[Pw_forum] variable cell MD target pressure

Davide Ceresoli davide.ceresoli at istm.cnr.it
Wed Oct 23 14:10:45 CEST 2013

Dear Luis,
     I'm the culprit who coded output of the kinetic stress, but
without adding it to the total stress. Are you sure it is added
to the total stress?

The issue is that it is not clear whether the "Ion kinetic term"
should be there. In my opinion it should be added to the total stress,
as it is done in classical MD and in the CPV code.

However, I tried to perform NVT MD runs of a liquid metal at different
volumes, and to compare by d^2<E>/dV^2 with the average stress computed
with and without the ions kinetic term. Unfortunately results were

Therefore, in order not to break the behavior of the code, I decided to
print it, without summing it to the total stress.

Comments on this topic are very welcome.


On 10/23/2013 10:21 AM, luisen wrote:
>   Dear all,
>   I am writing to ask for confirmation about a "strange" feature in
>   the program, that concerns molecular dynamics calculations at finite
>   temperature (non-zero) with variable cell.
>   As far as I have experienced it seems to me that the target pressure
>   that one states in the input file is the total pressure, I mean, including
>   the kinetic term coming from the motion of the ions due to the temperature,
>   while the stress tensor and pressure writen in the output file includes only
>   the potential terms, but not the kinetic ones.
>   This message is just to request a confirmation from the developers that I am
>   correct about this.
>   Note that this feature is very confusing for users; I could spot it because
>   the system I am studying is liquid Be, which has a very high temperature and
>   a very high number density, so that the kinetic part of the pressure is 20
>   kbar. So it is really confusing asking for 0 kbar and obtaining
>   oscillations around -20 kbar in the output file.
>   May I suggest to modify it in the future release of QE?
>   Yours,
>       Luis Enrique Gonzalez
>       Universidad de Valladolid

More information about the users mailing list