[Pw_forum] Balance the efficiency and veracity of PWscf vs. other DFT codes.

Nicola Marzari marzari at MIT.EDU
Sun Jul 6 07:57:36 CEST 2008


> On Sat, 5 Jul 2008, zhaohscas wrote:
> 

> HZ> When the system under study is large, i.e., including hundreds of 
> HZ> atoms or more, in this case, form the point of balancing the 
> HZ> efficiency and veracity, whether the PWscf is suitable for such 
> HZ> computation or not?  Furthermore, in this case, which will be more 
> HZ> promising DFT code(s)?



Besides PWscf, you might consider CP, as well - this is also part of
the QE distribution, and, albeit more finnicky to use, has been built
and optimized for large calculations - it can only do Gamma sampling
(hence it can impose to the k-s orbitals to be real, halving the
memory cost).

Traditionally, it would do static calculations (and ionic relaxations)
using damped dynamics on electrons (or electrons and ions); in this 
mode, it is very efficient and worth considering for large scale 
calculations. Recently, a conjugate-gradient strategy on electrons has 
been added, as  the possibility of doing metals (using ensemble-dft). 
Neither of these are extensively tested, but are being polished off - 
for ensemble-dft there is one last bug that used to affect parallel 
calculations, that has just been found (and not yet corrected, even in 
the CVS).

				nicola


---------------------------------------------------------------------
Prof Nicola Marzari   Department of Materials Science and Engineering
13-5066   MIT   77 Massachusetts Avenue   Cambridge MA 02139-4307 USA
tel 617.4522758 fax 2586534 marzari at mit.edu http://quasiamore.mit.edu


More information about the Pw_forum mailing list