Gregory, Even though your system is not too large (not a very large number of atoms, try decreasing the mixing weight to something like 0.05 or even lower. Works in most cases. Maybe a slightly higher cutoff (250-300 Ry, maybe) would help, but I never really made a systematic study on how cutoff values influence the convergence of the scf loop. However, the lower missing weight is almost sure to help.
Cheers, Marcos Vous avez écrit / You have written / Lei ha scritto / Você escreveu... Gregory Geneste > Quoting Marcos Verissimo Alves <[EMAIL PROTECTED]>: > >> Questions that come up on previous messages, also... :) What are the >> parameters you are using for your calculation? >> >> Cheers, >> >> Marcos >> > > > Dear Marcos, > > here follows my input file as well as the basis used for the calculation. > > > > # > # General system descriptors > # > > SystemName BFO > SystemLabel BFO > NumberOfAtoms 30 > NumberOfSpecies 3 > %block ChemicalSpeciesLabel > 1 83 Bi > 2 26 Fe > 3 8 O > %endblock ChemicalSpeciesLabel > > # > # Basis definition > # > > PAO.EnergyShift 0.01 Ry > PAO.BasisType split > block PAO.Basis < basis.fdf > %block PS.lmax > Bi 3 > Fe 3 > O 3 > %endblock PS.lmax > > # > # Lattice, coordinates, k-sampling > # > > LatticeConstant 5.517471 Ang > %block LatticeVectors > 1.00 0.00 0.00 > 0.00 1.00 0.00 > 0.00 0.00 3.20 > %endblock LatticeVectors > > AtomicCoordinatesFormat Bohr > %block AtomicCoordinatesAndAtomicSpecies > 0.00000 0.00000 0.00000 1 > 5.21326 5.21326 0.00000 1 > 0.00000 0.00000 7.40297 1 > 5.21326 5.21326 7.40297 1 > 0.00000 0.00000 14.80550 1 > 5.21326 5.21326 14.80550 1 > 0.00000 5.21326 3.70126 2 > 0.00000 5.21326 11.10424 2 > 5.21326 0.00000 3.70126 2 > 5.21326 0.00000 11.10424 2 > 0.00000 5.21326 18.50676 2 > 5.21326 0.00000 18.50676 2 > 0.00000 5.21326 0.00000 3 > 5.21326 0.00000 0.00000 3 > 0.00000 5.21326 7.40297 3 > 5.21326 0.00000 7.40297 3 > 2.60663 2.60663 3.70126 3 > 7.81988 7.81988 3.70126 3 > 2.60663 7.81988 3.70126 3 > 7.81988 2.60663 3.70126 3 > 2.60663 2.60663 11.10424 3 > 7.81988 7.81988 11.10424 3 > 2.60663 7.81988 11.10424 3 > 7.81988 2.60663 11.10424 3 > 0.00000 5.21326 14.80550 3 > 5.21326 0.00000 14.80550 3 > 2.60663 2.60663 18.50676 3 > 7.81988 7.81988 18.50676 3 > 2.60663 7.81988 18.50676 3 > 7.81988 2.60663 18.50676 3 > %endblock AtomicCoordinatesAndAtomicSpecies > > > # Magnetism > FixSpin .true. > TotalSpin 0.0 > NonCollinearSpin .false. > > %block DM.InitSpin > 7 + # Atom index, spin, theta, phi (deg) > 8 - > 9 - > 10 + > 11 + > 12 - > %endblock DM.InitSpin > > > %block kgrid_Monkhorst_Pack > 4 0 0 0.0 > 0 4 0 0.0 > 0 0 1 0.0 > %endblock kgrid_Monkhorst_Pack > > > # > # DFT, Grid, SCF > # > > XC.Functional LDA > XC.Authors CA # CA = Ceperley-Alder > SpinPolarized .true. > MeshCutoff 200 Ry > > > Diag.DivideAndConquer .false. > #note: this one has been added after the message: > #Failure to converge standard eigenproblem Stopping Program from Node: 0 > > > DM.MixingWeight 0.1000 # New DM amount for next SCF cycle > DM.Tolerance 1.d-5 # Tolerance in maximum difference > # between input and output DM > DM.NumberPulay 3 > #DM.NumberKick 21 > #DM.KickMixingWeight 0.200 > MaxSCFIterations 150 > > > # > # Eigenvalue problem: order-N or diagonalization > # > > SolutionMethod diagon > ElectronicTemperature 300 K > > # > # Molecular dynamics and relaxations > # > > MD.TypeOfRun CG # Type of dynamics: > # - CG > # - Verlet > # - Nose > # - Parrinello-Rahman > # - Nose-Parrinello-Rahman > # - Anneal > # - FC > MD.VariableCell .false. # The lattice is relaxed together with > # the atomic coordinates? > MD.NumCGsteps 300 # Number of CG steps for > # coordinate optimization > MD.MaxCGDispl 0.5 Bohr # Maximum atomic displacement > # in one CG step > MD.MaxForceTol 0.01 eV/Ang # Tolerance in the maximum > # atomic force > MD.MaxStressTol 0.0005 eV/Ang**3 > > # > # Output options > # > > LongOutput .false. > WriteCoorStep .true. > WriteForces .true. > WriteMullikenPop 0 # Write Mulliken Population Analysis > # - 0 = None > # - 1 = atomic and orbital charges > # - 2 = 1 + atomic overlap > population > # - 3 = 2 + orbital overlap > population > WriteCoorXmol .true. > WriteMDCoorXmol .false. > WriteMDhistory .false. > WriteEigenvalues .true. > AllocReportLevel 2 # Sets the level of the allocation > report > # - 0 = No report at all (default) > # - 1 = Only total memory peak > and where > # it ocurred > # - 2 = detailed report printed > only > # at normal program > termination > # - 3 = detailed report printed at > # every new memory peak > # - 4 = print every individual > # (re)allocation or > deallocation > > # > # Options for saving/reading information > # > > DM.UseSaveDM .false. # Use DM Continuation files > MD.UseSaveXV .false. # Use stored positions and velocities > MD.UseSaveCG .false. # Use CG history information > SaveRho .false. # Write valence pseudocharge at the > mesh > SaveDeltaRho .false. # Write RHOscf-RHOatm at the mesh > SaveElectrostaticPotential .false. # Write the total elect. pot. at the > mesh > SaveTotalPotential .false. # Write the total pot. at the mesh > #WriteSiestaDim .false. # Write minimum dim to siesta.h and > stop > WriteDenchar .false. # Write information for DENCHAR > > > My basis and pseudopotentials include semicore electrons for Bi and Fe > and perfectly work > for bulk R3c BiFeO3: > > PAO.BasisType split > %block PAO.Basis > Bi 4 > n=5 2 2 P 1 > 0.000 0.000 > 1.000 1.000 > n=6 0 2 P 1 > 0.000 0.000 > 1.000 1.000 > n=6 1 2 P 1 > 0.000 0.000 > 1.000 1.000 > n=7 0 1 > 0.000 > 1.000 > Fe 4 > n=3 1 2 P 1 > 0.000 0.000 > 1.000 1.000 > n=4 0 2 P 1 > 0.000 0.000 > 1.000 1.000 > n=3 2 2 P 1 > 0.000 0.000 > 1.000 1.000 > n=4 1 1 > 0.000 > 1.000 > O 2 > n=2 0 2 > 0.000 0.000 > 1.000 1.000 > n=2 1 2 P 1 > 0.000 0.000 > 1.000 1.000 > %endblock PAO.Basis > -- Dr. Marcos Verissimo Alves Post-Doctoral Fellow Unité de Physico-Chimie et de Physique des Matériaux (PCPM) Université Catholique de Louvain 1 Place Croix du Sud, B-1348 Louvain-la-Neuve Belgique ------ Gort, Klaatu barada nikto. Klaatu barada nikto. Klaatu barada nikto. Free translation: Gort, Google is your friend. Google is your friend. Google is your friend.