[gmx-users] diffusion coefficient

[Amit Choubey]

Hi Everyone,

I have been trying to calculate diffusion coefficient for water. I am trying
to reproduce the numbers published in journal papers.
I am using SPCE water model. I use the g_msd analysis tool.

g_msd -f traj.trr -n index.ndx -s npt.tpr -b 2 -e 8

I use a box of volume 6x6x6 nm^3 which has 7161 water molecules. I
equilibriate the system for a ns and then run for additional 10 ps for
analysis. Here are some of the numbers that i get

a. With Berendsen's T coupling and P coupling on i get 4.4941 (+/-
0.2992) 1e-5 cm^2/s
b. With Berendsen's T coupling on and P coupling off I get 3.2469 (+/-
0.1076) 1e-5 cm^2/s
c. With Berendsen's T coupling on and P coupling off for 1ns and then T,P
coupling both off (for analysis part) i get 2.8085 (+/- 0.0310) 1e-5 cm^2/s
.

c is closest to the widely accepted experimental value of 2.3 1e-5 cm^2/s
but its not quite right.

Could someone explain to me why the values obtained in above 3 cases are
widely different and may be give some tips about the right procedure to
calculate diffusion (method and invoking the g_msd tool)?

Thank you
Amit
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20100210/76d6c6a5/attachment.html>