This thread is locked.Only browsing is available.
Top Page > Browsing
Convergence problem for NEGF self consistent loop of Fe/MgO/Fe system
Date: 2015/11/20 21:01
Name: Hiroshi Tsukahara

Dear developers and users,

I am performing simulations using OPENMX code to calculate spin dependent condactance. I have been trying to calculate the conductance using NEGF method. However My simulation still does not converge even if number of NEGF loop is over 2000.

A simulation system is Fe/MgO/Fe which is mentioned in User's manual. I choose atomic positions and parameters by refering to Phys. Rev. B 81, 035116.

could anybody help me to get convergenc?

best regards


I use OPENMX3.5 and input file for NEGF calculation is as follows:

#
# File Name
#

System.CurrrentDirectory ./ # default=./
System.Name negf-Fe_MgO_Fe
level.of.stdout 1 # default=1 (1-3)
level.of.fileout 1 # default=1 (0-2)

DATA.PATH ../DFT_DATA # default=../DFT_DATA13

NEGF.filename.hks.l lead-l-Fe_fcc.hks
NEGF.filename.hks.r lead-r-Fe_fcc.hks

NEGF.Num.Poles 130 # default=150
NEGF.scf.Kgrid 7 7 # default=1 1

NEGF.scf.Iter.Band 100 # defalut=6
NEGF.Output.for.TranMain on

NEGF.bias.voltage 0.0 # default=0.0 (eV)
NEGF.bias.neq.im.energy 0.01 # default=0.01 (eV)
NEGF.bias.neq.energy.step 0.02 # default=0.02 (eV)

Dos.fileout off # on|off, default=off
NEGF.Dos.energyrange -15.0 25.0 5.0e-3 #default=-10.0 10.0 5.0e-3 (eV)
NEGF.Dos.energy.div 200 # default=200
NEGF.Dos.Kgrid 7 7 # default=1 1

NEGF.tran.energydiv 200 # default=200
NEGF.tran.energyrange -10 10 1.0e-3 # default=-10.0 10.0 1.0e-3 (eV)
NEGF.tran.Kgrid 160 160 # default= 1 1



#
# Definition of Atomic Species
#

Species.Number 3
<Definition.of.Atomic.Species
Fe Fe5.0-s2p2d1 Fe_LDA
Mg Mg5.5-s2p2 Mg_TM_PCC
O O5.0-s2p2d1 O_LDA
Definition.of.Atomic.Species>


#
# Atoms
#

Atoms.SpeciesAndCoordinates.Unit Ang # Ang|AU
Atoms.Number 8
<Atoms.SpeciesAndCoordinates
1 Mg 6.459 0.000 0.000 4.0 4.0 0.0 0.0 0.0 0.0 0 off
2 Mg 10.513 0.000 0.000 4.0 4.0 0.0 0.0 0.0 0.0 0 off
3 Mg 8.486 1.433 1.433 4.0 4.0 0.0 0.0 0.0 0.0 0 off
4 Mg 12.540 1.433 1.433 4.0 4.0 0.0 0.0 0.0 0.0 0 off
5 O 8.486 0.000 0.000 3.0 3.0 0.0 0.0 0.0 0.0 0 off
6 O 12.540 0.000 0.000 3.0 3.0 0.0 0.0 0.0 0.0 0 off
7 O 6.459 1.433 1.433 3.0 3.0 0.0 0.0 0.0 0.0 0 off
8 O 10.513 1.433 1.433 3.0 3.0 0.0 0.0 0.0 0.0 0 off
Atoms.SpeciesAndCoordinates>

#
# Lead-Left
#

LeftLeadAtoms.Number 4
<LeftLeadAtoms.SpeciesAndCoordinates # Unit=Ang.
1 Fe 0.000 0.000 0.000 8.0 6.0 0.0 0.0 0.0 0.0 0 off
2 Fe 1.433 1.433 1.433 8.0 6.0 0.0 0.0 0.0 0.0 0 off
3 Fe 2.866 0.000 0.000 8.0 6.0 0.0 0.0 0.0 0.0 0 off
4 Fe 4.299 1.433 1.433 8.0 6.0 0.0 0.0 0.0 0.0 0 off
LeftLeadAtoms.SpeciesAndCoordinates>

#
# Lead-Right
#
RightLeadAtoms.Number 4
<RightLeadAtoms.SpeciesAndCoordinates # Unit=Ang.
1 Fe 14.700 0.000 0.000 8.0 6.0 0.0 0.0 0.0 0.0 0 off
2 Fe 16.133 1.433 1.433 8.0 6.0 0.0 0.0 0.0 0.0 0 off
3 Fe 17.566 0.000 0.000 8.0 6.0 0.0 0.0 0.0 0.0 0 off
4 Fe 18.999 1.433 1.433 8.0 6.0 0.0 0.0 0.0 0.0 0 off
RightLeadAtoms.SpeciesAndCoordinates>

#
# SCF or Electronic System
#

scf.XcType LSDA-CA # LDA|LSDA-CA|LSDA-PW|GGA-PBE
scf.SpinPolarization On # On|Off|NC
scf.SpinOrbit.Coupling Off # On|Off, default=off
scf.ElectronicTemperature 1000.0 # default=300 (K)
scf.energycutoff 300.0 # default=150 (Ry)
scf.maxIter 10000 # default=40
scf.EigenvalueSolver NEGF # DC|GDC|Cluster|Band
scf.lapack.dste dstevx # dstegr|dstedc|dstevx, default=dstegr
scf.Kgrid 7 7 7 # means n1 x n2 x n3
scf.Mixing.Type rmm-diisk # Simple|Rmm-Diis|Gr-Pulay|Kerker|Rmm-Diisk
#scf.Kerker.factor 40.0 # default=1
scf.Init.Mixing.Weight 0.001 # default=0.30
scf.Min.Mixing.Weight 0.0001 # default=0.001
scf.Max.Mixing.Weight 0.02 # default=0.40
scf.Mixing.History 60 # default=5
scf.Mixing.StartPulay 30 # default=6
scf.Mixing.EveryPulay 1 # default=6
scf.criterion 1.0e-6 # default=1.0e-6 (Hartree)

#
# MD or Geometry Optimization
#

MD.Type nomd # Nomd|Opt|DIIS|NVE|NVT_VS|NVT_NH
MD.maxIter 1 # default=1
MD.TimeStep 1 # default=0.5 (fs)
MD.Opt.criterion 1.0e-5 # default=1.0e-4 (Hartree/bohr)
メンテ
Page: [1]

Re: Convergence problem for NEGF self consistent loop of Fe/MgO/Fe system ( No.1 )
Date: 2015/11/23 19:15
Name: Artem Pulkin

I never managed to converge anything magnetic in OpenMX NEGF. If you want to spend your time on it you should start with increasing your scattering region and including some Fe atoms in it. Then, scf.Mixing.History should be something huge like several hundreds. Afterwards you should play with mixing weight values and see what happens with your iterations when you decrease them.

Artem
メンテ
Re: Convergence problem for NEGF self consistent loop of Fe/MgO/Fe system ( No.2 )
Date: 2016/03/15 13:56
Name: T.Ozaki

Hi,

By repeating the calculation with the restart file, you may be able to obtain
the convergent result. Actually, I achieved the SCF convergence with the procedure
to get the result for Fe|MgO|Fe shown in the manual.

Regards,

TO
メンテ

Page: [1]