[gmx-users] Non-equilibraium MD: simulating a moving plate.

[Vitaly V. Chaban]

Hello,

I want to simulate a rigid plate (wall) moving along one direction and
reflecting all the particles it meets.

I have made a potential only for repulsion which acts at the distances
less than 1 nm (and equals zero at bigger ones) and bound it to a
particle. If the plate is immovable everything goes OK. To make it go
I use

acc-grps                 = WAL
accelerate               = 0 0 1

where the acceleration is in [nm/ps2] (from the gmx manual 4.0).

To make a plate motion insensitive to the molecules it reflects, its
mass was set to a very big value. But during the following run the
plane moved very slowly while all other particles in the system have
obtained an acceleration at the same direction. Even those ones which
are to far to interact with a plate. Why? Then if the mass of the
plate is reduced the above effect is not so pronounced but then the
plate becomes very sensible to the interactions, so its motion is not
as we want...
If the acceleration is given in  [nm/ps2], how the mass can influence
the motion? I thought this value is just added to the value obtained
from the interations, but it's evidently not true here.

Please let me know if it's possible in gromacs to make such moving
plate I described and if so, what is a good practice to simulate this
effect?


Thanks you,
Vitaly