For easy nonmagnetic cases, any of the three would get you convergence easily. For difficult magnetic cases or when a complicated functional is use, I've found that only linear mixing works.
Hi Lars, I am indeed David's PhD student. We met a long time ago in ORNL. Please send me the link to alaska[.]subedi[AT]polytechnique[.]edu. Thanks a lot, Alaska
Dear Lars and Kay, I read in the forum that an interface to Wannier90 was being developed for Elk. Can it be made available? I'd be happy to run the examples given in the Wannier90 distribution as tests and compare the results with that obtained using Quantum Espresso. Best, Alaska
Hi Thomas, Please check that state i and j are not degenerate. If they are from a...
Dear Kay, The file modvars.f90 does not compile with Cray fortran, presumably because...