[Pw_forum] Problem with PWCOND

[vborisov]

Dear all,

I noticed a problem while using the PWCOND code for calculating
the transmission for more than one k-point within the two-dimensional 
BZ.
Whereas there is no problem with the conductance calculation for
a single k-point, the following error message appears once I provide
a list with n>=2 points in the input file for conductance calculation.
It is shown below on a standard example for an Al nanowire (same 
behavior
also observed for different systems):


forrtl: severe (151): allocatable array is already allocated
Image              PC                Routine            Line        
Source
pwcond.x           0000000000BEC50D  Unknown               Unknown  
Unknown
pwcond.x           0000000000BEB015  Unknown               Unknown  
Unknown
pwcond.x           0000000000B866D0  Unknown               Unknown  
Unknown
pwcond.x           0000000000B1EDCF  Unknown               Unknown  
Unknown
pwcond.x           0000000000B5D7CB  Unknown               Unknown  
Unknown
pwcond.x           000000000047E8C4  init_gper_.A               72  
init_gper.f90
pwcond.x           000000000045CAB9  do_cond_.A                462  
do_cond.f90
pwcond.x           000000000045388A  MAIN__                     22  
condmain.f90
pwcond.x           0000000000437FDC  Unknown               Unknown  
Unknown
libc.so.6          00002B17BB725CDD  Unknown               Unknown  
Unknown
pwcond.x           0000000000437ED9  Unknown               Unknown  
Unknown
forrtl: severe (151): allocatable array is already allocated
Image              PC                Routine            Line        
Source
pwcond.x           0000000000BEC50D  Unknown               Unknown  
Unknown
pwcond.x           0000000000BEB015  Unknown               Unknown  
Unknown
pwcond.x           0000000000B866D0  Unknown               Unknown  
Unknown
pwcond.x           0000000000B1EDCF  Unknown               Unknown  
Unknown
pwcond.x           0000000000B5D7CB  Unknown               Unknown  
Unknown
pwcond.x           000000000047E8C4  init_gper_.A               72  
init_gper.f90
pwcond.x           000000000045CAB9  do_cond_.A                462  
do_cond.f90
pwcond.x           000000000045388A  MAIN__                     22  
condmain.f90
pwcond.x           0000000000437FDC  Unknown               Unknown  
Unknown
libc.so.6          00002B76A0D2CCDD  Unknown               Unknown  
Unknown
pwcond.x           0000000000437ED9  Unknown               Unknown  
Unknown
--------------------------------------------------------------------------
mpirun has exited due to process rank 1 with PID 17595 on
node node327.cruncher exiting without calling "finalize". This may
have caused other processes in the application to be
terminated by signals sent by mpirun (as reported here).
--------------------------------------------------------------------------

The input for PWSCF:

  &control
     calculation='scf'
     restart_mode='from_scratch',
     pseudo_dir = '/scratch/fhgfs/vladislav/espresso-5.0.2/pseudo/',
     outdir='/scratch/fhgfs/vladislav/espresso-5.0.2/tmp/test/',
     prefix='al'
  /
  &system
     ibrav = 6,
     celldm(1) =5.3,
     celldm(3) =1.414,
     nat= 2,
     ntyp= 1,
     ecutwfc = 15.0,
     occupations='smearing',
     smearing='methfessel-paxton',
     degauss=0.01
  /
  &electrons
     conv_thr = 1.0e-8
     mixing_beta = 0.7
  /
ATOMIC_SPECIES
  Al 26.98 Al.pz-vbc.UPF
ATOMIC_POSITIONS
  Al 0. 0. 0.0
  Al 0.5 0.5 0.707
K_POINTS (automatic)
  4 4 4 1 1 1


The input for PWCOND:

  &inputcond
     outdir='/scratch/fhgfs/vladislav/espresso-5.0.2/tmp/test/',
     prefixl='al'
     prefixs='al'
     tran_file ='trans.al.Ef'
     ikind = 1
     iofspin = 1
     energy0 = 0.00d0
     denergy = -0.01d0
     ewind = 1.d0
     epsproj = 1.d-3
     delgep = 1.d-12
     cutplot = 3.d0
     nz1 = 22
  &
     2
     0.0  0.0  0.5
     0.5  0.5  0.5
     1

************************************************************
************************************************************

The calculation stops after the first k-point and gives
the aforementioned message. Has anyone encountered the same problem
and possibly know how to solve it?

Another issue is related to the use of a distributed network
filesystem. The problem is a heavy workload noticed in our
filesystem that leads to performance problems on our high-performance
cluster. In particular, the meta servers report a very high work
request. This isn't observed using any other code. Exemplarily,
below there is the time report of PWCOND during one of such runs:

PWCOND       :     0h18m CPU        5h19m WALL

init         :    432.25s CPU  18519.07s WALL (       1 calls)
poten        :      0.49s CPU      0.52s WALL (       2 calls)
local        :     49.24s CPU     51.44s WALL (       1 calls)

scatter_forw :    598.01s CPU    599.04s WALL (       2 calls)

compbs       :     12.88s CPU     12.89s WALL (       1 calls)
compbs_2     :     10.84s CPU     10.84s WALL (       1 calls)

Notice the large difference between the CPU and the WALL times
for the init subroutine. This was observed during the parallel
execution with different number of processors both for 5.0.1 and
5.0.2 versions, and on different architectures using OPENMPI 
environment.

I would very much appreciate any help with these matters.

With kind regards,
Vladislav Borisov


Max Planck Institute of Microstructure Physics
Weinberg 2, 06120, Halle (Saale), Germany
Tel No: +49 345 5525448
Fax No: +49 345 5525446
Email: vborisov at mpi-halle.mpg.de