[gmx-users] computing REMD swap proba

pascal.baillod at epfl.ch pascal.baillod at epfl.ch
Wed Dec 28 17:14:29 CET 2005

Thank you very much, Mark and David!

3 more REMD questions if you will..

1) I would like to modify the REMD protocol so that the probability of a swap is
computed with the potential energy of a subsystem (prot-prot and prot-solvent
interactions, but no solv-solv).. To do so, I have been studying the routines in
repl_ex.c. I am tracking back the epot variable, which so far brought me to
ener[F_EPOT] in md.c. Is there any description, somewhere, of the ener array
elements I need in order to get the above mentioned protein pot energy? With
energy groups in the .mdp files, I can compute these protein energies..

2) I tried to recompute the swap probability. I only get the same result as the
gromacs code if I use the target temperature and not the current simulation
temperature in the formula, i.e. using the target 300.0 of the mdp input instead
of using the current 301.2 of the log file for that step. Shouldn't one use the
current temperature, more representative of the conformation to be exchanged? I
wonder if there are similar problems for the P*V term (although it is negligible)..

3) I still do not manage to get PT running with more than 2 replicas, in the NPT
ensemble.. For reminder:

I have tried running NPT-REMD with 3 replicas, at 300, 302 and 304 K. My .mdp
files are identical, with the exception of the target temperature. The log files
of the aborted simulation also indicate that:

Multi-checking the pressure coupling ... OK
Repl  There are 3 replicas:
Repl      0     1     2
Repl  T 300.0 302.0 304.0
Repl  p  1.00  1.00  1.00-------------------------------------------------------
Program mdrun_mpi, VERSION 3.3
Source code file: repl_ex.c, line: 142
Fatal error:
The reference pressure decreases with increasing temperature

Could that have something to do with the size of the system (38373 atoms)?

Thanks a lot for any help!!


Pascal Baillod (PhD student) 
Swiss Federal Institute of Technology EPFL	        Tel: +41-(0)21-693-0322
Institute of Chemical Sciences and Engineering ,	Fax: +41-(0)21-693-0320
Laboratory of Computational Chemistry and Biochemistry	pascal.baillod at epfl.ch
Room BCH 4121, Avenue Forel,	                        http://lcbcpc21.epfl.ch
CH-1015 Lausanne	

More information about the gromacs.org_gmx-users mailing list