The ABINIT Users Mailing List ( CLOSED )

["Matteo Giantomassi" <Matteo.Giantomassi@uclouvain.be>]

> Dear Matteo,
>  The log of the aborted work is attached,
> thanks again for your kind reply and help.
>

 The log you sent is obtained with 5.6.3 and it's
 pretty normal that the calculation is crashing.
 In 5.6.3 no check on the size of the FFT mesh was done,
 and the Fourier interpolation was not available.

 Directly from the source of rdqps.F90 (v5.6.3):

 ! === Read the QP density ===
 ! Here take care because the FFT grid must be the same

 Therefore, you cannot use different FFT meshes in v5.6.3
 unless you (change|hack) the code.

 I'm interested in the first log file you posted on the mailing list,
 the one that has been obtained with 5.7.3 where the FT interpolation
 is used.



> Best wishes,
>
> Hai-Ping
>
> On Fri, May 1, 2009 at 8:55 PM, Matteo Giantomassi <
> Matteo.Giantomassi@uclouvain.be> wrote:
>
>>
>> Dear Hai-Ping,
>>
>> I also need the log file to have a better idea of
>> what's going on.
>>
>> Regards
>> Matteo Giantomassi
>>
>>
>>
>> > Dear Matteo:
>> >
>> >  Thanks for quick reply.  Below is my settings for  scQP calculation.
>> >
>> > ###
>> >   ndtset 2
>> >   jdtset  7 8
>> > ## Common setttings for SCR and SIG calculations
>> >  # gwmem11 00
>> >  # gwpara  2
>> >   fftgw   11
>> >   awtr 1
>> >   gwcomp 1
>> >   gwencomp  0.80 Ha
>> >   ppmfrq    35.60 eV
>> >   gwcalctyp  28  # 28 Self-GW, 27 COHEX
>> > ##SCR
>> >   gwpara7   2
>> >   optdriver7  3
>> >   getkss7   2
>> >   getqps7   6
>> >   ecutwfn7  16.00 Ha
>> >   ecuteps7  10.00 Ha
>> >   nband7  96
>> > ##SIG
>> >   optdriver8  4
>> >   getkss8 2
>> >   getqps8 6
>> >   getscr8 3
>> >   ecutwfn8  20 Ha
>> >   ecutsigx8  20.0 Ha
>> >   nband8    96
>> >
>> > #Common settings for GW
>> >   rhoqpmix     0.5
>> >   nkptgw    13    #
>> >   kptgw
>> >
>> >
>> > I think i maybe used fftgw wrongly for scQP calculations, since  i
>> could
>> > succeed finishing scQP calculation without this setting.  But the test
>> is
>> > just going on due to relative large computational loads.
>> >
>> > Thanks again.
>> >
>> > Regards,
>> > Hai-Ping
>> > On Fri, May 1, 2009 at 7:34 PM, Matteo Giantomassi <
>> > Matteo.Giantomassi@uclouvain.be> wrote:
>> >
>> >>
>> >> Please provide input and log files.
>> >> The FFT mesh used for the GW calculation
>> >> differs form the mesh used to calculate the density
>> >> store in the QPS file.
>> >> This was causing a SIGFAULT in old versions.
>> >> Now we are performing a Fourier interpolation
>> >> to have the QP rho on the same mesh as that used in the
>> >> main code.
>> >> There's something wrong (likely not implemented)
>> >> in the way the FFT routine is called.
>> >> Are you using input variables related to the FFT parallelism?
>> >>
>> >> I've also coded a naive linear interpolation in rdqps.F90.
>> >> In case you might set "use_FFT_interpolation" to .FALSE.
>> >> in the code and recompile although the FFT approach is more
>> >> accurate.
>> >>
>> >> Regards
>> >> Matteo Giantomassi
>> >>
>> >> > Dear  Developers:
>> >> >    I carried out a  scQP calculation with abinit5.6.4/5.7.3.  But
>> the
>> >> > calculation cannot continue after first calculation.  I found  some
>> >> > messages
>> >> > was given in the log file:
>> >> > "
>> >> >   rdqps : FFT meshes differ, performing Fourier interpolation
>> >> > -P-0000
>> >> > -P-0000  sg_ctrig: BUG -
>> >> > -P-0000   number of factors ic=        12
>> >> > -P-0000   exceeds dimensioned mfac=   11
>> >> > -P-0000
>> >> > -P-0000
>> >> > -P-0000  leave_new : decision taken to exit ...
>> >> > -P-0000  leave_new : synchronization done...
>> >> > -P-0000  leave_new : exiting...
>> >> > "
>> >> >
>> >> > Therefore, would you please tell me  how to handle this error ?
>> >> >
>> >> >
>> >> > Regards,
>> >> > Hai-Ping
>> >> >
>> >> > --
>> >> > Hai-Ping Lan
>> >> > Department of Electronics ,
>> >> > Peking University , Bejing, 100871
>> >> > lanhaiping@gmail.com, hplan@pku.edu.cn
>> >> >
>> >>
>> >>
>> >>
>> >
>> >
>> > --
>> > Hai-Ping Lan
>> > Department of Electronics ,
>> > Peking University , Bejing, 100871
>> > lanhaiping@gmail.com, hplan@pku.edu.cn
>> >
>>
>>
>>
>
>
> --
> Hai-Ping Lan
> Department of Electronics ,
> Peking University , Bejing, 100871
> lanhaiping@gmail.com, hplan@pku.edu.cn
>