Dear all,
I am doing a relaxation with vanadium on graphene, and the scf loop goes crazy:
siesta: iscf Eharris(eV) E_KS(eV) FreeEng(eV) dDmax Ef(eV)
siesta: 1 -97094.38150 -99070.91635 -99070.91635 39.07032 -4.23154
timer: Routine,Calls,Time,% = IterSCF 1 882.442 95.03
elaps: Routine,Calls,Wall,% = IterSCF 1 223.031 94.74
siesta: 2 -97140.17446 -99051.35598 -99051.50902 34.72834 -4.20753
siesta: 3 -97226.76276 -98965.36619 -98965.50463 35.42815 -4.12217
siesta: 4 -94936.08243 -99371.74467 -99371.89250 35.10492 -4.22078
siesta: 5 -90678.74567 -99715.25553 -99715.47141 24.75070 -4.07217
siesta: 6 -86315.83616 -100050.89144 -100051.12562 36.83745 -3.67364
siesta: 7 -97702.64129 -98510.48955 -98510.724571051.40098 -3.89029
-And it only gets worse after more iterations.
I have tried almost all thinkable combinations of Pulay mixing, mixing rate
etc, but without success. I have also increased temperature, disabled spin,
changed energyshift, tried different basis sets for all elements....No luck at
all! Sometimes I can get it to converge, but then after a CG step or two it
goes nuts again.
If I leave out vanadium, the problem disappears. I have used this vanadium PS
many times in different calculations, but have never come across this issue
until now.
Have you got any ideas? I have listed my .fdf file below (the long list of
coordinates is left out).
Hope you have some good advice. Thanks!
NumberOfSpecies: 3
NumberOfAtoms: 631
LatticeConstant 1.0 Ang
%block LatticeVectors
55. 0. 0.
0. 8. 0
0. 0. 36.84
%endblock LatticeVectors
%block ChemicalSpeciesLabel
1 6 C_pbe
2 1 H
3 23 V
%endblock ChemicalSpeciesLabel
AtomicCoordinatesFormat Ang
%block AtomicCoordinatesAndAtomicSpecies
40.470458 0.000000 0.000196 1
39.049818 0.000000 0.000213 1
36.210059 -0.000001 0.000237 1
0.709653 0.000000 0.000252 1
.....
.....
%endblock AtomicCoordinatesAndAtomicSpecies
%block kgrid_Monkhorst_Pack
1 0 0 0.0
0 1 0 0.0
0 0 1 0.0
%endblock kgrid_Monkhorst_Pack
DM.UseSaveDM T
#SaveTotalPotential T
#SaveRho T
#SaveElectrostaticPotential T
#MD.UseSaveXV T
SolutionMethod Diagon
%block PAO.BasisSizes
C_pbe DZ
H DZ
%endblock PAO.BasisSizes
%block PAO.Basis
V 4 # Species label, number of l-shells
n=4 0 2 P 1 # n, l, Nzeta, Polarization, NzetaPol
7.497 6.783
1.000 1.000
n=3 1 2 # n, l, Nzeta
3.048 1.943
1.000 1.000
n=4 1 2 # n, l, Nzeta
9.870 8.285
1.000 1.000
n=3 2 2 # n, l, Nzeta
4.901 2.758
1.000 1.000
%endblock PAO.Basis
XC.functional GGA
LongOutput .true.
XC.authors PBE
SystemLabel bulk
DM.MixSCF1 T
MaxSCFIterations 600 # Maximum number of SCF iter
DM.MixingWeight 0.01 # New DM amount for next SCF cycle
DM.Tolerance 1.d-4 # Tolerance in maximum difference
DM.UseSaveDM true # to use continuation files
DM.NumberPulay 4
SpinPolarized true
WriteMullikenPop 1
MeshCutoff 175 Ry
MD.TypeOfRun CG
MD.NumCGSteps 190
MD.UseSaveCG true
MD.UseSaveXV true
MD.MaxCGDispl 0.02 Ang
MD.MaxForceTol 0.05 eV/Ang
WriteCoorXmol true
WriteMDXmol true
WriteMDhistory true
