[Pw_forum] problem with DFT+U

Hongsheng Liu lhs.happy2007 at gmail.com
Tue Feb 14 10:31:04 CET 2017


Dear Paolo,
    Thank you. The error always appears after some steps of scf before the
calculation of forces. I changed the machine. And the job can run on
another machine with the same input. So it is machine dependent.

Best regards,
Hongsheng

2017-02-14 10:03 GMT+01:00 Paolo Giannozzi <p.giannozzi at gmail.com>:

> If this happens during the calculation of forces, try the patch below.
> Otherwise it is impossible to say anything without an input and an
> output
>
> Paolo
>
> --- trunk/espresso/PW/src/force_hub.f90 2016/09/12 16:34:38 12959
> +++ trunk/espresso/PW/src/force_hub.f90 2017/02/07 14:22:26 13289
> @@ -466,11 +466,11 @@
>     !                           wfatbeta(iwf,ih)*dbetapsi(ih,ibnd)
>     !
>     IF ( mykey == 0 .AND. nh(nt) > 0 ) THEN
> -      CALL ZGEMM('N','N',nwfcU, nb_e-nb_s+1, nh(nt), 1.0_dp,  &
> -           wfatdbeta, nwfcU, betapsi(1,nb_s), nh(nt), 1.0_dp,&
> +      CALL ZGEMM('N','N',nwfcU, nb_e-nb_s+1, nh(nt), (1.0_dp,0.0_dp), &
> +           wfatdbeta, nwfcU, betapsi(1,nb_s), nh(nt),(1.0_dp,0.0_dp), &
>             dproj(1,nb_s), nwfcU)
> -      CALL ZGEMM('N','N',nwfcU,nb_e-nb_s+1, nh(nt), 1.0_dp,  &
> -           wfatbeta, nwfcU, dbetapsi(1,nb_s), nh(nt), 1.0_dp,&
> +      CALL ZGEMM('N','N',nwfcU,nb_e-nb_s+1, nh(nt), (1.0_dp,0.0_dp),  &
> +           wfatbeta, nwfcU, dbetapsi(1,nb_s), nh(nt),(1.0_dp,0.0_dp), &
>             dproj(1,nb_s), nwfcU)
>     END IF
>     ! end band parallelization - only dproj(1,nb_s:nb_e) are calculated
>
> Paolo
>
> On Thu, Feb 9, 2017 at 12:32 PM, Hongsheng Liu <lhs.happy2007 at gmail.com>
> wrote:
> > Dear Matteo,
> >   Thank you for your reply. I'm sorry for the unclarity of my problem. It
> > can work when lda_plus_u = .false.. I'm using PWSCF v.5.4.0. The error
> > always appears after some steps of scf (before convergency), before the
> > calculation of forces and stresses.
> >
> > Best regards,
> > Hongsheng
> >
> > 2017-02-09 9:54 GMT+01:00 Matteo Cococcioni <matteo at umn.edu>:
> >>
> >> Dear  Hongshen,
> >>
> >> does it work when lda_plus_u = .false.? from your email it is not clear
> >> whether the error is due to the +U part or not. also, are you using the
> >> latest version of the code or an older one? at what point of the
> calculation
> >> does the error appear? is it in the calculation of forces, stresses or
> >> before?
> >>
> >> Matteo
> >>
> >>
> >> On Wed, Feb 8, 2017 at 5:49 PM, Hongsheng Liu <lhs.happy2007 at gmail.com>
> >> wrote:
> >>>
> >>> Dear all,
> >>>   I'm trying to do a geometry optimization for bulk magnetite with
> PBE+U.
> >>> The job succeeded with primitive cell (including 14 atoms), but failed
> with
> >>> conventional cell (including 56 atoms). The error message is shown
> below,
> >>>   'Fatal error in PMPI_Bcast: Other MPI error, error stack:
> >>>   PMPI_Bcast(2434)........: MPI_Bcast(buf=0x998c5a0, count=3366,
> >>> MPI_DOUBLE_PRECISION,   root=4, comm=0x84000002) failed
> >>>   MPIR_Bcast_impl(1807)...:
> >>>   MPIR_Bcast(1835)........:'
> >>>
> >>> Similar question was asked before on pw_forum, but it seems no solution
> >>> till now.
> >>>  My input is listed below,
> >>>  &CONTROL
> >>>   calculation    = "vc-relax",
> >>> ......
> >>> /
> >>> &SYSTEM
> >>>   ibrav = 1
> >>>   A = 8.522631
> >>>   nat = 56,
> >>>   ntyp = 3,
> >>>   ecutwfc = 42.0,
> >>>   ecutrho = 340.0,
> >>>   nspin = 2,
> >>>   starting_magnetization(1) = 0,
> >>>   starting_magnetization(2) = -0.25,
> >>>   starting_magnetization(3) = 0.25,
> >>>   occupations = 'smearing',
> >>>   smearing = 'gaussian',
> >>>   degauss = 0.0008,
> >>>   lda_plus_u = .TRUE.
> >>>   lda_plus_u_kind = 0
> >>>   Hubbard_U(1) = 0,
> >>>   Hubbard_U(2) = 3,
> >>>   Hubbard_U(3) = 3,
> >>> /
> >>> .........
> >>> /
> >>> &CELL
> >>>   cell_dofree = 'volume'
> >>> /
> >>> ......
> >>> K_POINTS automatic
> >>> 6 6 6 0 0 0
> >>>
> >>> I appreciate any help!
> >>>
> >>> _______________________________________________
> >>> Pw_forum mailing list
> >>> Pw_forum at pwscf.org
> >>> http://pwscf.org/mailman/listinfo/pw_forum
> >>
> >>
> >>
> >> _______________________________________________
> >> Pw_forum mailing list
> >> Pw_forum at pwscf.org
> >> http://pwscf.org/mailman/listinfo/pw_forum
> >
> >
> >
> > _______________________________________________
> > Pw_forum mailing list
> > Pw_forum at pwscf.org
> > http://pwscf.org/mailman/listinfo/pw_forum
>
>
>
> --
> Paolo Giannozzi, Dip. Scienze Matematiche Informatiche e Fisiche,
> Univ. Udine, via delle Scienze 208, 33100 Udine, Italy
> Phone +39-0432-558216, fax +39-0432-558222
> _______________________________________________
> Pw_forum mailing list
> Pw_forum at pwscf.org
> http://pwscf.org/mailman/listinfo/pw_forum
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20170214/2eed4b3d/attachment.html>


More information about the users mailing list