when I did an unrestricted optimization on an open-shell system, I
often used "keeporbitals=Nkeep" and "freeze" if SCF convergence
failed in case of customizing the occupations myself. However,
keeporbitals function seemed don't work as well as I anticipated. It
was still of high probability that scf convergence failed or wf was
optimized to a excited state. Furthmore, the occ remained changeable
after "Nkeep" iterations, for example a electron jumped among two or
three orbitals----I tried to use vshift to prevent this jumping,but
failed.
thus, I'm confused that how the keeporbitals=nkeep works?

--
so let us be up and doing
with a heart for any fate
still pursuing and still achieving
learn to labor and to wait