RE: Problem with running FEP calculations

From: Eric Lang (eric.lang_at_pg.canterbury.ac.nz)
Date: Mon Apr 15 2013 - 20:21:53 CDT

Dear Chris,

Thanks for getting back to me.

I am sorry, I forgot to mention that I am trying to calculate binding free energies using the double annihilation procedure. Thus, if I am correct, I do not have to use Alchemify in this case (I got the message: “Either no atoms appearing, or no atoms disappearing. PSF file requires no modification.” if I try to use it anyway).

I also tried to modify the configuration file of the crown ether example using the parameters that can be found in the “vanilla” directory of your tutorial:

# LOOP OVER LAMBDA-STATES

alchEquilSteps 50

set nSteps 200

set dLambda 0.1

set init {0.0 0.0000001 0.000001 0.00001 0.0001 0.001 0.01 0.05 0.1}

set end {0.9 0.95 0.99 0.999 0.9999 0.99999 0.999999 0.9999999 1.0}

runFEPlist $init $nSteps

runFEP 0.1 0.9 $dLambda $nSteps

runFEPlist $end $nSteps

If I do this, it runs for some time and then stop suddenly (no error indicated in the log file) with a "*** glibc detected *** double free or corruption (!prev)" error in my error file.

If I try to increase the number of steps for alchEquilSteps and nSteps, I do not have this problem but at some point I got an error message like:

ERROR: Stray PME grid charges detected: 56 sending to 63 for planes 26 in my log file

Do you have any idea of what could be wrong?

Many thanks in advance,

Eric

________________________________
From: chipot [chipot_at_ks.uiuc.edu]
Sent: Monday, 15 April 2013 5:03 p.m.
To: Eric Lang
Cc: namd-l_at_ks.uiuc.edu
Subject: Re: namd-l: Problem with running FEP calculations

Hello,

did you "alchemify" explicitly your PSF, or do you let NAMD clean up the
file? In the latter case, I suggest you run "alchemify" separately, as the
symptom you describe looks like the two topologies are "talking" to each
other. Cheers,

Chris Chipot

Chris Chipot, Ph.D.
Beckman Institute. University of Illinois at Urbana-Champaign

sent from my iPad

On Apr 15, 2013, at 1:51 AM, Eric Lang <eric.lang_at_pg.canterbury.ac.nz<mailto:eric.lang_at_pg.canterbury.ac.nz>> wrote:

Hello,

I would like to run some alchemical Free Energy Perturbation calculations on a Blue Gene P system. Although my calculations work well on my PC and on a Power 7 cluster they do not work on the Blue Gene system. (NAMD 2.9 used on all platforms)

I tried different simulations, including the crown ether example from the FEP tutorial, and I always have the same kind of errors with the Blue Gene P:

I got messages like "Constraint failure in RATTLE algorithm for atom ... simulation has become unstable" at the very beginning of the simulation. If I deactivate the RATTLE algorithm, it works for the first window, but at the beginning of the second window I got the messages “Atoms moving too fast”. (please note that all the simulated systems are well equilibrated)

I therefore strongly suspect the build of NAMD on the Blue Gene system I am using, but I do no know what can be wrong (classical simulations work well). Does anyone have an idea of what could be the problem? Are there any important thing related to FEP to take care of when building NAMD form source?

Many thanks in advance for your help,

Eric

Eric Lang

Department of Chemistry

University of Canterbury

Christchurch – New Zealand

This archive was generated by hypermail 2.1.6 : Wed Dec 31 2014 - 23:21:07 CST