[gmx-users] question regarding friction coefficient in stochastic dynamics

[Thomas Schlesier]

hi all,
i want to perform stochastic dynamics (sd) simulations in implicit 
mesitylene. and first i wanted to understand how the value of the 
friction coefficient for water sd simulations is chosen.

gromacs manual suggests a value of 0.5 1/ps (which is lower then the 
internal friction of water; but for this i couldn't find a value). in 
the literature i have found most often the value 10 1/ps, but it even 
goes up to 50 1/ps (so the values i find differ by 2 orders of
magnitude).
but nobody really tells why they use this value.

then i tried to calculate the friction coefficient, via
gamma = kT / D m
gamme - firction coefficient
kT - thermal energy
D - diffusion constant of water (calculated via the msd of a 10ns water 
simulation with around 2000 molecules - D = 4*10^-5 cm^2/s)
m - mass of water

in the end i get a value of around 0.03 1/ps, which is an order of 
magnitude lower then the lowest value which i found in the literature.
i literature i find for D(h2o) a value of about 2*10^-5 cm^2/s, but this 
doesn't really change the picture.

so can anyone shed some light on the topic?
and has anybody an idea, how i can find a value for the friction 
coefficient of mesitylene?

greetings
thomas