Re: [ESPResSo-users] how to run NPT simulation

[Axel Arnold]

On Friday 11 February 2011 09:24:47 Ulf Schiller wrote:
> Yonglei Wang wrote:
> > Hej
> > 
> > I noticed that we can simulate NPT ensemble with this package but I
> > checked all scripts and found there is only one npt.tcl file which is
> > used to validate this package during installation. So I want to know how
> > can I implement the NPT simulation in detail? I mean, if we generate all
> > particle coordinates with very low number density, then run the NPT
> > simulation to shrink the simulation box. How should I do fulfill this
> > simulation?
> 
> Yonglei,
> 
> the following reference describes the NPT algorithm in more detail:
> 
> J. Chem. Phys. 111, 4453 (1999); doi:10.1063/1.479208
> 
> Note in particular the discussion on oscillating behavior. I believe
> there is no general recipe how to setup an NPT simulation. You will have
> to tune the parameters for your specific system.

Hi,

since we are often interested in NVT simulations, we use normally other tricks 
to setup our systems, which I guess will also be faster when using NPT. 
Starting from a rather large box, it might take rather long for NPT to 
equilibrate to the correct box size, therefore, starting at higher densities 
might be faster.

To start a high densities reliably, the short ranged interactions can be 
capped, that is, their singularity is replaced by a low barrier, which 
gradually can be risen. This barrier is set by inter ljforcecap for the 
Lennard-Jones potential. See for example the script samples/lj_liquid.tcl.

Regards,
Axel

-- 
JP Dr. Axel Arnold
ICP, Universität Stuttgart
Pfaffenwaldring 27
70569 Stuttgart, Germany
Email: address@hidden
Tel: +49 711 685 67609