[Wien] Problem for convergence

[Peter Blaha]

>     I was running spin-polarized calculation to Ni (a=b=c=14.0 Ang).
> But the big system had the difficulty of converging.The charge distance
> was oscillatory.Could you tell me how to do if I was need to be adjusted
> some parameters.

Are you trying a single Ni atom in a big box ?

Do it spin polarized.

Use just ONE k-point (if you want to simulate a single atom!)

Otherwise I suspect your problem is similar to what was discussed before:
TM atoms (Fe, Co,Ni,..) have the problem that they do NOT converge properly.
This means: When the 4s state is occupied by 2 electrons, it will rise
in energy and the 3d levels become lower in energy. As a result of this,
only the 3d states will be occupied, but the 4s level will be empty. Of
course this occupation drives the 4s eigenvalue down until it is lower than
3d and the occupation changes again. ......

A possible solution is a fairly large broadening parameter and GAUSS or TEMP
method in case.in2. With that it should be possible to get some "partial"
occupation of the 4s state keeping it at the same energy as the 3d.

Regards

PS: The same problem happens in lstart, but there it is solved by fixed
occupations (given in the input).

                                      P.Blaha
--------------------------------------------------------------------------
Peter BLAHA, Inst.f. Materials Chemistry, TU Vienna, A-1060 Vienna
Phone: +43-1-58801-15671             FAX: +43-1-58801-15698
Email: blaha at theochem.tuwien.ac.at    WWW: http://info.tuwien.ac.at/theochem/
--------------------------------------------------------------------------