[Wien] To LO or not to LO ?

[Peter Blaha]

> I seem to have a blind spot when it comes to using LO's. I've noticed that
> with bulk rutile TiO2 (running LDA) write_in1 is picking up a D LO (GGA
> does not); Q-d-low is 0.0545 which is just enough. With a reduced TiO2-x
> calculation run spin-polarized (LDA) the LO is not being picked up in part
> because the charge is now up/dn partitioned so is less than 0.05, the
> default cutoff. I could raise -ql in write_in1 for the bulk, or reduce -ql
> for the reduced TiO2-x calculation; any suggestions as to which is more
> "correct"?

Using additional LOs should enrich the basis set and thus always be better,
(unless you get ghostbands because of approximate linear dependencies).

Of course, it might be that you do not gain much (E-tot does not decrease too 
much). Just test it, but judge if the additional effort (5 extra basis functions
per atom) justify the gain.


                                      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/
--------------------------------------------------------------------------