Treatment of Virial Data in training

[sivonxay]

When training a model using VASP AIMD data, should the kinetic contributions to the virials be subtracted out?

I took a look at how it's done in https://github.com/deepmodeling/dpdata/blob/master/dpdata/vasp/outcar.py and that seems to include the kinetic contributions, since it parses the lines beginning with "in kB".

I have run VASP AIMD of my system at various temperatures (1000, 1500, 2000 etc.), each at a volume that yields approximately 0 external pressure (Using ISIF=2). Thus, the virial data for each of these would be similar, although their volumes are very different. Not excluding the kinetic contribution would imply that the training/model has some concept of the temperature/velocity of the training data - which is data not included in the training data.

Thanks in advance for your response,
Eric

[jameswind]

Hi Eric, There are two kinds of kinetic contributions to the stress: the electronic one and the ionic one. If we understand correctly, what you suggest removing, with the keyword "Kinetic", is actually the electronic contribution, which should be included and does not has the concept of (ionic) temperature/velocity. Therefore, we think the line we used with the keyword "in kB" should be correct. BTW, the ionic contribution should be in the line with keyword "kinetic pressure (ideal gas correction)". I didn't find a good reference but if you have VASP 5.4.4, you could check the source code in force.F below the line "sum the ionic kinetic contributions to the stress tensor". Best,

…

On Thu, Apr 9, 2020 at 8:13 PM sivonxay ***@***.***> wrote: When training a model using VASP AIMD data, should the kinetic contributions to the virials be subtracted out? I took a look at how it's done in https://github.com/deepmodeling/dpdata/blob/master/dpdata/vasp/outcar.py and that seems to include the kinetic contributions, since it parses the lines beginning with "in kB". [image: Screen Shot 2020-04-09 at 4 59 48 PM] <https://user-images.githubusercontent.com/25111118/78950414-86da9d80-7a83-11ea-95eb-60e776e443bc.png> I have run VASP AIMD of my system at various temperatures (1000, 1500, 2000 etc.), each at a volume that yields approximately 0 external pressure (Using ISIF=2). Thus, the virial data for each of these would be similar, although their volumes are very different. Not excluding the kinetic contribution would imply that the training/model has some concept of the temperature/velocity of the training data - which is data not included in the training data. Thanks in advance for your response, Eric — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <https://github.com/deepmodeling/deepmd-kit/issues/213>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AEJ6DCZPFC2JWLBTIHATN3TRLZQEHANCNFSM4MFDWZHA> .