[QE-users] Clarification on the diffusion coefficient calculation

Kazume NISHIDATE nisidate at iwate-u.ac.jp
Fri Mar 24 14:37:17 CET 2023


Dear Mauro,

> I suppose that the value printed by Quantum Espresso is calculated
> using a certain number of MD timesteps. So it should discard the
> initial part of the trajectory.

It may be always safe to discard the first part of the trajectories 
since it started from a particular initial configuration in a quasi 
equilibrium state.


Below is only my thought but may work.

You should first calculate the system for a few thousands time steps
with temperature regulation (say at 300 K). And you obtain the
diffusion coefficient D1 at 300 K evaluated by the QE.

Now, hopefully, the system is in an equilibrium state.

Next calculate the subsequent MD steps using the option

restart_mode = 'restart’,

After the calculation, you will obtain a diffusion coefficient D2
calculated by QE. D2 at the equilibrium state is much more reliable 
than the D1 at the quasi equilibrium state.

Repeat the above procedure until you convince can get a converged
diffusion coefficient D at 300K.



Note that it is also not so difficult to evaluate the diffusion 
coefficient directory from the trajectories written in the 
output file.



西館数芽
Kazume NISHIDATE Ph.D

Department of Systems Innovation Engineering,
Graduate School of Science and Engineering, Iwate University
4-3-5 Ueda, Morioka, Iwate, 020-8551 JAPAN
Phone:+81-19-621-6391
kazume.nishidate at gmail.com, nisidate at iwate-u.ac.jp
https://sites.google.com/site/nisidatelab/



More information about the users mailing list