[QE-users] a problem related to the electric field calculations

[Rostami Samare]

Dear all,

I am working on a Ti-doped Fe2O3 systems and want to calculate the dipoles
in the presence of electric field. I face many difficulties for these kind of
calculations. I was wondering if some one could help me to fix my problems.

I am using DFT+U for this work. my system contains 60 atoms and is a
ferromagnetic.  and the electronic structure is not symmetric due to Ti
doping.  when I use a pure Fe2O3, I do not have any problem , but by
adding
Ti, I get the convergence problem with both SCF and Dipole parts.

First, for this system without electric field ,  using different k-points
I will get different values for dipoles in each directions and it does not
seem that I reach to a absolute value by increasing the k points.

K points Dipole(x) Dipole(y)  Dipole(z)
3 6 2    -12.03    -3.57       -7.36
3 6 4    -10.58    -2.22       -7.26
6 12 4   -11.41    -5.87       -7.84

I actually do not now how to choose k point and to rely on the dipole
results.

The other problem is the convergence of Dipole in the presence of electric
field. for example, I applied field for this system (0.0005 in z
direction) with 3 6 4 kpoints , but there are two problem for
converging. First, after many times of restarting, it crashed and get an
error for computing cholesky . An important problem is the dipoles do not
seem to converge and vary in each iteration. It is the result of dipoles
in the last iterations before crashing : (For these iterations ethr is in
the order of 1.0E-09)

   Electronic Dipole per cell in x,y,z direction :  -10.370  -4.809  4.756
   Electronic Dipole per cell in x,y,z direction :  -9.7570  -3.731  4.603
   Electronic Dipole per cell in x,y,z direction :  -8.8681  -4.904  1.382
   Electronic Dipole per cell in x,y,z direction :  -10.383  -2.574  1.838
   Electronic Dipole per cell in x,y,z direction :  -7.8399  -3.152 -2.605
   Electronic Dipole per cell in x,y,z direction :  -8.4782  -3.036  1.031

I increased nberrycyc or used a smaller field but it do not help.
I do not know what the problem is and how can I fix it.


I thank you in advance for your help.

Samare Rostami
Posdoc at ICTP, Italy