AW: replica exchange module

From: Norman Geist (norman.geist_at_uni-greifswald.de)
Date: Tue Oct 27 2015 - 03:08:42 CDT

Basically by modifying replica.namd, you can treat each replica individually, as it can be identified by replica_id. You just need to understand that replica.namd plays in jobscript context, whereas tclforces plays in tclforces context. It's not the same TCL context. But you could give each replica a different tclforces configuration, which I hope can be changed between "run" commands.

Norman Geist

-----Ursprüngliche Nachricht-----
Von: owner-namd-l_at_ks.uiuc.edu [mailto:owner-namd-l_at_ks.uiuc.edu] Im Auftrag von Jessen Lucas
Gesendet: Dienstag, 27. Oktober 2015 04:49
An: Norman Geist <norman.geist_at_uni-greifswald.de>
Cc: namd-l_at_ks.uiuc.edu
Betreff: Re: namd-l: replica exchange module

Hello, Dear Norman,

I am sorry to disturb you again, if I want to add different force to
different replicas, then should I do like the following,

add the following codes (tclforces on, tclforcescript tclforce.tcl) in
step 5 you mentioned above. how about colvar bias scriptfunction in each
replica if the scriptfunction in replica 1 is different from replica 2?

actually, what I want to do is to write code in each replica, so all this
could be done in step 5?

Thanks a lot.

J. Lucas

2015-10-26 16:21 GMT+08:00 Norman Geist <norman.geist_at_uni-greifswald.de>:

> Replica.namd will do everything basically.
>
>
>
> Workflow:
>
>
>
> 1. Use make_output_dirs to create the replicas individual output
> folders
>
> 2. Create a general jobscript without some like run/numsteps or
> temperature that are set by replica.namd later when sourced.
>
> 3. Replica.namd itself needs some options to be set like
> stepsperrun, runsperframe etc.. Therefore you create a remd.conf which
> replica.namf will source
>
> 4. You start namd with +replicas option which which will run that
> number of individual namd simulations (number of nodes/cores/gpus should be
> devisible by number of replicas, you need at least two cores per replica)=

This archive was generated by hypermail 2.1.6 : Thu Dec 31 2015 - 23:22:11 CST