The ABINIT Users Mailing List ( CLOSED )

["Anglade Pierre-Matthieu" <anglade@gmail.com>]

Dear Yusia, dear GW developpers

5.4.4 is a modern enough version of abinit.
Yet it looks like 15gw/sizefft.F90 has kept a local copy of Abinit fftdata with its old limitation. 
It would be nice to improve on that but I fear this data has  more constraints than usual fftdata. That's the reasons why they have not been touched when upgrading Abinits fftdata.

Do GW developpers have some clues about that ? Can I/we simply update 15gw/sizefft.F90 with an include of defs/defs_fftdata.F90 ? 

Best

PMA

On Jan 29, 2008 4:12 PM, Yusia Rosee <yrosee@gmail.com> wrote:

Hello, Dear Anglade,

Thank you! The version I am using is 5.4.4.  I think this is a recent
version already, right?

Yusia

Anglade Pierre-Matthieu wrote:
> Hi,
>
> Wich version of Abinit are you using ? Recent versions can stand much
> larger FFT grids (up to ~65000). May be upgrading to a newer version
> of Abinit would help ?
>
> Best
>
> PMA
>
> On Jan 28, 2008 7:59 PM, Yusia Rosee <yrosee@gmail.com

> <mailto:yrosee@gmail.com>> wrote:
>
>     Dear All,
>
>     When I do the screening calculation in the GW calculation without
>     using
>     the plasmon pole model, I get the following error:
>
>     ......
>     timing point number   20
>     -cpu time  =       0.02 seconds
>     -real time =       0.06 seconds
>
>     setmesh: npwwfn =             6043 ; Max (m1,m2,m3) =     11    11
>        11
>     setmesh: npweps/npwsigx =     2801 ; Max (mm1,mm2,mm3) =     10
>      10    10
>             addshell =    2
>     setmesh: using method 1 with aliasing_factor =   1
>
>     setmesh : finding a FFT mesh compatible with all the symmetries
>
>     setmesh : divisor mesh119119119
>
>     sizefft : ERROR-
>     nbest =     2049
>     is larger than any allowable FFT
>
>     leave_new : decision taken to exit ...
>
>
>     It seems related to the FFT grid. But when I decrease the cutoff
>     energies, I always get the same error.  So, how to solve this problem,
>     please?
>
>     Thank you very much!
>
>     Yusia
>
>
>
>
> --
> Pierre-Matthieu Anglade

-- 
Pierre-Matthieu Anglade