[gmx-users] Re: T-Coupling and COM removal

[Markus Miettinen]

On Feb 20, 2008, at 19:36 , gmx-users-request at gromacs.org wrote:

> Message: 4
> Date: Wed, 20 Feb 2008 09:15:33 -0800
> From: "Ilya Chorny" <ichorny at gmail.com>
> Subject: [gmx-users] T-Coupling and COM removal
> To: "Discussion list for GROMACS users" <gmx-users at gromacs.org>
> Message-ID:
> 	<5f5f916d0802200915y1beff759ye244b11775355553 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
>
> HI All,
>
> I ran some membrane protein simulations and got some strange  
> results. The
> RMSD (~.18 nm) of the protein equilibrated quite nicely but when I  
> look at
> the trajectory the protein seems to be diffusing way to much over  
> 20ns. The
> protein has about 1000 AA. I calculated the KE of the protein,  
> lipid, and
> water and got a KE that was  1.5RT, 1.2RT, and .8RT respectively.  
> My T is
> 298K.
> My COM removal and T-coupling is for the whole system. I am running  
> 3.3.1.
> Is there a problem with T-coupling and COM removal on the whole  
> system?

If you have set nstcomm=1, you should not experience a drift.

For some reason, however, the center of mass of the whole simulation
can slowly drift, even if explicit removal (nstcomm = 1) has been set.
A Bugzilla has been posted, please see

   http://bugzilla.gromacs.org/show_bug.cgi?id=165

for further details.


Cheers,
Markus Miettinen