Dear Nick Sir, I am thankful for your reply.
Regards Ananya On 9 Feb 2018 2:35 a.m., "Nick Papior" <nickpap...@gmail.com> wrote: > Dear Anaya, > > If you search the mailing list about mixing you will find that there is no > way to generalize parameters for optimal convergence for *all* systems. > > I.e. *you* have to test. > If it converges extremely slowly, it may be because the mixing weight is > too low. Secondly, increasing the history may also help in certain cases. A > reasonable number is around 6-12, but you have to try it. > > As for > SCF.Mix.Spin all > I would suggest to not use this flag. It shouldn't influence how it > performs (in its current state). > > 2018-02-07 11:12 GMT+01:00 Ananya Rajpoot <rajpootanaya1...@gmail.com>: > >> Dear users, >> >> I am interested in simulation of Graphene ribbons doped with Fe. When I >> proceed with the spin-polarized calculations, the SCF in first CG run >> didn't converge. It is taking 500 SCF cycles or even more during >> convergence. >> >> Should I change pulay parameters? Or Linear mixing schemes? If pulay then >> should I increase weight or number of pulay?? >> >> Also, what else should I correct to get faster convergence?? or >> Electronic temp should be reduced? >> >> >> My fdf file parameters are as follows: >> >> PAO.BasisSize DZP >> >> XC.functional GGA >> >> XC.authors PBE >> >> %block kgrid_Monkhorst_Pack >> >> 1 0 0 0.0 >> >> 0 1 0 0.0 >> >> 0 0 49 0.0 >> >> %endblock Kgrid_Monkhorst_Pack >> >> MeshCutoff 370.0 Ry >> >> MaxSCFIterations 5000 >> >> SCF.Mixer.Method Pulay >> >> SCF.Mixer.Variant original >> >> SCF.Mixer.Weight 0.1 >> >> SCF.Mixer.History 2 >> >> SCF.Tolerance.DM 1.d-4 >> >> >> MD.TypeOfRun CG >> >> MD.MaxForceTol = 0.01 eV/Ang >> >> MD.NumCGsteps 5000 >> >> MD.VariableCell .false. >> >> >> SCF.Mix.First false >> >> SCF.Mix.AfterConvergence false >> >> SCF.RecomputeHAfterScf false >> >> SCF.Mix density >> >> SCF.Converge.DM true >> >> >> Spin polarized >> >> SCF.Mix.Spin all >> >> FixSpin false >> >> DM.InitSpinAF false >> >> >> WriteCoorXmol true >> >> SolutionMethod diagon >> >> Thank u so much... >> >> With Regards >> Anaya >> >> > > > -- > Kind regards Nick >
