[gmx-users] MSD and frequency of writing trajectories

[Anna Akinshina]

Dear Gromacs Users,

I have a question interpreting obtained msd data.
I need to calculate diffusion coefficient for a single argon atom in a box of water (500 molecules).
During the calculations (50ns) I write both trr and xtc trajectories, but to save space I write trr for whole system very seldom (every 50000 steps = 100 ps)
but in xtc I have argon only (no water) and write more often (every 1000 steps =  2 ps).

When I analysed the results using g_msd I was surprised that the msd  curves obtained from trr and xtc trajectories differ a lot, what affect the resulting diffusion coefficients.

If I save trr more often, like every 10000 steps (20ps) the results from both  files coincide.

Does the msd results depend on how often you write trajectories?
If yes, what frequency for writing the trajectories for future calculation of the diffusion coefficient would be reasonable and how long should be the calculations?

Thanks you
Anna
-----------------------
Dr Anna Akinshina
School of Chemical Engineering & Analytical Science
The University of Manchester, UK