Dear expert users of the SPRKKR package,
Can someone give me guidelines to resolve convergence issues in the SCF run? For pure compound, the SCF cycle gets converged quickly (within 50 electronic iterations), but with substitutional doping (~ 2%), all the maximum allowed iteration (300) gets exhausted, and the system doesn't converge. The details of the input file and error message are as follows:
Input
MODE SP-SREL
LLOYD
TAU BZINT= POINTS NKTAB= 250
ENERGY GRID={5} NE={30}
ImE=0.0 Ry EMIN=-0.2 Ry
SCF NITER=300 MIX=0.101 VXC=PBE
TOL=0.00001 MIXOP=0.20 ISTBRY=1
QIONSCL=0.80
NOSSITER
Error
*******************************************************************************
no problems with CPA-cycle !
*******************************************************************************
*******************************************************************************
300 ERR 0.113E+01 0.698E-01 EF 0.78274 -0.00105 D 83.132 M 3.0776 0.0000
ETOT -10814.43503634 SCF - cycle not converged - NSCFITER exhausted
*******************************************************************************
Dear Rohit
Do you optimize crystal structure before any calculations?
Thank you, Jaafar Jalilian , for your quick reply!
Yes, the parent compound was optimized using VASP, but not the doped structures. I am unsure whether any relaxation (ionic positions or volume) procedure is available in the SPRKKR package. Please let me know if it is available, and I am very new to the package.
The package uses CPA approximations for the doping calculations, so it doesn't provide any supercell that could be optimized using VASP. I am interested in calculating exchange interaction with the Lichtenstein approach using the SPRKKR package.
Basically, force and lattice constant optimizations are necessary for doped system.
Please check SPRKKR for this option.
Many thanks Shahram Yalameha for you kind help!
I will compile the latest version and follow your suggestions.
I tried reducing the mixing parameter, and it worked in one case.
Please let me know if the lattice parameter and atomic positions optimization (relaxation) process are available in SPRKKR?
- Badges
- Science topic