[Q-e-developers] [Pw_forum] FFT on a small mesh

[Paolo Giannozzi]

On Sep 22, 2011, at 2:13 , dario rocca wrote:

> Dear Developers

(reminder: the proper place where to contact developers is the q-e- 
developers
mailing list on qe-forge.org)

> I would be interested in calculating the FFT with a small  
> reciprocal space cutoff.
> Namely beyond the FFT for wavefunctions and density I would like to  
> use an
> additional FFT on an "object" that requires a cutoff smaller than  
> the wavefunctions
> (but of course the real space mesh would still be the same). All  
> the standard
> FFT machinery seems very specialized to the wavefunction/density case

it is, by construction. In the most recent versions, however, the FFT  
stuff is
encapsulated into a few "descriptors" that are initialized in a few  
specific
routines. Note that the CP code has a third grid, the "small box"  
grid, a
subset of the dense grid in real space.This is used to deal with  
augmentation
charges in the USPP framework. You may want to have a look at it.

I have heard of work done to speed up exact exchange and GW  
calculations,
consisting in performing the needed FFT with a lower cutoff. Nothing  
has yet
percolated to the current development version, though.

P.
---
Paolo Giannozzi, Dept of Chemistry&Physics&Environment,
Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222