Dear all, We have a problem concerning the phonon calculation in QE-7.4.1, which aborted in dependence of the used compiler suite either with numerical errors or with a segmentation fault.
Because calculations with QE-6.8 completed successfully we are doubtful if some
input data has to be adjusted or if we do something wrong during compilation or
if there is some other issue.
What we did is the following:
- Running all calculations on the same hardware with mpi. No other processes
run at the same time.
- We compiled QE-6.8 with the intel compiler suite 2022.2.0 including mkl
2022.2.0 and intel mpi 2021.7.0 (referred to intel-6.8)
- We compiled QE-7.4.1 with the same intel compiler suite (referred to
intel-7.4.1)
- We compiled QE-7.4.1 with the gnu compiler version 10.2.1. (referred to
gnu-7.4.1)
- We have two materials to calculate: NbOCl2 and NbOBr2. For each material we
did three calculations with the above described binaries: After the
self-consistent calculation (pw.x) we did a phonon calculation with the ph.x
binary which belongs to the specific pw.x binary. The input data are the same
for the calculation with the different binaries.
The results with intel-6.8 are numerically stable and correct.
The calculations with intel-7.4.1 fail consistently for both materials after
the first SCF phonon step, and no output is generated.
The phonon calculation with gnu-7.4.1 of NbOCl2 finished without errors, but
the numerical results are completely incorrect. The eigenvalues and mode
symmetries show significant discrepancies compared to the reference
(intel-6.8), and the values are clearly unphysical. For NbOBr2, the calculation
crashes with a NaN error, so in both cases, this binary is unreliable.
We have no further idea to solve the problems with 7.4.1. On the other side we
like to use some of the new features in this version.
What can we do?
The scf and ph input files for both materials are attached.
If necessary I can provide the output files and temporary files on a dropbox
like share.
Many thanks in advance and best regards,
Ulrike Nitzsche
--
Ulrike Nitzsche | email: u.nitzs...@ifw-dresden.de
| phone: +49-351-4659-463
&CONTROL calculation = 'scf' etot_conv_thr = 8.0000000000d-05 forc_conv_thr = 1.0000000000d-04 restart_mode = 'from_scratch' outdir = './tmp/' prefix = 'br' pseudo_dir = './' tprnfor = .true. tstress = .true. verbosity = 'high' / &SYSTEM degauss = 0.02 ecutrho = 400 ecutwfc = 50 ibrav = 0 nat = 8 ntyp = 3 occupations = 'smearing' smearing = 'methfessel-paxton' vdw_corr='grimme-d2' / &ELECTRONS conv_thr = 1.00d-8 mixing_beta = 0.7 / &IONS ion_dynamics='bfgs' / &CELL cell_dynamics='bfgs', press=0.0, press_conv_thr=0.1, / ATOMIC_SPECIES Cl 35.453 cl_pbe_v1.4.uspp.F.UPF Nb 92.90638 Nb.pbe-spn-kjpaw_psl.0.3.0.UPF O 15.9994 O.pbe-n-kjpaw_psl.0.1.UPF ATOMIC_POSITIONS crystal Nb 0.0248982578 0.9754504248 0.7790154193 Nb 0.0245495752 0.9751017422 0.2209845807 Cl 0.1127193297 0.1557301266 0.5696825202 Cl 0.8442698734 0.8872806703 0.4303174798 Cl 0.1389424627 0.1688333441 0.0816080077 Cl 0.8311666559 0.8610575373 0.9183919923 O 0.4918204317 0.5095466864 0.7714330597 O 0.4904533136 0.5081795683 0.2285669403 K_POINTS automatic 8 8 5 0 0 0 CELL_PARAMETERS angstrom 6.547461840 -1.945052017 -0.011052571 6.547461840 1.945052017 -0.011052571 -1.792698061 0.000000000 6.466848685
Phonon dispersion &inputph prefix='br', tr2_ph = 1.0d-15, outdir='../scf2/tmp/', fildyn='br.dyn', ldisp=.true., nq1=2,nq2=2,nq3=2, /
&CONTROL calculation = 'scf' etot_conv_thr = 8.0000000000d-05 forc_conv_thr = 1.0000000000d-04 restart_mode = 'from_scratch' outdir = './tmp/' prefix = 'br' pseudo_dir = './' tprnfor = .true. tstress = .true. verbosity = 'high' / &SYSTEM degauss = 0.02 ecutrho = 400 ecutwfc = 50 ibrav = 0 nat = 8 ntyp = 3 occupations = 'smearing' smearing = 'methfessel-paxton' vdw_corr='grimme-d2' / &ELECTRONS conv_thr = 1.00d-8 mixing_beta = 0.7 / &IONS ion_dynamics='bfgs' / &CELL cell_dynamics='bfgs', press=0.0, press_conv_thr=0.1, / ATOMIC_SPECIES Br 79.904 br_pbe_v1.4.uspp.F.UPF Nb 92.90638 Nb.pbe-spn-kjpaw_psl.0.3.0.UPF O 15.9994 O.pbe-n-kjpaw_psl.0.1.UPF ATOMIC_POSITIONS crystal Nb 0.7822165592 0.2170595504 0.2163285646 Nb 0.7829404496 0.2177834408 0.7836714354 Br 0.9028273400 0.4077256410 0.0839962778 Br 0.5922743590 0.0971726600 0.9160037222 Br 0.6092921406 0.1241678578 0.4303403698 Br 0.8758321422 0.3907078594 0.5696596302 O 0.2529978606 0.7451808512 0.2229042548 O 0.2548191488 0.7470021394 0.7770957452 K_POINTS automatic 8 8 5 0 0 0 CELL_PARAMETERS angstrom 6.986724451 -1.939733069 0.000457675 6.986724451 1.939733069 0.000457675 -1.889184394 0.000000000 6.794686330
Phonon dispersion &inputph prefix='br', tr2_ph = 1.0d-15, alpha_mix(1)=0.5, outdir='../scf/tmp/', fildyn='br.dyn', search_sym=.true., ldisp=.true., nq1=2,nq2=2,nq3=2, /
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________________________________________ The Quantum ESPRESSO Foundation stands in solidarity with all civilians worldwide who are victims of terrorism, military aggression, and indiscriminate warfare. -------------------------------------------------------------------------------- Quantum ESPRESSO is supported by MaX (www.max-centre.eu) users mailing list users@lists.quantum-espresso.org https://lists.quantum-espresso.org/mailman/listinfo/users
