interface for deepmd with the centroids atoms

[DavideTisi]

The following request is aimed to allow the deepmd code to pass to lammps the full non symmetric atomic_virial matrix (dvatom) using the cvatom matrix defined for centroid/atom/stress. This will allow to take full advantage of the use of the computed "atomic_virial" that is non-symmetric, beeing deepmd potential inherently many-body. This allows to compute correctly the heat current with LAMMPS. It has been shown in the literature that the usual formula implemented in compute_hat_flux in LAMMPS for the heat current is not correct for many body potential where the atomic virial (called vatom in the deepmd interface) is not symmetric (see Fan et al, Phys. Rev. B 92, 094301 or Phys Rev E, 99, 051301(R) (2019) or even the LAMMPS manual page on compute heat/flux ).

Now lammps has an option called compute centroid/atom/stress to have a non symmetric atomic viral, and to compute the thermal conductivity from that. With this request I have included in deepmd interface for lammps the use of these 3*3 atomic stress matrix called cvatom.



To have a better inside on the equations the heat flux formula is:

J= sum_i e_i v_i + sum_{n,m} (r_m-r_n) de_m/dr_n v_n

Being r_n position of atom, v_n velocity of atom and e_n local atomic energy.
deepmd computes and stores in dvatom the matrix sum_m(r_n-r_m) de_m/dr_n .
Thus I passed this 9 component non-symmetric array to cvatom that can be used with the compute heat/flux.

[njzjz]

Please submit commits to devel branch

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 75.72%. Comparing base (e2fc5e3) to head (2f190ae).
Report is 1459 commits behind head on devel.

Additional details and impacted files

@@           Coverage Diff           @@
##            devel    #1093   +/-   ##
=======================================
  Coverage   75.72%   75.72%           
=======================================
  Files          88       88           
  Lines        6998     6998           
=======================================
  Hits         5299     5299           
  Misses       1699     1699           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[DavideTisi]

Please submit commits to devel branch

I changed the target to devel now.

[mayankaditya]

Thanks for the details. I am just wondering if I could use the option "compute centroid/atom/stress" in current deepmd-2.0.3 lammps. It would be great if an example input for the same is provided.

Thanks,
Mayank

[mayankaditya]

Dear Developers,

It would be great if you could provide an input script for thermal conductivity calculation. I am facing an issue to compute the 'cvatom' and use it in compute heat/flux as mentioned above.

Thanks,
Mayank

[mayankaditya]

Hello everyone,

I am getting the following error while calculating the stress matrix with "compute myStress all centroid/stress/atom NULL virial"

ERROR: Pair style does not support compute centroid/stress/atom (src/compute_centroid_stress_atom.cpp:131)

Please see the attached input file.

Thanks,
Mayank

[DavideTisi]

Dear Mayank,

In the documentation at the end of the page at this link
Doc/lammps-command.md there are the combinations of commands to add to the lammps script in order to use compute heat/flux.
They are working for me, and I am using deepmd-2.0.3 and lammps version 29Oct2020

Best,
Davide

[DavideTisi]

Dear Mayank,

Sorry, I saw your last post after I send mine.
The input seems ok which version of lammps and deepmd are you using ?
Davide

[mayankaditya]

Dear Davide,

Thanks for the prompt response. I am also using deepmd-2.0.3 and default lammps executable (installed with deepmd).

Mayank

[DavideTisi]

Dear Mayank,

for what I see deepmd2.0.3 use as default now lammps stable_29Sep2021, I can try to do some test to see if there are problems with the interface and that version of lammps.
Can you try to see which version of lammps are you using, and in the case can you use lammps version 29Oct2020?
if the lammp version is the problem, changing it might solve it and you can continue in your research.

Davide

[mayankaditya]

Dear David, Thank you for the help, I will try with lammps version 29Oct2020. Best regards, Mayank

…

On Thu, Jan 27, 2022 at 11:58 PM Davide Tisi ***@***.***> wrote: Dear Mayank, for what I see deepmd2.0.3 use as default now lammps stable_29Sep2021, I can try to do some test to see if there are problems with the interface and that version of lammps. Can you try to see which version of lammps are you using, and in the case can you use lammps version 29Oct2020? if the lammp version is the problem, changing it might solve it and you can continue in your research. Davide — Reply to this email directly, view it on GitHub <#1093 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADC5MSVLE64C5URKBZBYLVTUYGFEJANCNFSM5DLPRBXQ> . Triage notifications on the go with GitHub Mobile for iOS <https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675> or Android <https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub>. You are receiving this because you commented.Message ID: ***@***.***>

[neojie]

The link does not work now. What are the combinations of commands to add?

[neojie]

Dear David, Thank you for the help, I will try with lammps version 29Oct2020. Best regards, Mayank
…
On Thu, Jan 27, 2022 at 11:58 PM Davide Tisi @.> wrote: Dear Mayank, for what I see deepmd2.0.3 use as default now lammps stable_29Sep2021, I can try to do some test to see if there are problems with the interface and that version of lammps. Can you try to see which version of lammps are you using, and in the case can you use lammps version 29Oct2020? if the lammp version is the problem, changing it might solve it and you can continue in your research. Davide — Reply to this email directly, view it on GitHub <#1093 (comment)>, or unsubscribe https://github.com/notifications/unsubscribe-auth/ADC5MSVLE64C5URKBZBYLVTUYGFEJANCNFSM5DLPRBXQ . Triage notifications on the go with GitHub Mobile for iOS https://apps.apple.com/app/apple-store/id1477376905?ct=notification-email&mt=8&pt=524675 or Android https://play.google.com/store/apps/details?id=com.github.android&referrer=utm_campaign%3Dnotification-email%26utm_medium%3Demail%26utm_source%3Dgithub. You are receiving this because you commented.Message ID: @.>

Hi Mayank,
Do you make it work using different version of lammps? I tried
compute myStress all centroid/stress/atom NULL virial
and get the same problem as you reported.
I am using the latest stable version of lammps and deepmd-kit
Thank you.
Jie