[CP2K-user] [CP2K:19113] Re: Ehrenfest dynamics, inconsistent results from versions 9.1 and 2023.1

[Guillaume Le Breton]

Dear Natalia,

My apologies: the updates I have done are in the 2023.2 version, not the 
2023.1 one. Hence this should not be the issue. 
Discussing with other colleagues, they have been able to compare results 
from 9.1 and the current version of the code. 

Looking closer at your input, here would be some advice: 

- You can decrease the  &REAL_TIME_PROPAGATION&EPS_ITER by 2 orders of 
magnitude to be safer
- The field you are applying has a too long envelope compared to the 
simulation time. At t=0 the field envelope is already quite large so that 
at the very begging of the simulation you apply an electric field: this may 
cause instability.  Try T0=5 fs
- Try to compare the results from the 9.1 and 2023.1 versions without a 
time-dependent field.

Finally, in my point of view, the results you have obtained from the 2023.1 
version seem more acceptable than the ones from 9.1. The kinetic energy of 
9.1 for instance seems weird to me.  The results from the 2023.1 version 
seem much more smooth for all the results. 

Hope that this will help you, do not hesitate to reply! 

Best of luck,
Guillaume

Le lundi 10 juillet 2023 à 09:43:42 UTC+2, Natalia K a écrit :

> Dear Guillaume,
>
> thank you for your answer. Unfortunately, it doesn't seem to be the case 
> that the energy is recovered after the pulse is off. Neither the dipole 
> moment is the same. Please see the updated figures attached.
>
> Best regards,
> Natalia
>
> On Friday, July 7, 2023 at 11:16:37 AM UTC+2 Guillaume Le Breton wrote:
>
>> Dear Natalia,
>>
>> I have modified recently the calculation of the electron-field energy for 
>> real-time field. Previously, this calculation was made on the grid, now it 
>> is done as an expectation value of the dipole operator. This was done in 
>> order to compare the total electronic energy with the velocity gauge 
>> approach (now available for Real-Time Propagation, not yet for Ehrenfest 
>> MD). 
>>
>> This modification of the electronic energy calculation should have an 
>> effect only when the electric field is on. If you continue this calculation 
>> after the pulse end you should recover the same energy as for the previous 
>> version of CP2K. 
>> Moreover, the other quantities such as the dipole moment or the nuclei 
>> displacement should be the same for this version and the 9.1 throughout the 
>> whole simulation. 
>>
>> Please let us know if this is not the case.
>>
>> Best,
>> Guillaume 
>>
>>
>> Le jeudi 6 juillet 2023 à 12:45:27 UTC+2, Natalia K a écrit :
>>
>>> Dear community,
>>>
>>> I am performing test calculations for a system of 2 silver atoms under 
>>> an external field. Using the same input files (attached), I get completely 
>>> different energies from the two versions of the code (see images attached). 
>>> The results from cp2k-9.1 look correct to me. Is this the code problem or 
>>> should I modify the input file? It seems consistent when compared to the 
>>> files from the /tests/regtests.. files. 
>>>
>>> P.S. the xyz file is obtained from the geometry optimization with the 
>>> version 9.1, but I wouldn't think this is the problem.
>>>
>>

-- 
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/3da07f53-8b2e-4d33-8c9b-1cfbae80a999n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20230714/8909e90f/attachment.htm>