[QE-users] GIPAW crashes with nspin=2

[E. Lora]

Dear QE developers

By computing a spin polarized system, with the following scf input:

&CONTROL
   calculation='scf' ,
   verbosity='high' ,
   etot_conv_thr=1.0D-5 ,
   forc_conv_thr=1.0D-3 ,
   pseudo_dir = './',
   outdir='./tmp'
   prefix='CaMnGe2O6U4_paw_spin' ,
/
&SYSTEM
   a     =  10.444750
   b     =  9.312284
   c     =  5.548569
   cosac = -0.2437201
   ibrav = -12
   nat   = 40
   ntyp  = 5
   starting_magnetization(1) = 1.d-6 ,
   starting_magnetization(2)= 0.5 ,
   starting_magnetization(3)= -0.5 ,
   starting_magnetization(4) = 1.d-6 ,
   starting_magnetization(5) = 1.d-6 ,
   ecutwfc = 70 ,
   ecutrho = 540.0 ,
   occupations='smearing' , smearing='mv' , degauss=1d-4 ,
   nspin=2 ,
!  lda_plus_u=.true. , Hubbard_U(2)=4 , Hubbard_U(3)=4 ,
  ! nosym=.true.
/
&ELECTRONS
   electron_maxstep=1000000,
   conv_thr=1.D-8 ,
   mixing_beta = 0.3 ,
/

ATOMIC_SPECIES
   Ca   40.07800  Ca.pbe-spn-kjpaw_psl.1.0.0.UPF
  Mn1  54.93800  Mn.pbe-spn-kjpaw_psl.0.3.1.UPF
  Mn2  54.93800  Mn.pbe-spn-kjpaw_psl.0.3.1.UPF
   Ge   72.63000  Ge.pbe-dn-kjpaw_psl.0.2.2.UPF
    O   15.99900   O.pbe-n-kjpaw_psl.1.0.0.UPF
ATOMIC_POSITIONS {crystal}
Ca           0.0000000000       0.3045100000       0.2500000000
Ca           0.0000000000       0.6954900000       0.7500000000
Ca           0.5000000000       0.8045100000       0.2500000000
Ca           0.5000000000       0.1954900000       0.7500000000
Mn1          0.0000000000       0.9065930000       0.2500000000
Mn2          0.0000000000       0.0934070000       0.7500000000
Mn1          0.5000000000       0.4065930000       0.2500000000
Mn2          0.5000000000       0.5934070000       0.7500000000
Ge           0.2863390000       0.0970150000       0.2323850000
Ge           0.7136610000       0.9029850000       0.7676150000
Ge           0.7136610000       0.0970150000       0.2676150000
Ge           0.2863390000       0.9029850000       0.7323850000
Ge           0.7863380000       0.5970160000       0.2323840000
Ge           0.2136620000       0.4029840000       0.7676160000
Ge           0.2136620000       0.5970160000       0.2676160000
Ge           0.7863380000       0.4029840000       0.7323840000
O            0.1131650000       0.0970310000       0.1439770000
O            0.8868350000       0.9029690000       0.8560230000
O            0.8868350000       0.0970310000       0.3560230000
O            0.1131650000       0.9029690000       0.6439770000
O            0.6131640000       0.5970310000       0.1439770000
O            0.3868360000       0.4029690000       0.8560230000
O            0.3868360000       0.5970310000       0.3560230000
O            0.6131640000       0.4029690000       0.6439770000
O            0.3681360000       0.2528710000       0.3564940000
O            0.6318640000       0.7471290000       0.6435060000
O            0.6318640000       0.2528710000       0.1435060000
O            0.3681360000       0.7471290000       0.8564940000
O            0.8681360000       0.7528710000       0.3564940000
O            0.1318640000       0.2471290000       0.6435060000
O            0.1318640000       0.7528710000       0.1435060000
O            0.8681360000       0.2471290000       0.8564940000
O            0.3562090000       0.0352240000       0.9781070000
O            0.6437910000       0.9647760000       0.0218930000
O            0.6437910000       0.0352240000       0.5218930000
O            0.3562090000       0.9647760000       0.4781070000
O            0.8562080000       0.5352240000       0.9781060000
O            0.1437920000       0.4647760000       0.0218940000
O            0.1437920000       0.5352240000       0.5218940000
O            0.8562080000       0.4647760000       0.4781060000
K_POINTS {automatic}
4 4 8 0 0 0

we intend to obtain the electric field gradient:

&inputgipaw
job='efg'
prefix='CaMnGe2O6U4_paw_spin'
tmp_dir='./tmp/'
verbosity='high'
spline_ps= .true.
Q_efg(1) = -6.65 !43Ca
Q_efg(2) = 33.0 !55Mn
Q_efg(3) = 33.0 !55Mn
Q_efg(4) = -19.6 !73Ge
Q_efg(5) = -2.56 !17O
q_gipaw = 0.01
/

However, when running gipaw.x for this system we obtain an error message 
related to segmentation fault problems. By increasing RAM, nodes or even 
decreasing the K-point mesh, the problem persists. By performing the 
gipaw calculation on another system, which is spin-unpolarized the 
calculation runs smoothly without any problem. We think it might be 
related to the calculation being spin polarized, since after the 
'q-space interpolation' the calculation crashes, after outputting the line:

(RHO,ZETA) => (RHO_UP,RHO_DOWN)

      select_spin: s_maj=1 s_min=2 rho_diff=    0.000000

Can you please advise on what the problem might be related to and if 
there is a workaround this issue? We have also tried removing the LDAU 
calculation, to check if the problem could be related to the inclusion 
of the Hubbard paramters, but unfortunately with no luck.

Regards

Estelina Silva