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