[gmx-users] Supradiffusivity in a bilayer

[Erik Lindahl]

Hi,

On Oct 10, 2005, at 5:10 PM, Rama Gullapalli wrote:

> Dear GMX ers
> I had a simple question regarding the calculation of
> lipid diffusion.
>
> 1) How do i remove the supradiffusivity of the bilayer
> when calculating the diffusion constant?

If you haven't started the simulation yet, define three separate  
groups for center-of-mass motion removal: water and a separate group  
for each lipid layer.

If you already started it, you can use a trick that might require  
some coding. Go through the trajectory, remove periodicity and  
calculate the center-off-mass for each frame. Calculate how much the  
COM moves, and then remove this motion from each frame.

>
> 2) How do i couple the COM removal of the individual
> bilayer leaflets while running the simulation? Is it
> through make_ndx, assign names and couple them
> seperately? Also, is it necessary to do the same in
> order to obtain correct diffusion coefficients?

You've essentially got it :-)

The COM motion needs to be taken into account one way or the other,  
since the periodic system means there isn't any natural friction for  
collective XY motion. Doing it before is easier, but we managed to  
publish the first reasonably accurate diffusion constants by coming  
up with the second method - 100ns of trajectories was a lot at the  
time, and we weren't exactly thrilled with the idea of rerunning it  
all :-)

Cheers,

Erik



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20051010/bd59e607/attachment.html>