infrequent rattle failure with charmm36

From: Edward Lyman (edward.lyman_at_gmail.com)
Date: Wed Apr 20 2011 - 08:47:59 CDT

Hi all,

I am simulating a bilayer with the charmm36 parameters, which was
extensively minimized, heated, and relaxed. I find the simulation always
crashes due to a rattle failure, typically between 4M and 8M timesteps.
Nothing special in the cfg file except for the use of force-based switching,
as required by the charmm36 parameterization. (details of the cfg below) I
am wondering if anyone else has seen this issue?

The only other thing that is out of the ordinary is that I am running the
simulation on 176 procs, since that is what I have available. I plan to test
and see if it is stable on a more normal partition, like 128, but a careful
reading of the fft cfg section suggests 176 should be ok in NAMD.

Any insights appreciated. The error and cfg below.

Thx,
Ed

log file end:
WRITING EXTENDED SYSTEM TO RESTART FILE AT STEP 20051000
WRITING COORDINATES TO DCD FILE AT STEP 20051000
WRITING COORDINATES TO RESTART FILE AT STEP 20051000
FINISHED WRITING RESTART COORDINATES
ERROR: Constraint failure in RATTLE algorithm for atom 15602!
ERROR: Constraint failure; simulation has become unstable.
ERROR: Exiting prematurely; see error messages above.
====================================================

WallClock: 68395.648438 CPUTime: 66929.414062 Memory: 244.082031 MB

The relevant cfg sections:

#******************************run parameters*****************************
# number of steps set in FEP section
firsttimestep 13800000
numsteps 103800000
timestep 2.0
#temperature 310K #comment out when
restarting

#*****************************pressure control*****************************
useGroupPressure yes
useFlexibleCell yes
useConstantRatio yes
#useConstantArea no
LangevinPiston on
LangevinPistonTarget 1.01325
LangevinPistonPeriod 500
LangevinPistonDecay 100
LangevinPistonTemp 310

#******************************electrostatics******************************
fullElectFrequency 1
exclude scaled1-4
1-4scaling 1.0
switching on
vdwForceSwitching on
switchdist 10.0
cutoff 12.0
pairlistdist 13.5

#**********************************bonds***********************************

rigidBonds all
rigidTolerance 1.0e-5
#
#*******************************PME options********************************
PME yes
PMEPencils 1

# New Style
PMETolerance 1.0e-6
PMEInterpOrder 6
PMEGridSpacing 1.0

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 05:23:51 CST