm>