The ABINIT Users Mailing List ( CLOSED )

[BOTTIN Francois <francois.bottin@cea.fr>]

Hi,
David is right, it seems that the band/FFT parallelism is switched on.
But I thought that MPI_IO can solve the problem of WF in cut3d.
If you use MPI_IO capabilities (compilation with enable_mpi_io="yes" and 
execution with accesswff=1), you should be able:
(i) to write the WF file in parallel as it is performed in sequential
(ii) then execute cut3d in sequential and read the WF file without any 
trouble.
Regards,
Francois

David Waroquiers a écrit :
> Hello,
>
> I guess you are using band/FFT parallelism and that you are using 32
> processors with npband 32 or something like that ? (next time provide
> your input file, it's easier to see where the problem might be).
> In this way of parallelization, one has to use the MPI_IO capabilities
> of ABINIT to read the wavefunctions. The problem is that it has been
> implemented in ABINIT but not (yet ?) in cut3d.
> At the moment, one way to avoid this problem is to run in sequential.
> Maybe you can try to converge your wavefunction with bands/FFT
> parallelisation and then run again a calculation in sequential by just
> reading the wavefunction and writing it (just put nstep 1). Maybe
> someone could confirm that ?
>
> Hope it helps you ..
>
> On Wed, 2009-10-14 at 16:51 +0200, Philip Murphy wrote:
>   
> > Dear abinit developers,
> >
> > While trying to process a wavefunction file with the latest version of 
> > abinit's
> > cut3d, I get the following message:
> >
> > rwwf : BUG -
> >  Reading option of rwwf. One should have npw=npw1
> >  However, npw= 3097, and npw1=99133.
> >  Action : contact ABINIT group.
> >
> >
> > There is nothing out of the ordinary in the way I generated the WF file: a 
> > self
> > consistent calculation of a supercell with only kpt 0.0 0.0 0.0. I have tried
> > with and without time reversal symmetry, to no avail.
> >
> > It seems an odd error, since I am not performing an RF calculation at all, and
> > npw1 shouldn't be of any consequence.
> >
> > Please help.
> >
> > Kind regards,
> > Philip Murphy
> >     


--
##############################################################
Francois Bottin                    tel: 01 69 26 41 73
CEA/DIF                            fax: 01 69 26 70 77
BP 12 Bruyeres-le-Chatel         email: Francois.Bottin@cea.fr
##############################################################