From: Josh (
Date: Wed Aug 09 2023 - 13:49:47 CDT

Hi Nakshatra,

Did you use a wrapped trajectory? Knowing very little about the
diffusion coefficient tool, which isn't part of a standard VMD
installation, that is where I'd guess things went wrong. If you have a
simulation that is wrapped, the mean squared displacement found in the
trajectory is bounded by how big your box is, which is why you often
will unwrap your trajectory prior to calculating a diffusion coefficient.


On 8/9/23 12:25 PM, Nakshatra Upadhyay wrote:
> Dear all,
> I have been trying to calculate the diffusion coefficient of water
> using VMD Diffusion Coefficient Tool. However, I got an extremely low
> diffusion coefficient value for water (Experimental value ~2.3 m^2/s),
> 0.0002266 angstrom^2/ns (2.66 m^2/s) was the fit value for diffusion
> coefficient. I used a 100ns trajectory having 5000 frames (data saved
> in every 10000 steps and time step was 2.0 fs) to calculate the
> diffusion coefficient. I used 500 to 2501 frames as lag time and
> calculated the diffusion coefficient with a 100-frame interval
> (Although I got almost the same values with different lag times). The
> VMD Diffusion Coefficient Tool calculated MSD value was almost
> constant (~720 angstrom^2) for the selected lag time. Could you please
> suggest me how to calculate the diffusion coefficient using VMD?
> Sincerely yours,
> Nakshatra

Josh Vermaas
Assistant Professor, Plant Research Laboratory and Biochemistry and Molecular Biology
Michigan State University