[gmx-users] How to calculate the diffusion coefficient.

李佳翔 jxli at ciac.ac.cn
Tue Sep 20 08:29:52 CEST 2016

Dear all,

    I want to calculate the diffusion coefficient of a single coarse-grained polymer adsorbed on a wall with Gromacs 5.0.7, where the wall's type is 9-3 and the density is set as 1. Firstly, I use the "gmx msd" command to calculate the diffusion coefficient. The calculation error is very large and the results of the parallel samples are always different. For example, I find the curve of MSD-time is approximately linear, but the result of the "gmx msd" is incorrect, which shows:

       <D> = 97.9395 Std. Dev. = 0.0000 Error = 0.0000

       Fitting from 20 to 100 ps

       D[    System] 99.2119 (+/- 31.8833) 1e-5 cm^2/s

Then, I try to use "gmx velacc" command to calculate the diffusion coefficient. Similarly, the fluctuation of different parallel samples is still very large, and I occasionally find the integral of the VACF is negative. In addition, I try to change the “dt” and the “nstxout”, which can’t work. I have made numerous attempts to solve this problem. Unfortunately, I have not succeeded.

     I would be grateful if someone can help me solve this problem or provide some suggestions.


Best regards.

Sincerely yours,

Jiaxiang Li

More information about the gromacs.org_gmx-users mailing list