[gmx-users] Re: temperature scaling and frozen atoms

[Berk Hess]

Date: Thu, 31 Jul 2003 16:20:05 +0200
Mime-Version: 1.0
Content-Type: text/plain; format=flowed

Hi,

I the past I have fixed several details concerning freeze groups.
According to me everything should work fine now.

I just checked a system of 4 atoms with 2 atoms frozen and 2 atoms free.
Everything works fine, even with seperate comm removal groups and seperate
temperature coupling groups, using both Berendsen and Nose-Hoover coupling.

When constraints are present in the frozen group things get a bit more 
tricky,
since if the starting structure does not obey the constraints perfectly one 
can not
have the whole group frozen and the correct constraint lengths at the same 
time.
The current implementation also corrects constraints in frozen groups.

Still I don't see why you can not use comm removal or seperate temperature 
coupling groups.
Note that comm removal is often not what you want when a large part of the 
system is frozen,
since you might not want to restrict motion with respect to the frozen part.

A problem in one of the mdp files you showed is that the tau_t of the 
Nose-Hoover
thermostat is 0.02 while the timestep is 0.002. The tau_t sets the period of 
the bath oscillations
and 10 timesteps per oscillation is not enough for accurate intergration.

Also you want to have rlist equal to rcoulomb and not to rcoulomb_switch.

Berk.

_________________________________________________________________
Protect your PC - get McAfee.com VirusScan Online 
http://clinic.mcafee.com/clinic/ibuy/campaign.asp?cid=3963