molecular dynamics in VASP

Message

whzhang · #1 Post by **whzhang** » Wed Feb 07, 2007 10:49 am

I want to perform a molecular dynamics calculaiton for one system . This system contains water molecules and TiO2(110) surface.

I use the input file as below:

SYSTEM = dipeptide

PREC = Low
EDIFF = 1E-04
ISPIN = 2
NELMDL = 4
NELMIN = 8
!BMIX = 2.0 ! mixing parameter
!MAXMIX = 50 ! keep dielectric function between ionic movements

ISYM = 0
VOSKOWN = 1
LREAL = A

ISMEAR = 0
SIGMA = 0.05

ISIF = 2

SMASS = 4.0
TEBEG = 300

POTIM = 0.17

IALGO = 48

NSW = 3000

IBRION = 0

LWAVE = .FALSE.
LCHARG = .FALSE.

I have tried to change the PARAMETER SMASS and POTIM, but temperature I get have a big oscillate magnitude.

I have some question:
1) the criterion of the MD quality?
2) How to choose the SMASS and POTIM..

Can anybody give some suggestion about this?

thanks a lot

dtrinkle · #2 Post by **dtrinkle** » Thu Feb 08, 2007 7:22 pm

What is a "big" oscillation in temperature? If your initial condition is not a previously thermalized MD run, there will always be some ringing in the temperature as your system equilibrates. Since your POTIM is very small (0.17 fs, as opposed to, say 1 or 2fs), this ringing will happen for many timesteps. Is there a reason you chose such a small POTIM? You can monitor the overall energy conservation during an MD run to ensure that your POTIM isn't too large, but 0.17 is rather small.

Remember too that the size of thermal oscillations--which will always be present--will go as 1/sqrt(number of atoms). So if you have a small system, expect "large" oscillations.

whzhang · #3 Post by **whzhang** » Mon Feb 12, 2007 2:02 pm

thanks for your reply. I change POTIM to 1.0fs, and the calculaiton has performed 1ps. The temperature osscilator is in the acceptable range. As for the energy conservation, which value in the OSZICAR file should be considered ?

Thanks a lot.

dtrinkle · #4 Post by **dtrinkle** » Tue Feb 13, 2007 11:36 pm

From http://cms.mpi.univie.ac.at/vasp/vasp/node55.html , E= is the total free energy, including the Nose thermostat and the ion kinetic energy. This is the conserved quantity that should not drift "too much" over the length of the run (something like meV/atom/ps).

whzhang · #5 Post by **whzhang** » Mon Mar 12, 2007 9:41 am

I found only double surface adsorption model can give correct results for my system. So I want to use double surface model for MD simulation. But it seems that the program adops no symmetry when perform MD, and the system lost it's symmetry... what can I do ?

thanks a lot.