[gmx-users] Re: About the parameters for REMD in vacuum

[Xavier Periole]

Dear Ozge,

lets keep the discussion in the gmx-users list.

> Thank you for your quick reply! I used the temperature distribution that was 
>used in your paper: starting from 275.6 to 399.6. 

This distribution was chosen to give an exchange ratio of ~0.3 on the
system used: a beta-peptide in methanol. Given that you change the
system and thus the degrees of freedom you have adapt your temperatures.
Their distribution has to be chosen such that the overlap of total
potential energy of the system at consecutive temperatures allows
exchange. You can tune the separations between the temperatures to
vary the exchange rate.

> I used the 0.002 ps as time step, but I could not understand what you 
>suggested exactly. You asked the time step for temperature coupling? Is is 
>true? if so, the temperature coupling constant was 0.1. 

I was referring to the integration time step, which is the time step you
use to integrate the equation of motion that are used to propagate your
system in time.
If you go to higher temperature the thermal energy is such that the
motions of the atoms are increased and you need to decrease the time
step of integration to be able to describe the evolution of the system.

That might be the reason you see LINCS warnings.

Then the fact that you do not use explicit solvent (no viscosity)
will increase the motion of the particles (atoms) and may add to the
problem described above.

> What temperature coupling constant and temperature interval will be 
> appropriate for simulation in vacuum via REMD?

delta=0.001 ps might be enough. You need to make some trials.

But again, simulating in vacuo is almost useless, unless you want to
show that the sampling is inaccurate.

> Have you ever performed REMD in vacuum? 

Yes to show that you get erroneous sampling. This is well accepted
also there are few papers showing it explicitly. Same thing for the
use of implicit solvents but this is more documented in the literature.
Look for papers of AE. Garcia and co-workers.

> Ozge

XAvier

-----------------------------------------------------
XAvier Periole - PhD

NMR & Molecular Dynamics Group
University of Groningen
The Netherlands
http://md.chem.rug.nl/~periole
-----------------------------------------------------