Bug: Fix convergence problem of spin-polarized metaGGA calculation for nspin=2#3328
Conversation
kirk0830
left a comment
There was a problem hiding this comment.
I see, undefined behavior(s) eliminated
|
I've test the performance of this repository on Fe-bcc and C-diamond system, which both have good convergence performance. For C-diamond, the TMAG and AMAG after converge to For Fe-bcc, if One more thing to notice is that if I set And |
I will realize a new mixing method for tau, which maybe improve convergence for metaGGA. After that, we can double-check if these confusions are still there. |
Fix #3260 and #3146, in which scan calculation.
What's changed?
psiToRho()mix_rho()set_mixing()andmix_reset()After this PR, convergence problem of spin-polarized metaGGA calculation is almost solved.