[gmx-users] SD velocity correction

[Berk Hess]

>I am using sd for my simulations and have a question about it. Could
>someone explain to me why the kinetic energy is corrected for the
>friction and where the form of the correction (i.e. exp(-e
>delt)+sqrt(2)(1-exp(-e delt)) comes from?

When the friction is very high, the directions of v(t-dt/2) and v(t+dt/2)
are almost not correlated. If one still wants a measure fot the velocity
at time t, the average needs to be corrected by sqrt(2).
The deterministic component of the velocity is exp(-xi dt) and
the random component is 1-exp(-xi dt).

Note that this correction only influences the kinetic energy output values
and temperature output values, and in no way influences the simulation 
itself.

Berk.

_________________________________________________________________
MSN 8 helps eliminate e-mail viruses. Get 2 months FREE*. 
http://join.msn.com/?page=features/virus