[QE-users] hp.x root not converged for LiNiO2

["Börries von Seggern"]

Dear QE-Community,
I am trying to calculate the Hubbard U parameter for Nickel in LiNiO2
using hp.x in qe-7.1 (compiled using intel compilers). Unfortunately the
calculations keep crashing, even when using input files adapted from an
older forum post, in which someone managed to do the hp.x calculation for
LiNiO2 using qe-6.5... (
https://www.mail-archive.com/users@lists.quantum-espresso.org/msg38839.html
).
So far I have tried using different k-point meshes in the scf run, as well
as using a PBE-relaxed geometry and smearing, however using very soft
Pseudopotentials, the calculation keeps crashing for q-point 4, as hp.x
seems to start with a somewhat faulty response function and fails at
converging this giving output such as:
     atom #  1   q point #   4   iter #   2
     kpoint 146 sternheimer_kernel: root not converged, thresh <  2.256E+06
     chi:   1 **************   residue:  **************
     Average number of iter. to solve lin. system:   33.5
Using very soft PPs (for example excluding semicore states for Ni) from
pslib, hp.x also gives the "S matrix not positive semidefinite error",
which doesn`t happen for harder PPs, as to be expected. However the
sternheimer_kernel: root not converged error unfortunately persists. :/
Do you know, what might be the issue here? Any help would be greatly
appreciated.
Pls find a example input file using harder pps below, as well as the
input/outputfiles in this OneDrive Folder:
https://1drv.ms/u/s!An1DvoEOkGMHhCisvOFIG4uQRE2O?e=cASPwY

Best wishes,

Börries von Seggern

Helmholtz Institute for Materials and Energy Berlin



The SCF-input file (restarted from a less tightly converged SCF run with
smearing):

 &control
 calculation = 'scf'
 restart_mode = 'from_scratch',
 prefix = 'lno'
 pseudo_dir = 'pseudo'
 outdir = 'tmp'
 verbosity = 'normal'
 tstress=.true.
/

&system
ibrav = 5,
celldm(1) = 9.5631,
celldm(4) = 0.83635,
nat = 4,
ntyp = 3,
ecutwfc = 110.0,
ecutrho = 660.0,
nspin = 2,
occupations ='fixed',
tot_magnetization = 1.00,
nbnd = 17
/

 &electrons
conv_thr =  1.d-15
mixing_beta = 0.2
electron_maxstep=250
startingpot='file'
startingwfc='file'
/

ATOMIC_SPECIES
Ni       58.6934  Ni.pbe-spn-rrkjus_psl.1.0.0.UPF
O        15.999   O.pbe-n-rrkjus_psl.1.0.0.UPF
Li       6.941    Li.pbe-s-rrkjus_psl.1.0.0.UPF

ATOMIC_POSITIONS {crystal}
Ni       0.0000000000   0.0000000000   0.0000000000
O        0.2582132811   0.2582132811   0.2582132811
O        0.7417867189   0.7417867189   0.7417867189
Li       0.5000000000   0.5000000000   0.5000000000

K_POINTS {automatic}
  12 12 12 0 0 0

HUBBARD {ortho-atomic}
   U Ni-3d 0.0001

And the corresponding hp.x input file:

 &inputhp
prefix = 'lno',
outdir = 'tmp',
nq1 = 2, nq2 = 2, nq3 = 2,
alpha_mix=0.1            !default 0.3, try 0.1 when in trouble
slightly
iverbosity = 2

/