[gmx-users] Diffusion Coefficient from VAC

[Frederic Leroy]

What do you mean quantatively by "VAC does not agree with one calculated 
from MSD"?



Alexandre Suman de Araujo wrote:

> Hi Gmxers.
>
> I´m trying to calculate coefficient diffusion values for ions in water.
>
> I made some calculations using MSD and the results were in good 
> agreement with experimental values.
>
> The problem is when I try to calculate it from integration of VAC 
> function for the ions, like in Rasaiah paper (J. Chem Phys. 101(8) 6964).
>
> I made 40 simulations of 250 ps with 1 fs time step and output of 
> velocities every 1 fs. From these simulations I calculated the 
> normalized VAC function from these 40 simulations and calculated the 
> mean. This methodology is implemented to have a good statistic and a 
> more smooth VAC function (without fluctuations that could make the 
> integral diverge). With this VAC function I use this equation to 
> calculate do diffusion coefficient:
>
> D = kT/M*Y, where Y is the integral of the VAC function
>
> The problem is that the value calculated from VAC does not agree with 
> one calculated from MSD.
>
> Does anyone knows any trick to perform this kind of calculation? In 
> the paper cited above these two values are very near and I´m using the 
> same methodology.
>
> Thank´s!
>