AMBER: Replica exchange rate in REMD

From: Seongeun Yang <>
Date: Thu, 21 Dec 2006 20:42:59 +0900

Hello all,

I have some general questions in running replica exchange and conventional MD simulations.

1. Do the replicas always have to be exchanged after their potential energies are relaxed enough to equilibrium?

2. If the total potential energy relaxes quite slowly due to a big system size (e.g. a polypeptide+solvent molecules),
then is it reasonable to reduce the number of solvent molecules to a degree
that the buffer size in making a cubic simulation box is as small as 3.5 angstrom?

This is what I found in some papers cited quite frequently.
(The nonbonding cutoff could not be found explicitly.)

3. If the nonbonding cutoff is set to be 8.0 angstrom in md imput,
then how big the buffer size in making a cubic box should be at least?

4. If the solute polypeptide can be unfolded during simulations,
then the buffer size should be set larger than at least the longest end-to-end distance of the unfolded solute, right?

I'm wondering the answer to the last question is really 'yes' to so many simulation studies reported in the literature.

If the answer is 'no', what is the 'secret' in using a seemingly very small simulation boxes?

Thanks for your reply in advance.

Seongeun Yang
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Dec 24 2006 - 06:07:23 PST
Custom Search