[gmx-users] temperature scaling and frozen atoms

[Xavier Periole]

Michael

> Using restraints (harmonic potentials) instead of
> constraints is probably no good idea since I would
> have to use a very large force constant and
> probably that would mean a shorter time step.

I would not worry about that. I did use big restrains on 
a system (for equilibration though) and did not have any 
problem time step. It is not a real bond you describe.

> However, I did manage to get it working,
> There are two things one has to do:
> 1) never set the center of mass motion to zero
> (no problem if the comm is zero or close to zero
> at the beginning of the simulation)
> 2) don't define two seperate temperature scaling
> groups for the frozen and the not frozen parts but
> only one (the entire system).
> 
> If both of the above conditions are given it works,
> strange, isn't it ?
> The system would not explode anymore and the dynamics
> looks reasonable. I am still a bit concerned, however,
> about whether the EOM are solved correctly with this
> setup and whether the temperature of the non-frozen
> part given in the output is really correct.
> 

I don't know how the program works but probably takes into 
account that part of the system is frozen : loss of degrees of 
freedom that should be considered in the calculation of the 
temperature and velocity scaling. Perhaps you can rerun the 
trajectory defining your temperature groups differently 
(separate frozen and non-frozen) and see if the frozen is at 
0 K and the non-frozen at the T you asked.

XAvier