Re: box size changes under anisotropic pressure coupling

From: Philip Blood (
Date: Fri Apr 08 2005 - 14:28:11 CDT

Now that I think about it some more, I am not surprised to see this
behavior. I think this is just a small system size effect. I heard
from another person who saw the opposite box size change: 1 angstrom in
the x direction and 8 angstroms in the y. I think with a small system,
for any given configuration there will be very large differences in the
pressure in the two directions that can cause it to relax much more in
one direction. In fact, I checked a simulation I did that was 10x
larger (720 lipids) with anisotropic pressure coupling and I saw nearly
isotropic box size changes in the x and y directions.

For the small system I think I will allow it to equilibrate under the
constant x-y ratio condition and then go from there.


Jerome Henin wrote:

>Hi Philip,
>Actually, I did see a similar behaviour on some lipid bilayer I have simulated.
>Until now, I assumed it to be a consequence of intrinsic properties of my
>system, but I would be interested to know if many people experienced the same
>situation, and if it somehow could be linked to the barostat.
>>I am equilibrating some lipid bilayer configurations (72 lipids) with a
>>fully flexible cell and I see something rather odd. I start the
>>configurations at a higher area per lipid than given by experiment and
>>I expect the bilayer to relax to a lower area per lipid, which it does.
>>The odd thing is that all the relaxation occurs in the x direction,
>>while the y dimension of the box stays at about the same average
>>length. I realize that this could be a result of some weird thing with
>>the initial conditions, but I have seen this same behavior with two
>>separate configurations.
>>So I was wondering if those of you who have simulated lipid bilayers
>>with anisotropic pressure coupling or who have used anisotropic pressure
>>coupling in general could comment. These are isotropic systems (in the
>>x-y plane), so it is strange that the cell would shrink over 8 angtroms
>>in the x direction and only 1 angstrom in the y direction (this is after
>>10 ns of simulation time). Could there be anything in the pressure
>>routine itself that might cause some bias? Again I can't rule out that
>>this happens by chance, but it might be worthwhile to look into this.
>>Phil Blood

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:40:40 CST