The ABINIT Users Mailing List ( CLOSED )

[Nuno Galamba <ngalamba@cii.fc.ul.pt>]

On Thursday 02 March 2006 12:20, Per Andersson wrote:
> vinayak@ipr.res.in wrote:
> > Dear Abinit users.
> >
> >   I'm trying to do MD simulation of the compressed Hydrogen fluid with
> > the time step of 12atu to calculate the energy . I found that after 100
> > MD steps particles are away from the simulation box. Is the simulation
> > running properly?? Your reply will be highly appreciated.
> >
> > Thanks & Regards
> > Vinayak Mishra,
> > Institute for Plasma research,
> > Bhat, Gandhinagar-382428, India.
>
> I am doing similar calculations with 128 atoms. Your
> timestep seems rather large, I had to go as low as 3 a.u. to get
> good convergence for cryogenic temperatures (and as low as 0.5
> for higher temperatures). Please take a look at your results.
>
> The atoms leaving the box will be "folded" back as the box is
> a part of a repeated lattice.
>
> /Per Andersson

I do not know much about hydrogen simulations, but my first comment would be 
that the time-steps are too small, and not to large. I used ABINIT molecular 
dynamics algorithm for NVE simulations for a while and had some problems. So 
perhaps I am not the best person to answer you. I do hope it helps though.

My comment to the first e-mail is that, ABINIT only prints out positions NOT 
subjected to periodic boundary conditions (PBC), i.e., the forces must be 
calculated based on some internal coordinates subjected to PBC but those 
printed out are not. What you can do is to fold them back by writting a small 
code that applies PBC to the sets of coordinates ABINIT gives you in the 
output file.

Regarding the second e-mail I think you should check on energy conservation 
and linear momentum conservation if you are doing NVE. I observed for some 
simple simulations that the linear momentum was not beeing conserved properly 
and I could not figure out why. I decided to stop using it at that time and 
to follow a different approach and did not pursue this problem further but I 
would like to understand it. Perhaps you may have had a similar experience.


-- 
Nuno Galamba
Grupo de Fisica-Matematica da Universidade Lisboa
Lisboa
Portugal