[gmx-users] some remarks about REMD and analysis

[Luca Mollica]

Dear all,

I would like to write down here some remarks about my Replica Exchange 
Molecular Dynamics simulation that finally thanks to all the advices has 
been successfully (partially performed).

1. The real problem of my simulation (described in the following post: 
http://www.gromacs.org/pipermail/gmx-users/2005-October/017715.html ) was 
inded the temperature distribution and the difference of temperature I had 
between replicas. therefore, I have changed completely my mind and used 
this T setup:

280.0 281.0 282.1 283.2 284.3 285.6 286.8 288.2 289.6 291.0 292.5 294.1 
295.7 297.3 299.1 300.8 302.7 304.6 306.5 308.5 310.6 312.7 314.8 317.1 
319.3 321.7 324.1 326.5 329.0 331.6 334.2 336.8 339.6 342.3 345.2 348.1 
351.0 354.0 357.1 360.2 363.3 366.6 369.8 373.2 376.6 380.0

reducing the global time interval and also the difference of temperature 
between the single replicas according to Berk Hess advice.
Now the REMD is really ... exchanging replicas ! And everything sems to be 
fine and it works really well. Many thanx to all who gave me advices.

2. A question/ suggestion: REMD implementation is only set up for single 
CPU .... but what about giving the possibility of distributing a replica 
on many CPUs instead of using one processor only ? Is it a choice or are 
there any problems of implementationof such a feature ??

Bye

LM


 
-----------------------------------------------------------------------------

Luca Mollica
ZRL-IBM Research Center
Saumerstrasse 4
8803, Rueschlikon (Zurich)
CH

Tel: 0041 - 447248217

email:        YLU at zurich.ibm.com 
email(alt): ZRLYLU at ch.ibm.com 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20051028/fc53d272/attachment.html>