[Wien] QTL-B VALUE

[Peter Blaha]

> I do calculations on VPO systems and run into the following problem:
> the scf converges nicely (without any indications for ghost-bands)
> - but when i increase emax in .in1 for the
> calculation of ELNES spectra from 1.5 to 2.0 or 2.5 eV,     <== Ry ?????
> lapw2 -qtl produces an error.
> in the .help file for the P-atom I can see that some numbers are replaced by
> ********
> and in .scf2 the warning
>
> QTL-B VALUE .EQ. **********   !!!!!!
>
> is printed.
>
> The RMT radii are not too different (V: 1.55, P: 1.45 and O: 1.35),
> Gmax=14, rkm=7 - 8, and the energy to separation in lstart was -10Ry
> (otherwise charge is leaking out)
> i calculated with the option -in1new.
>
> Is there any obvious thing I do wrong, or
> does anyone have a suggestion about what I can do?

LAPW (or APW+lo) is a "linearized" method, i.e. the E-variation of the
radial basisfunction is "linearized". This linearization is good only for
a limited E-range, maybe about +-1 Ry (depending on the specific n,l) around
the "linearization energy", i.e. the energy where you expand your u_l.
Usually these E_l are set just below EF.

When you increase EMAX you calculate states far away from E_l and the
linearization may break down, leading to "ghoststates".

Possible solutions:
a) Check (from the ***** in the help files) for which atom and which l the
problems occur.
b) Sometimes it is enough to raise the corresponding E_l by eg. 0.5 -1 Ry
c) If there was no LO selected for this atom and l, add an LO at high energy (
e.g. at 2 Ry)
d) If there is a LO at low energy (and with your small spheres you probably
have lots of LOs for V and P !), change the LO-energy from the low
energy (e.g. -6 Ry) to a high E (2 Ry). This will remove the e.g. the P-2p
states from the basis, but add a 4p basis.

PS: All this not for scf, but just for the SPECTRA !!

> another thing: the columns in the .help03X files: the first is the sum (ul,
> dul/de), the second is ul,
> the third corresponds to the energy derivative
> but what is displayed by the other three?

These are the coefficients of a local orbital!

Regards


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