[gmx-users] gmx potential and water

[Alex]

Hi all,

What we have here is a well-equilibrated, well-behaved, 300 ns long 
simulation of an aqueous system (tip4p, ~50K atoms) at 300K and a 
version of gmx_potential.c hacked by one of our very well known members. 
It allows plotting potential in a all of the slices as functions of 
simulated time.

I am looking at the time data produced by selecting just water as the 
contributing charge group. The value of epsilon is default. The result 
(for a given slice) is a massively noisy dataset, basically oscillations 
around zero, as one should expect. However, I have run this data through 
a very stringent lowpass filter (basically a recursive running average), 
so everything beyond a few hundred of MHz is gone. The filtered data 
features oscillations with a magnitude of, well, 0.5 V.

Is this expected? Is this normal? Does it need more smoothing? Does it 
need rescaling by eps=80? I thought water should "self-screen" here... 
Also, there's a caveat: the trajectory is spit out every 50 ps (can't go 
finer, huge files), so it could be a classic case of aliasing. Any 
attempt at an explanation would be great. :)

Thanks you,

Alex