Dear Sir/Madam;
I compiled the Parallel/REM version of amber8 and tried testing it.
A problem arises when I try to test the REM simulation (all the other
tests pass)
I noticed that the Makefile (in the test directory) calls for the
following actions when testing rem:
test.sander.REM:
cd rem_gb; Run.rem
cd rem_vac; Run.rem
cd rem_wat; Run.rem
I don't have those directories (in amber8/test). I do have a directory
named rem. In there, there is a Run.rem script which I ran (I edited the
Makefile to go into that directory and run the script). I used 4
processors and the simulation ran to completion. However the output files
look different (although, not unreasonable).
The first few lines of the rem.log I produced looks like this:
# replica exchange log file
# Replica #, Velocity Scaling Factor, T, Eptot, Temp0, NewTemp0
# exchange 1
1 -1.00 275.68 -547.98 267.00 267.00
2 -1.00 298.43 -544.73 283.00 283.00
the rem.log.save file looks like this:
# replica exchange log file
# Replica Number, Scaling Factor, T, Eptot, Temp0, NewTemp0
# exchange 1
1 -1.00 268.20 -541.88 267.00 267.00
2 -1.00 277.70 -529.46 283.00 283.00
Using the coordinates and mdin file for replica 1, I performed the
simulation up to the point right before exchange using a serial
version of amber8 (one that passses all the
sander tests as well). I get
NSTEP = 100 TIME(PS) = 45646.100 TEMP(K) = 275.68 PRESS = 0.0
Etot = -353.4957 EKtot = 194.4827 EPtot = -547.9784
A V E R A G E S O V E R 100 S T E P S
NSTEP = 100 TIME(PS) = 45646.100 TEMP(K) = 268.44 PRESS = 0.0
Etot = -352.5915 EKtot = 189.3727 EPtot = -541.9643
So it looks like the rem.log.save file used the average values for energy
and temperature. Now, I know that's incorrect, and perhaps that's from
some earlier version. So it looks like the code is working properly (it's
certainly picking out the adequate energy and temperature). Although I'm
left uncertain if the compilation worked properly. Has anyone else come
across this problem before?? Any simple tests that can be run??
Thanks
-Sergio
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Fri Sep 10 2004 - 13:25:59 PDT
