[gmx-users] COM motion removal problem

[Erik Lindahl]

Christoph Freudenberger wrote:

> I already guessed so, because if you don't set any comm-parameters
> grompp will automaticaly set comm_mode=linear and nstcomm=1.
> but why than the "should only be used for vacuum simulations"-remark? 

That's the "angular" option to comm_mode (which used to be set by 
providing a negative number to nstcomm, but comm_mode is the preferred 
way now).

When you simulate in vacuum you can get a rotation buildup in the system 
in the same way you get a linear velocity buildup with a periodic box, 
and this resets the overall rotation of the system.


Cheers,

Erik