[QE-users] HSE calculation on two different machines: OK with QE-7.3.1, charge error with QE-7.2.2

[Thierry JOLICOEUR]

Hi dear QE users, 

I am trying to capture the band structure of PbSe - a semiconductor with a small gap. 
The literature says that this is a tricky case requiring a hybrid functional treatment 
(Hummer et al PhysRevB 75, 195211 (2007)). 
I did such a calculation with QE-7.3.1 compiled locally on a 6-core machine. 
The calculation ends without problems. 

I decided to try to use a bigger HPC cluster with QE-7.2.2 
and as a test I ran the very same calculation which now fails 
with the infamous "charge error" message. 
This happens with the same number of MPI threads and also with more threads. 

There are not so many pseudos recommended for HSE so I use SG15 variants 
in both cases. I just perform the SCF run of pw.x because the K points 
include the L point where the gap is. I don't need the high-res non SCF 
calculation (yet) which is anyway not possible with HSE. 

the input file : 

======================================================================= 
&CONTROL 
outdir = 'scratch/' 
prefix = 'PbSe' 
pseudo_dir = '/home/tjoli/handson_pwscf/sg15pseudos/' 
calculation='scf' 
nstep = 200 
wf_collect = .true. 
verbosity='high' 
/ 

&SYSTEM 
celldm(1) = 11.6 
ecutwfc = 48.0 
ecutrho = 300.0 
nbnd = 24 
ibrav = 2 
lspinorb = .true. 
nat = 2 
noncolin = .true. 
ntyp = 2 
input_dft='hse' 
nqx1 = 3 
nqx2 = 3 
nqx3 = 3 
/ 

&ELECTRONS 
conv_thr = 1e-8 
diagonalization = "cg" 
mixing_beta = 0.5 
mixing_mode = 'plain' 
startingwfc = 'atomic-random' 
/ 

ATOMIC_SPECIES 
Pb 207.2 Pb_ONCV_PBE_FR-1.0.upf 
Se 78.96 Se_ONCV_PBE_FR-1.1.upf 


ATOMIC_POSITIONS alat 
Pb 0.000000000000 0.000000000000 0.000000000000 
Se 0.500000000000 0.000000000000 0.000000000000 

K_POINTS automatic 
6 6 6 0 0 0 

==================================================================== 
the error message: 

ethr = 1.00E-02, avg # of iterations = 10.9 

negative rho (up, down): 1.120E+00 0.000E+00 

total cpu time spent up to now is 2.4 secs 

WARNING: integrated charge= 10.23051228, expected= 20.00000000 

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 
Error in routine electrons (1): 
charge is wrong 
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

stopping ... 

====================================================================== 

the same calculation on my local QE-7.3.1 gives : 

starting charge 19.8764, renormalised to 20.0000 

which is what is expected from the electron numbers in the pseudos. 


hints appreciated ! 
Th Jolicoeur 

senior researcher CNRS 
================================================================== 
Thierry Jolicoeur 
Institut de Physique Theorique-DRF 
room 16 blg 774 
Orme des Merisiers 
91190 Gif sur Yvette 
tel +33 169087512 
P44W+V7 Saint-Aubin 
ICBM: 48.708004N, 2.145753E 

"It is unworthy of excellent men to lose hours like slaves 
in the labour of calculation which could safely be 
relegated to anyone else if machines were used." 
Gottfried Wilhelm Leibniz 
================================================================== 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20241112/a4ac77c8/attachment.html>