Dear Harsh Shah,it may sound banal, but - if you hope for a useful answer you should outline your problem in a meaningful way. "A certain system with 110 atoms did converge but the other one with 769does not" - this is not much of information.You define 2000 SCF iterations and 5000 CG steps - who of them do not converge?What is your mixing parameter, anyway? How does your failed convergence look like? In some systems (no idea what is yours) with many nearly degenerate statesa very drastic reduction of mixing might help(to get the convergence trend right, before trying more efficient mixing schemes).Best regards Andrei Postnikov ----- Harsh Shah <shah19041...@gmail.com> a écrit : >Dear Siesta User,
I am running a file in parallel with around 769 atoms in stampede2 supercomputer and my file is not converging. It has been running without giving any error for around 10-12 days and I think it should not take that much time.I ran few other files having 110 atoms and it is giving proper output. I am using:MD.MaxForceTol 0.025MeshCutoff(in Ry) 80MaxSCFIterations 2000MD.NumCGSteps 5000DM.Tolerance 1.d-4 Is there anything which I should change from above? What can be other possibilities do to which the file is not converging? I would really appreciate if you could help me out. Regards,Harsh Shah