[gmx-users] osmotic pressure in Gromacs

[chris.neale at utoronto.ca]

>    We want to simulate water transport through channel by osmotic   
> pressurce using Gromacs software. The force on the water molecules   
> is usually applied only in a buffer zone, not on all water   
> molecules. How to set up my system. In my opion, we can only define   
> a simple group by index. Water molecules are moving, so the water   
> molecules in a buffer zone change frequently. How to define the   
> group of water molecules in a buffer zone. Gromacs can do it?
>     thanks!

First: what is a 'buffer zone'? And why do you need one? Also, you  
should be able to model osmotic pressure with the application of  
special forces, no?

Perhaps I entirely missed the point. If not, here is what came to mind.

One admitedly strange way to apply forces based on position would be  
to run gromacs in very short segments using the pull code to apply the  
forces and every time you stop gromacs run your own scripts or gromacs  
utilities to determine which waters should be feeling your force and  
make a new .ndx file to cause that to happen in the next mrdun  
segment. You should then determine how long to run your segments based  
on the diffusion of normal water and take an average of how long it  
before recalculating. I figure 0.4A^2/ps diffusion suggests segments  
of < 10ps (but don't take my word for it) making this a very time  
consuming endeavour.

Also not that this method would destroy dynamics information about the waters.