<div dir="ltr"><div>Context: I'm interested in a series of computations regarding
properties of compounds of the skutterudite family . As a test I tried
to run cell optimization of NaFe4Sb12 which is reported to have
magnetisation of about 1.5-1.7 bohr/per Fe. When I tried to run
relpbe/pslib calculation with ultrasoft pseudos and noncolin/spinorb, I
got oscillation of about ~0.04 eV magnitude in SCF procedure. I was not
able to achieve convergence playing with mixing and broadening setup.
When I tried colin setup with n-nc pseudos, I got quick convergence, but
resulting magnetic moment is zero. The system is supposedly half-metal.<br></div><div><br></div><div class="gmail-adL">The question: What are suggested parameters (requirements for pseudopotentials, (super)cell size. number of k-points) for electronic structure calculation of half-metals including heavy atoms? <br><br></div></div>