AMBER Archive (2005)

Subject: Re: AMBER: replica exchange

From: Carlos Simmerling (carlos_at_ilion.bio.sunysb.edu)
Date: Fri Aug 26 2005 - 10:16:51 CDT


the initial structures can be the same or not, both have
been used in published studies. The conformations can be random,
from equilibrated runs, etc. They can be equilibrated in advance
or they you can discard part of the REMD results as the equilibration
period. Temperature spacing depends on system size and should be
exponential, articles have been published on optimal replica
spacing (do a google search for replica exchange optimal spacing).

Overall, REMD provides many more options than normal MD and
you'll need to read the literature to see what others have done.
It's not a method to use until you are very comfortable with regular
MD and especially with analyzing simulation convergence. REMD is not
magic and you need to be much more careful with convergence than
you need to be with normal MD (since many REMD simulations can
appear to be converged long before they really are).

quoted message:
I am trying to run a replica exchange simulation. I have some doubts in how to
setup the system. First, the initial structure for each replica is the same
or different conformations??? If different conformetions are needed, how to
obtain those conformations. Second, the initial structure for the replicas
need to be at the temp0 specified in the mdin? or all replicas start for the
same temperature (tempi=0.0 temp0=temp) of the replica?. Third and last, How
to select the temperatures for the replicas?? any method???.

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu