The ABINIT Users Mailing List ( CLOSED )

[Vincenzo Fiorentini <vincenzo.fiorentini@dsf.unica.it>]

My personal view: VASP is very efficient and well parallelized, and  
has well-constructed and tested PAW and ultrasoft potentials, and is  
probably on of the top codes available for total energy calculations.  
It does not, however, do linear response (so far: version 5.0 will,  
it appears), which is of course the strong point of abinit (of course  
well-tuned PAW is also at a premium over norm-conserving: abinit will  
be much better off once PAW is up and running, especially in linear  
response). That aside, I am pretty sure that version matching is  
entirely accidental.

If you are after medium (~100 atom) LDA/GGA structure/dynamics  
calculations, especially with nasty atoms (magnetic stuff, etc.) VASP  
is probably your best bet. For larger systems, SIESTA may be a  
solution (and besides it has been interfaced with a transport code by  
Sanvito and colleagues), but localized orbitals are appreciably more  
complicated to handle. If you are interested in linear response  
calculations of course abinit is your choice. A serious alternative  
(also open source) you may try out is Espresso (pwscf.org: the SISSA,  
Trieste code) which uses ultrasoft potentials and does total energy/ 
dynamics as well as linear response. There is also the transport code  
Want (www.wannier-transport.org, by Marzari, Nardelli and others)  
interfaced to Espresso.

Philosophical issues aside, VASP is not free of charge. I don't know  
how much they charge these days. Personally, the equivalent of a  
couple of thousand euro I payed years ago have been well spent.

best all,
        Vincenzo



On 11 May 2006, at 7:23 PM, Nichols A. Romero wrote:

> I you are dealing with very large unit cells and are computer
> resources are at a premium. I will recommend SIESTA
> http://www.uam.es/departamentos/ciencias/fismateriac/siesta/
>
> Is not as straight forward to use as VASP? But it has a mailing list
> like ABINIT and people who are willing to help.
>
> Bests,
> On 5/11/06, spamrefuse@yahoo.com <spamrefuse@yahoo.com> wrote:
> > Hello,
> >
> > I have joined an academic group that is interested
> > in simulating properties of carbon nanotubes. To
> > that purpose, the 'boss' here has purchased VASP
> > (the Vienna Abinitio Simulation Package).
> >
> > Meanwhile I myself have discovered ABINIT. Although
> > I am supposed to work with VASP, as a fan of
> > Opensource I would like to explore ABINIT as well.
> >
> > I'm still in an early stage of exploring, but to me
> > it looks as if the two packages are quite similar.
> > Is that so?
> >
> > Is it more than an coincidence that even the version
> > numbering is the same (also VASP's latest stable
> > release is 4.6 and 5.X is 'under construction').
> >
> > Apart from the finances (VASP is not free), could
> > someone point out what are the key differences
> > between the two packages?
> >
> > If there's a difference, which package is most
> > suitable for simulating carbon nanotubes related
> > topics?
> >
> > Thank you,
> > Rob.
>
>
> --
> Nichols A. Romero, Ph.D.
> 1613 Denise Dr. Apt. D
> Forest Hill, MD 21050
> 443-567-8328 (C)
> 410-306-0709 (O)