[QE-users] k-point parallelization with localization_thr>0

Zeeshan Ahmad azeeshan at uchicago.edu
Thu Oct 29 21:05:29 CET 2020


Hi,

I got a segmentation fault when I tried to parallelize over k-points using localization_thr > 0 when they are specified in crystal. Works fine with automatic. Is this the expected behavior?

Here is an input file for reproducing. I tried -nk 2 in both cases, -nk 4 gives an error even with automatic (both -nk 2 and 4 work fine without localization_thr).

 &control
    calculation = 'scf'
    restart_mode='from_scratch',
    prefix='Si-HSE',
    pseudo_dir = '/home/azeeshan/pseudopot/sg15_ONCV/',
    outdir='./si/'
 /
 &system    
    ibrav=  2, celldm(1) =10.20, nat=  2, ntyp= 1,
    ecutwfc =30.0,  nbnd = 8,
    input_dft='hse', 
    nqx1 = 1, nqx2 = 1, nqx3 = 1, 
    x_gamma_extrapolation = .true., 
     localization_thr = 0.002
    exxdiv_treatment = 'gygi-baldereschi',
    nosym = .true., noinv = .true
 /
 &electrons
    mixing_beta = 0.7
 /
ATOMIC_SPECIES
 Si  28.086  Si_ONCV_PBE-1.1.upf
ATOMIC_POSITIONS alat
 Si 0.00 0.00 0.00 
 Si 0.25 0.25 0.25 
K_POINTS crystal
8
  0.00000000  0.00000000  0.00000000  1.250000e-01
  0.00000000  0.00000000  0.50000000  1.250000e-01
  0.00000000  0.50000000  0.00000000  1.250000e-01
  0.00000000  0.50000000  0.50000000  1.250000e-01
  0.50000000  0.00000000  0.00000000  1.250000e-01
  0.50000000  0.00000000  0.50000000  1.250000e-01
  0.50000000  0.50000000  0.00000000  1.250000e-01
  0.50000000  0.50000000  0.50000000  1.250000e-01


Thanks,
Zeeshan
--
Zeeshan Ahmad
Postdoctoral Researcher
Pritzker School of Molecular Engineering
University of Chicago

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20201029/20ac0fe8/attachment.html>


More information about the users mailing list