[gmx-users] g_msd vs g_velacc

[Vitaly Chaban]

I've got a discrepancy calculating the diffusion coefficient via
Green-Kubo equation and via Einstein equation.

For methanol (200ps, dt=0.001):
frequency to write velocities = 5
frequency to write coordinates = 100

$ g_msd
2.3 (and it is true)

$ g_velacc -nonormalize -acflen 1001
$ g_analyze -f vac.xvg -integrate
1.3 (and it's almost two times lower...)

Do I miss any important keyword here? It seems the result of integration is not
correct in this case.


Thanks,
Vitaly V. Chaban