Dear all, I am doing some phonon calculations and have some convergence problems in the "Self-consistent Calculation" at some q-points and Representations. I noticed that the "|ddv_scf|^2"-value jumps in the second step to a greater value, like in the scf calculation I have done for this phonon calculation. I minimized the jump in the scf calculation, when I use "mixing_mode='TF' ". Is this option also implemented for ph.x? I also tried decreasing the "alpha_mix"-value, it helps but I hope to get better results with mixing_mode (as it does in an scf test calculation). Do I have other options to decrease the cpu time and number of iterations? (a normal convergence takes 10 iterations and not 30 with alpha_mix 0.3) Thanks for help
Thomas Gruber Institute for Theoretical Physics TU Bergakademie Freiberg
