[AMBER] Replica exchange restart file

From: Mele N. <nm10g13.soton.ac.uk>
Date: Fri, 28 Mar 2014 12:15:17 +0000

Dear Amber user,

I running a replica exchange simulation for a peptide in explicit water. When the simulation was done I restart it. When I had a look on the log file from the second simulation run I realized that the simulation did not take care of the target temperature change.
The target temperature in the second run did not correspond to the temperature at the end of the first one.

For the second run I used the same remd.mdin file and I just change the coordinate input file in sander command:

Input file:

Production
 &cntrl
   irest=0, ntx=1,
   nstlim=1000, dt=0.002,
   ntt=3, gamma_ln=1.0,
   temp0=XXXXX, ig=RANDOM_NUMBER,
   ntc=2, ntf=2, nscm=1000,
   ntb=1,
   cut=10.0,
   ntpr=100, ntwx=1000, ntwr=100000,
   nmropt=1,
   numexchg=15000,
 /
 &wt TYPE='END'
 /
DISANG=file.dat


Sander command with the previous run rst file:

-A -rem 1 -remlog rem_r2.log -i remd.mdin.001 -o remd_r2.mdout.001 -c remd.rst.001 -r remd_r2.rst.001 -x remd_r2.mdcrd.001 -inf remd_r2.mdinfo.001 -p file.prmtop
........
........

I had a look on Amber manuel and to restart a simulation it's write that the temperature change is taken into consideration.
Did I forget something?

Amber manual:

"Restarting REMD simulations
It is recommended that each REMD run generate a new set of output files (such as mdcrd),
but for convenience one may use -A in the command line in order to append output to existing
output files. This can be a useful option when restarting REMD simulations. If -A is used,
files that were present before starting the REMD simulation are appended to throughout the
new simulation. Note that this can seriously affect performance on systems where the file
writing becomes rate limiting, although the new implementation of REMD should help with
this somewhat. If -O is used, any files present are overwritten during the first iteration, and then
subsequent iterations append to these new files.
At the end of a REMD simulation, the target temperature of each replica is most likely not
the same as it was at the start of the simulation (due to exchanges). If one wishes to continue
this simulation, sander will need to know that the target temperatures have changed. Since the
target temperature is normally specified in the mdin file (using temp0), the previous mdin files
would all need to be modified to reflect changes in target temperature of each replica. In order
to simplify this process, the program will write the current target temperature as additional
information in the restart files during an REMD simulation. When an REMD simulation is
started, the program will check to see if the target temperature is present in the restart file. If
it is present, this value will override the target temperature specified using temp0 in the mdin
file. In this manner, one can restart the simulation from the set of restart files and the program
will automatically update the target temperature of each replica to correspond to the final target
temperature from the previous run. If the target temperature is not present (as would be the case
for the first REMD run), the correct values should be present in the mdin files."


Thanks a lot,




_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Mar 28 2014 - 05:30:03 PDT
Custom Search