# [SIESTA-L] Fwd: LDA+U band gap with very large U

Dear List,

We are studying the Cu2O system with LDA+U theory (please see the last
portion of the mail). We found that when we used a very LARGE U parameter
for oxygen, Up, like 10000, Siesta still calculates the band energy and did
not warn about its numerical values.

<sample setting>
%block LDAU.proj
Cu 1         # number of shells of projectors
n=3 2        # n, l
10.000 0.000 # U(eV), J(eV)
0.000  0.000 # rc, \omega (default values)
O 1          # number of shells of projectors
n=2 1        # n, l
10000.00 0.000  # U(eV), J(eV) <== the Up,J
0.000  0.000  # rc, \omega (default values)
%endblock LDAU.proj

Physically, large, Up means a very large separation of the electronic
states - so it should be avoided (pardon me If I am wrong in this case).

*From the Log,*
*Energies are (when Ud for Cu = 10. ev, Up = 10.eV):*
siesta: Final energy (eV):
siesta:  Band Struct. =    -472.209961

*Energies are (when Ud for Cu = 10. ev, Up = 10000.eV):*
siesta: Final energy (eV):
siesta:  Band Struct. =   -5730.637191

Corresponding band diagram is also attached.

So will it be a better idea to have a warning message for very large U
values for LDA+U calculations?

Our other concern is, how one can systematically vary U? Can we use large
values like Up=19, to fit an experimental band gap?

with thanks and regards

Krishna

*****************************************************************************************************************************************
INPUT
*******
SystemName Cu4O2

LatticeConstant  1.0 Ang
%block LatticeVectors
4.287813 0.000000 0.000000
0.000000 4.287813 0.000000
0.000000 0.000000 4.287813
%endblock LatticeVectors

NumberOfAtoms    6
NumberOfSpecies  2

%block ChemicalSpeciesLabel
1  29   Cu
2  8   O
%endblock ChemicalSpeciesLabel

AtomicCoordinatesFormat Fractional

%block AtomicCoordinatesAndAtomicSpecies
0.250000 0.250000 0.750000 1
0.750000 0.250000 0.250000 1
0.250000 0.750000 0.250000 1
0.750000 0.750000 0.750000 1
0.500000 0.500000 0.500000 2
0.000000 0.000000 0.000000 2
%endblock AtomicCoordinatesAndAtomicSpecies

%block kgrid_Monkhorst_Pack
8 0 0 0.
0 8 0 0.
0 0 8 0.
%endblock kgrid_Monkhorst_Pack

# Basis definition

PAO.EnergyShift 200 meV
PAO.SplitNorm 0.15
PAO.BasisSize DZ
PAO.BasisType           split

XC.Functional LDA
XC.authors PZ
kgrid_cutoff          50. Ang
MeshCutoff 250.0 Ry

MaxSCFIterations 50
DM.MixingWeight 0.5
DM.NumberPulay 3
ElectronicTemperature 300 K
#--------------------------------------------- LDA U
# For an AFM-Type II structure, choose the following

%block DM.InitSpin       # Describe the initial magnetic order (on Ni only)
1 +
2 -
3 +
4 -
5 -
6 +
%endblock DM.InitSpin

DM.UseSaveDM            .false.      # Use DM Continuation files

WriteIonPlotFiles     .true.

LDAU.FirstIteration              .false.
LDAU.ThresholdTol                1.0d-3
LDAU.PopTol                      4.0d-4
LDAU.ProjectorGenerationMethod   2
LDAU.CutoffNorm                  0.90

%block LDAU.proj
Cu 1         # number of shells of projectors
n=3 2        # n, l
10.000 0.000 # U(eV), J(eV)
0.000  0.000 # rc, \omega (default values)
O 1          # number of shells of projectors
n=2 1        # n, l
1000 0.000  # U(eV), J(eV)
0.000  0.000  # rc, \omega (default values)
%endblock LDAU.proj

#---------------------------------------------

LongOutput          .true.
WriteKpoints        .true.
WriteCoorXmol       .true.
WriteCoorStep       .true.
WriteCoorCerius     .true.
Diag.DivideAndConquer .false.
WriteKbands             T

BandLinesScale ReciprocalLatticeVectors

%block BandLines
1   0.0   0.0   0.0    \Gamma          #Beginat\Gamma
22  0.0   0.5   0.0    X               # 22 points from \Gamma to X
22  0.5   0.5   0.0    M               # 22 points from X to M
33  0.0   0.0   0.0    \Gamma           # 33 points from M to \Gamma
39  0.5   0.5   0.5    R                # 39 points from \Gamma to R
33  0.0   0.5   0.0    X                # 33 points from R to X
%endblock BandLines

-----------------------------------------------------------

--