[gmx-users] seperating bilayers (again)

[Mark Abraham]

Alan Dodd wrote:
> Hello all,
> I've been having some trouble recently with the
> leaflets in my bilayer gently drifting apart over a
> period of a few hundred picoseconds.  Increasing the
> vdW radius hasn't helped (up to 2nm), although it did
> delay the seperation.  Likewise, increasing tau_p from
> 5 up to 10.  The really bizarre thing is that I've
> been simulating peptides in bilayers using 3.2.1 on
> our cluster for about 18 months now, using the
> lipid-modified version of ffgmx and the parameters
> listed below, and it's only since I've started using
> 3.3 on a new machine that I've had any trouble.  Any
> suggestions on what might be going wrong?

> pme_order           =  8

pme_order != 4 is known to be buggy in the distribution version of 3.3 - 
see multiple mailing list threads. There is a fixed pme.c on the ftp 
site that you can find somewhere on the gromacs webpage.

Mark