[gmx-users] Problem with accelerations

[Kamps, M.]

Dear GMX users,

I have some strange behaviour which I cannot explain.

I want to accelerate atoms through my box at a certain velocity. Since
I can only adjust the acceleration, I have to trial-and-error my way
to the right accelerations.

To do this, I create a smaller 'testing' simulation, which is
continued from an extensive equilibrium. I apply an acceleration of
0.05 nm/ps2, and after 200ps my velocity (gathered from gmx traj with
an gmx select input) says that the velocity is stable at around 0.04
nm/ps. throughout the simulation the velocity slightly increases due
to the atoms rearranging etc.

Now I want to simulate the same behaviour, but for a longer amount of
time. I therefore take the exact same MDP file, and change nothing
except the time-related parameters. I change the number of total steps
and the timestep, but leave the acceleration intact. After analysing
the data I get a MUCH higher velocity! I can understand this due to
rearranging of the atoms on the longer term, but after the same 200ps
the velocity is also way higher.

So during my smaller run the velocity was 0.04 nm/ps after 200 ps, in
this longer run, the velocity is 0.2 nm/ps after the same 200ps. How
is this possible? The exact same acceleration is applied.

Am I missing something?