Re: Problem in energy conservation in NVE ensemble with 1fs timestep

From: Peter Freddolino (petefred_at_ks.uiuc.edu)
Date: Thu Aug 14 2008 - 11:40:32 CDT

Hi,
you're doing two things that can adversely affect energy conservation:
-using a very large rigidtolerance (see
http://www.charmm.org/ubbthreads/showflat.php?Cat=0&Number=472&page=91&fpart=2&vc=1)
-not using switching (I know this is the standard for amber family FFs,
but thought I'd still point it out)

You probably can't do much about the switching, since you're being
faithful to your forcefield, but you should certainly try the default
namd rigidtolerance (1e-8) and see if that improves things.

Best,
Peter

tinni sona wrote:
> Hi All,
>
> Please ignore my previous mail, the energy value is not given
> correctly there. Sorry for the inconvenience. Please look after the
> following one.
>
> I'm doing a NVE simulation in NAMD using AMBER force field. I used
> input coordinate and velocity files from a NVT simulation which was
> equilibrated at desired temperature (300 K). Looking at TOTAL it is
> seen that my system's energy underwent a slow linear
> drift (from -87435.3852 to -88170.7969 in 200 ps ). The TEMP of the
> system is also fluctuating in the range 295K to 304K.
>
> The .conf file I used is given below:
>
> # NAMD configuration file for CD44_HA (NVE) EQUILIBRATION
> #Initialization
>
> firsttimestep 6100000
>
> # input file
> set input CD44_HA_equilibration30
> coordinates CD44_HA_minimization_restart_f.pdb
> bincoordinates $input.restart.coor
> binvelocities $input.restart.vel
>
> # output file
> set output CD44_HA_equilibration31
> outputName $output
> restartName $output.restart
> binaryoutput yes
> binaryrestart yes
>
> #PBC
> cellBasisVector1 70.263 0.0 0.0
> cellBasisVector2 0.0 70.263 0.0
> cellBasisVector3 0.0 0.0 70.263
> wrapWater on
> wrapAll on
>
> # force field
> amber on
> parmfile CD44_HA_ff99SB.prmtop
> exclude scaled1-4
> 1-4scaling 0.833333
>
> # approximations
> switching off
> cutoff 12
> pairlistdist 13.5
> margin 0
>
> #integrator
> timestep 1.0
> rigidBonds all
> rigidTolerance 0.0005
> nonbondedFreq 1
> fullElectFrequency 1
> stepspercycle 10
>
> #PME (for full-system periodic electrostatics)
> PME yes
> PMEGridSizeX 72
> PMEGridSizeY 72
> PMEGridSizeZ 72
>
> #output
> restartfreq 1000 ;# 500steps = every 1ps
> dcdfreq 500
> xstFreq 1000
> outputEnergies 1000
> outputPressure 1000
>
> langevinPiston off
>
> run 200000
>
> I will highly be oblised if someone can give me an advice about how to
> get my simulations conservative. Please help.
>
> Thanks,
> Madhurima
>

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:48:11 CST