I am trying to run selfconsistency on graphite.
At first I used gwcalctyp 19, that If I'm not wrong, corresponds to the QuasiParticle SelfConsistent GW scheme where only energies are updated.
I have a strange behavior for many bands: the corrections DeltaE have strong oscillations around 0, and they do not seem to go towards convergence, as an example I quote what happens in Gamma for iterations 5,6,7,8,9
k = 0.000 0.000 0.000
Band E_lda <Vxclda> E(N-1) <Hhartree> SigX SigC[E(N-1)] Z dSigC/dE Sig[E(N)] DeltaE E(N)_pert E(N)_diago
1 -21.122 -15.174 -15.080 -5.859 -25.356 16.803 1.000 0.000 -8.552 0.668 -14.411 -14.411
Band E_lda <Vxclda> E(N-1) <Hhartree> SigX SigC[E(N-1)] Z dSigC/dE Sig[E(N)] DeltaE E(N)_pert E(N)_diago
1 -21.122 -15.174 -14.411 -5.947 -25.436 16.307 1.000 0.000 -9.129 -0.665 -15.076 -15.076
Band E_lda <Vxclda> E(N-1) <Hhartree> SigX SigC[E(N-1)] Z dSigC/dE Sig[E(N)] DeltaE E(N)_pert E(N)_diago
1 -21.122 -15.174 -15.076 -5.859 -25.356 16.806 1.000 0.000 -8.549 0.668 -14.408 -14.408
Band E_lda <Vxclda> E(N-1) <Hhartree> SigX SigC[E(N-1)] Z dSigC/dE Sig[E(N)] DeltaE E(N)_pert E(N)_diago
1 -21.122 -15.174 -14.408 -5.947 -25.436 16.311 1.000 0.000 -9.125 -0.664 -15.072 -15.072
Band E_lda <Vxclda> E(N-1) <Hhartree> SigX SigC[E(N-1)] Z dSigC/dE Sig[E(N)] DeltaE E(N)_pert E(N)_diago
1 -21.122 -15.174 -14.417 -5.947 -25.436 16.303 1.000 0.000 -9.132 -0.663 -15.080 -15.080
I tried also to use gwcalctyp 12 and I don't have the same behavior: DeltaE converges monotonically to 0 and 4-5 iterations are enough to reach convergence.
Can you comment on this oscillation of the corrections?
Thanks a lot.
So what is the difference between the two gwcalctyp I used (12 and 19) and what is recorded in the output QPS files?