[Wien] LDA+U

[Stefaan Cottenier]

> while I was trying to carry out the LDA+U calculation, I found that the
> scf procedure heavily oscilates as stated in the UG. therefore, I wanna to
> use it in combination with the spin-orbital coupling. in the UG, it
> recommends that " preferably running first LSDA+so and then slowly
> switching on the LDA+O orbital field". who can give a detailed description
> about how to operate in the calculation process.

First converge LDA+SO, as usual. Then take a small U (one tenth of what you
need), and a small mixing factor in case.inm (down to 0.0005 might be
necessary from time to time). Converge it, save and increase U in as much
steps as you need to arrive at the wanted U.

Stefaan