[Wannier] pw2wannier90.x terminates after Amn calculation

[Varadharajan Srinivasan]

Dear all,

This is in continuation with my request (trail mail below). My attempts to
run pw2wannier90 on a double perovskite are ending up in a crash at the
computation of M matrix. I did a little bit of debugging and found out that
the crash is originating from the following part of the subroutine
compute_mmn :

1139       DO nt = 1, ntyp
1140          IF (upf(nt)%tvanp ) THEN
1141             DO ih = 1, nh (nt)
1142                DO jh = 1, nh (nt)
1143                   CALL qvan2 (nbt, ih, jh, nt, qg, qgm, ylm)
1144                   qb (ih, jh, nt, 1:nbt) = omega * qgm(1:nbt)
1145                ENDDO
1146             ENDDO
1147          ENDIF
1148       ENDDO

I put some write statements at various points in the code to track the
crash and it seems that after looping a few times over ih and jh the code
crashes (without any clear error message).

I am using a mixture of kjpaw and rrkjus pseudo potentials in my system.
Could this be the source of the problem? I read in an earlier post that
USPP and PAW along with non-collinear spin or spinors is not implemented.
My calculation uses nspin=2 and should not have this issue hopefully.

Would really appreciate any help to track down the error. I am using QE-5.1
for my DFT calculations. The MLWF codes along with QE-5.1 seem to working
for other test cases (BaTiO3, PbTiO3 and I also tried BaFeO3 as a
spin-polarised case).

Best,
Vardha.

On Sun, Mar 27, 2016 at 11:38 PM, Varadharajan Srinivasan <
varadharajan.srinivasan at gmail.com> wrote:

> Dear all,
>
> I am facing a very frustrating problem whose origin has been eluding me. I
> am trying to compute MLWF for a double perovskite system which is a
> ferrimagnetic insulator. I start by doing an nscf with 9x9x9 k-mesh, then a
> wannier90,x -pp to generate the nntp file and then a pw2wannier90,x. Here
> it computes the Amn matrix and then starts the computation of the M matrix
> but crashes without any error message. I've tried it several times on 64 to
> 128 processors (with upto 512 GB memory available in the nodes) and get the
> same problem. I even tried reducing the k-mesh to 8x8x8 but in vain. I
> attach the win and pw2wan input files below (for the 8x8x8 k-mesh).
>
> I would be very grateful if someone can help me figure out what is going
> wrong here. I am using version 2.0.0 of wannier90 along with qe-5.1 (an
> older version but currently the more functional one on our cluster).
>
> Thanks,
> Vardha.
> IISER Bhopal, India.
>
> input file for wannier90.x
> ---------------------------------
> num_bands         =   114
> num_wann          =   114
>
> #bands_plot        = true
> #bands_num_points  = 300
> #bands_plot_project : 50-59
>
> #dis_win_max       = 14.0d0
> #dis_froz_max      = 11.0d0
> #dis_num_iter      =  1000
>
> num_iter          = 1000
> mp_grid           = 8 8 8
>
> #dis_mix_ratio = 1.0
> translate_home_cell = .true.
> write_xyz = .true.
>
> begin atoms_frac
> Ca       0.984432491   0.048477445   0.251191257
> Ca      -0.484432582   0.548477537   0.248808687
> Zn      -1.030301130   0.088960360  -0.255530584
> Zn       0.530301078   0.588959551   0.755530723
> Fe1      0.493555846   0.019726928  -0.001347745
> Os2     -0.001156021   0.494716020   0.001621897
> Os3      0.501155940  -0.005284076   0.498378116
> Fe4      0.006444193   0.519726859   0.501346598
> O        0.099365991   0.461218172   0.245396389
> O        0.400633937   0.961218157   0.254603761
> O       -0.122487446  -0.515716414  -0.245277892
> O        0.622487602  -0.015716357   0.745278021
> O        0.710098763   0.285077220   0.055774606
> O       -0.210098933   0.785077075   0.444225230
> O       -0.693931902  -0.334335687  -0.049331190
> O        0.193932221   0.165664570   0.549331614
> O        0.175100402   0.178285762   0.950125464
> O        0.324899469   0.678285872  -0.450124955
> O       -0.196387210  -0.226409454  -0.945282994
> O        0.696387291   0.273590460   0.445282996
> end atoms_frac
>
> begin projections
> random
> end projections
>
> begin unit_cell_cart
> bohr
> 10.19148602 0.0 0.0
> 0.0 10.4093715322 0.0
> 0.0 0.0 14.54694742342
> end unit_cell_cart
>
> begin kpoints
> <....512 points here generated by kmesh.pl ...>
> end kpoints
>
>
> pw2wannier90 input file
> -------------------------------
>
> &inputpp
>    outdir = '/scratch/vardha/czfos/'
>    prefix = 'CFOS'
>    seedname = 'czfos_up'
>    spin_component = 'up'
>    write_mmn = .true.
>    write_amn = .true.
>    write_unk = .false.
> /
>
> pw2wannier90 output (last 20 lines)
> ---------------------------------------
>
>      501
>      502
>      503
>      504
>      505
>      506
>      507
>      508
>      509
>      510
>      511
>      512
>
>   AMN calculated
>
>   ---------------
>   *** Compute  M
>   ---------------
>
> APPLICATION TERMINATED WITH THE EXIT STRING: Killed (signal 9)
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/wannier/attachments/20160404/8a5fe17f/attachment.html>