[QE-users] problem with hq.x
José Carlos Conesa Cegarra
jcconesa at icp.csic.es
Tue Aug 4 17:02:06 CEST 2020
Dear colleagues,
I have found a problem when running hp.x after a pw.x calculation on a
nitride. The input to pw.x was the following:
....................
&CONTROL
title = 'calc CuGeSnN4'
calculation = 'scf'
restart_mode = 'from_scratch'
prefix = 'CuGeSnN4'
outdir = './tmp'
etot_conv_thr = 1.0D-5
pseudo_dir = '..'
/
&SYSTEM
space_group = 47
A=4.3628, B=4.1528, C=4.3120
nat= 7, ntyp= 4
ecutwfc = 50.0
nspin = 2
occupations ='smearing'
degauss= 0.003
starting_magnetization(1)=1
lda_plus_u = .TRUE.
Hubbard_U(1)=0.01
Hubbard_U(2)=0.01
/
&ELECTRONS
diagonalization='david'
electron_maxstep = 100
mixing_mode = 'plain'
mixing_beta = 0.7
conv_thr = 1.0d-8
/
ATOMIC_SPECIES
Cu 63.5 Cu_pbe_v1.2.uspp.F.UPF
N 14.0 N.pbe.theos.UPF
Ge 72.6 Ge.pbe-dn-kjpaw_psl.1.0.0.UPF
Sn 118.7 Sn_pbe_v1.uspp.F.UPF
ATOMIC_POSITIONS crystal_sg
Ge 0.50000 0.50000 0.00000
Cu 0.50000 0.00000 0.50000
Sn 0.00000 0.50000 0.50000
N 0.50000 0.50000 0.50000
N 0.50000 0.00000 0.00000
N 0.00000 0.50000 0.00000
N 0.00000 0.00000 0.50000
K_POINTS automatic
6 6 6 0 0 0
....................
The run proceeded seemingly smoothly. Then I undertook a hp.x
calculation using the following input:
....................
&INPUTHP
prefix = 'CuGeSnN4'
outdir = './tmp'
iverbosity = 2
nq1 = 4, nq2 = 4, nq3 = 4
/
....................
and found the following result in stderr:
forrtl: Operation not permitted (repeated as many times as cores were
used)
and
forrtl: severe (28): CLOSE error, unit 0, file "Unknown"
followed by several references on errors in hp.x (all these lines
repeated as well as many times as cores were used)
The output of hp.x just contained at the end this line:
WARNING! All Hubbard atoms must be listed first in the
ATOMIC_POSITIONS card of PWscf
But since this was the case I do not think that this may be the cause of
the problem.
I have been told that unit 0 is stderr; but how can this be, since these
lines appeared in the stderr output itself? I have been using qe-6.5
(for both pw.x and hp.x) and the SLURM queuing system
Please give advice.
Regards,
--
José C. Conesa
Research Professor
Instituto de Catálisis y Petroleoquímica, CSIC
Marie Curie, 2; Campus de Cantoblanco
28028 Madrid (Spain)
Telef. +34 915854766
More information about the users
mailing list