[gmx-users] nstcomm value

[avinash kumar]

Hello all,

           I am simulating pressure driven flow in a nanochannel. To
introduce a pressure gradient in the system I have put all the liquid
atoms in one group and applied acceleration to it. Also I am
periodically removing the centre of mass motion my nstcomm value. My
doubt is that when I am changing the value of nstcomm from say 100 to
200 the velocity profile of the channel is almost doubling. Also the
behaviour of the channel (read velocity profile) to values of nstcomm
is dependent on the value of the acceleration. My question is how will
we decide what is the value of nstcomm depending on a given
acceleration. Is there any criteria of deciding the above thing.

Avinash Kumar