[CP2K-user] [CP2K:19517] dRe: Ehrenfest + Field problem for version 2023.2

[Guillaume Le Breton]

Dear Natalia,

First of all, your threshold regarding the RTP seems reasonable and I do 
agree that using GGAs compared to Hybrids should not be a problem as long 
as you match the field frequency with the transition targetted. 

The dipole moment and the potential energy oscillate at the field frequency 
for both calculations as expected. The evolution of the kinetic energy and 
the NN bond for version 9.1 seems to match the one from the paper: the 
oscillation period is about 20 fs. Yet, this is a bit weird since it is not 
the same as the one of the field! In version 2022.2, the kinetic energy is 
quite different but the period of the oscillation seems to match the one of 
the field. I would assume that there is two time scale: one related to the 
''instantaneous adaptation of the system'' and another one related to a 
slower evolution due to the excitation process. For instance, bottom right 
part of Figure 6: the electronic population evolved slowly toward a full 
excitation but with an oscillation at the field frequency. Hence, not 
observing a response of the nuclei at the field period is a bit concerning 
from my point of view. Yet, I am not sure whether or not this difference 
should be expected (for instance due to the implementation of Ehrenfest in 
the Gaussian code compared to CP2K). 

One potential problem could be the initial conditions: the temperature is 
non-zero in your input file. Could you try a run with a zero temperature to 
get rid of any potential problems due to the initial conditions for both 
versions? 

Best regards,
Guillaume

Le lundi 13 novembre 2023 à 10:28:49 UTC+1, Natalia K a écrit :

> Dear Guillaume,
>
> thank you for your instructions. After applying the patch, and making sure 
> the test files are correct, I ran calculations comparing the latest version 
> 2023.2 (with the patch) with the version 9.1, that I was previously using, 
> for the system Ag6-N2. The results are different. Namely, the kinetic 
> energy is significantly different (and consequently the atomic motion). 
> This calculation is similar to the one from 
> https://dx.doi.org/10.1021/acs.jpcc.0c02979?ref=pdf, where on Fig.6 
> (middle left) they show the N-N bond change with time. Their result is 
> obtained with the Gaussian code. With all the parameters similar to theirs 
> (except the XC functional, they use hybrid and I use GGA, but I don't think 
> it is important), CP2K-v9.1 gives similar result, while v2023.2 does not. I 
> attach my results and the input file I used in both calculations below.
>
> Best regards,
>
> Natalia
>
> On Thursday, October 12, 2023 at 10:54:18 AM UTC+2 Guillaume Le Breton 
> wrote:
>
>> Dear CP2K community, 
>>
>> On 24 February 2023 I made some changes in the Real-Time calculation part 
>> of the code which turned out to break one specific behavior: Ehrenfest 
>> Dynamics with a time-dependent field. This has been fixed the 9 October 
>> 2023. 
>>
>> Thus, if you are using Ehrenfest Dynamics with a time-dependent field, 
>> you are NOT impacted if:
>>
>> - You are using a version before 24.02.2023. This includes V2023.1 and 
>> the previous ones. 
>>    - The master branch from 09.10.2023. 
>>
>> You ARE impacted if:
>> - You are using a version between 24.02.2023 and 09.10.2023. This 
>> includes V2023.2. 
>>
>> If you are impacted, I recommend you rerun your simulations (involving 
>> Ehrenfest + Field) with the newest version of the code as the issue is 
>> quite important. If you want to keep using version 2023.2, here is a patch 
>> to solve this problem. 
>>
>> 1) Download the patch and untar it.
>> 2) Apply the patch using the patch command in the shell: 
>>
>> $> patch cp2k/src/qs_force.F qs_force.patch
>> $> patch cp2k/tests/QS/regtest-rtp-2/TEST_FILES TEST_FILES.patch
>>
>> To check that the patch has been applied correctly, open the 
>> cp2k/tests/QS/regtest-rtp-2/TEST_FILES file. Line 11 should be:
>>
>> H2-emd-efield.inp                                      2      4e-11       
>>          -0.894818188949
>>
>> 3) recompile the code and run regtests. 
>>
>> I sincerely apologize for this problem. 
>>
>> Guillaume Le Breton
>
>

-- 
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/11c79ac7-3a7c-4ce3-9ba9-e48dd40a0e84n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20231116/9c9da709/attachment.htm>