[Pw_forum] Bug in a calculation of PDOS in a non-colinear system (igk should be igk_k ?)

[MitsuakiKawamura]

Dear all QE users and developers

Hello,

When I was performing the calculation of PDOS by using projwfc.x code, I
found a problem. The calculated system is bcc-Fe with non-colinear
magnetization. Input files for pw.x and projwfc.x are as follows:
$ cat scf.in
&control
  calculation = 'scf'
   pseudo_dir = './fe.save',
       outdir = './',
       prefix = 'fe'
wf_collect = .true.
/
&system
        ibrav = 3,
    celldm(1) = 5.217,
          nat = 1,
         ntyp = 1,
      ecutwfc = 25.0,
      ecutrho = 200.0,
       report = 1,
     noncolin = .true.
  starting_magnetization(1) = 0.5
     angle1(1) = 90.0
     angle2(1) =  0.0
  occupations = 'smearing',
smearing = "m-p"
degauss = 0.05,
/
&electrons
  conv_thr = 1.0e-8
  mixing_beta = 0.2
/
ATOMIC_SPECIES
  Fe 55.847 Fe.pz-nd-rrkjus.UPF
ATOMIC_POSITIONS
  Fe 0.0 0.0 0.0
K_POINTS automatic
  8 8 8 0 0 0

$ cat proj.in
&PROJWFC
prefix = "fe",
outdir ="./",
emin = 14.0,
emax = 15.0
deltae = 0.2
/

And I ran pw.x and proj.x compiled with the ifort15.0.3 + openmpi1.8.8  
as follows:
mpiexec -np 2 ~/programs/QE/espresso-5.4.0/bin/pw.x -npool 1 -in scf.in
mpiexec -np 2 ~/programs/QE/espresso-5.4.0/bin/projwfc.x -npool 1 -in  
proj.in

Although pw.x finished normally, projwfc.x stoped with the following error
  SIGSEGV、segmentation fault occurred
Image              PC                Routine            Line  
Source
projwfc.x          00000000009E8591  Unknown               Unknown  Unknown
projwfc.x          00000000009E6CE7  Unknown               Unknown  Unknown
libmpi_usempif08.  00002B7532D97952  Unknown               Unknown  Unknown
libmpi_usempif08.  00002B7532D977A6  Unknown               Unknown  Unknown
libmpi_usempif08.  00002B7532D85FFC  Unknown               Unknown  Unknown
libmpi_usempif08.  00002B7532D67198  Unknown               Unknown  Unknown
libpthread.so.0    00002B7533AE0340  Unknown               Unknown  Unknown
projwfc.x          0000000000435B2E  atomic_wfc_nc_pro         111  
atomic_wfc_nc_proj.f90
projwfc.x          000000000040D483  projwave_nc_             1000  
projwfc.f90
projwfc.x          0000000000407B9B  MAIN__                    182  
projwfc.f90
projwfc.x          000000000040731E  Unknown               Unknown  Unknown
libc.so.6          00002B7533D0FEC5  Unknown               Unknown  Unknown
projwfc.x          0000000000407229  Unknown               Unknown  Unknown

I also ran pw.x and projwfc.x with single process; this finished  
normally but yield PDOS which is different from that of QE5.3.0.

I think this problem may be traced back to the following thing:
In the change from QE5.3.0 to QE 5.4.0,
the variable "igk" (indices of PW of each k) in PW/src/atomic_wfc.f90
becomes "igk_k". However "igk" in PP/src/atomic_wfc_nc_proj.f90 have not  
changed. In fact I can calculate correctly when I apply this kind of  
modification to the atomic_wfc_nc_proj.f90 .
Is my idea correct ?

Best regards,
Mitsuaki

-- 
------------------------------------------------------
Dr. Mitsuaki Kawamura
Software Advancement Team
Supercomputer Center
Materials Design and Characterization Laboratory
The Institute for Solid State Physics, Kashiwa, Japan
e-mail : mkawamura at issp.u-tokyo.ac.jp
------------------------------------------------------