[gmx-users] how to set "gen_vel" in mdp file and continue

X.Periole X.Periole at rug.nl
Tue May 24 08:52:46 CEST 2005

> I still have another question of "gen_vel".
> Does "gen_vel=yes" or "gen_vel=no" influence the mdrun 
>result much?

> Or what kind of properties will be greatly changed as 
>"gen_vel" option changed?

You probably know that his option in mdp file gives you 
possibility to generate velocities according to a
temperature you choose for your system. The effect of
generating new velocities on a system to simulate is not
define. You need initial velocities to start a MD run ...
that is a requieremnt of the algorithm of propagation.
After this, one do not generate new velocities for the
system. Generating new velocities in a continuation run
is almost the same as doing a new simulation from the
start, except that the inital conf is different and "had"
a taste of the temperature already.
One can use this gen_vel option (with different seed 
to start several simulation in different "direction", and
thus increase the conformational space exploration, 
of running one simulation. The former solution is 
more efficient but at the end (full conformational space
explored) this should not make a difference on the
properties of the system.

Basicaly if you generate new velocities in the middle of a
simulation you could explore a new part of the
conformational space, which might be of different 
properties for the system. It is rather unlikely that the
properties change much, it is of you to check.

Hope this helps.

More information about the gromacs.org_gmx-users mailing list