Re: AMBER: Temperature Regulation

From: David A. Case <case.scripps.edu>
Date: Tue, 27 Jul 2004 08:24:31 -0700

On Mon, Jul 26, 2004, Johnson Agbo wrote:

> We are running periodic boundary condition dynamics on 432 ice system
> in a rectangular box of size 19.074, 19.074, 38.148. we are interested
> in a constant temperature and constant pressure simulation. In order
> to get a higher density of the ice to about 1.3 g/cm3, we set the
> pressure to 10000 bars, the density came out as we expected but from
> the output, we noticed that the temperature is far from the temperature
> we want (temp0=50.0). Below are our input and output files.
>
>
> molecular dynamics run
> &cntrl
> imin=0, irest=1, ntx=5,
> ntt=1, temp0=50.0, tautp=2.,
> ntp=1, taup=1., ntave=1000,
> ntb=2, ntc=2, ntf=2,pres0=10000,
> nstlim=500000,
> ntwe=1000, ntwx=1000, ntpr=1000,
> &end
>
>
> A V E R A G E S O V E R 500000 S T E P S
>
>
> NSTEP = 500000 TIME(PS) = 600.000 TEMP(K) = 80.40 PRESS = 9983.6
> Etot = -4942.7336 EKtot = 206.8291 EPtot = -5149.5627
> BOND = 0.0000 ANGLE = 0.0000 DIHED = 0.0000
> 1-4 NB = 0.0000 1-4 EEL = 0.0000 VDWAALS = 1168.6964
> EELEC = -6318.2591 EHBOND = 0.0000 RESTRAINT = 0.0000
> EKCMT = 101.6708 VIRIAL = -2001.3074 VOLUME = 9759.0138
> Density = 1.3248
> Ewald error estimate: 0.2385E-02
>

Look at the subaverages to see if things are still drifting (in either
temperature or pressure). It is a concern that your Ewald error estimate is
so high. My guess is that you may need a small timestep (e.g. 0.0005), given
the density and temperature you are using. You should get better temperature
regulation by using ntt=3 rather than 1, but you should aim to being able
to do Newtonian dynamics following equilibration (ntt=0). Try setting ntave
to 10000 or 50000; that will give you more useful information.

You probably should run some test calculations at ordinary pressures and
see if you are getting expected behavior, before moving to higher pressures.
And you will probably need to look for advice to the physics literature for
suggestions about doing these sorts of simulations -- they are way outside
the realm of experience for most of us.

....dac
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber.scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo.scripps.edu
Received on Tue Jul 27 2004 - 16:53:00 PDT
Custom Search