[Wien] [Fwd: Re: Stuck in LAPW2 after increasing the number of k-points]

[jadhikari@clarku.edu]

Prof. P Blaha,

The ghost band problem was solved by repeated adjustment of the energy
parameters according to the values in the scf2 file or by the use of
-in1new option. The converged calculation had a QTL-B value of 4.04, which
may have been the source of error for higher k-points calculations.

Now the calculation for 15 k-points in IBZ has converged and still has the
QTL-B value of 4.04.

Thank you very much for your kind suggestion for this problem.
Subin





More k-points do NOT need more memory (need more disk space).

You said, also the previous calculations had "ghost bands", which could
be "solved". What did you do ? I'd expect you still have ghostbands.

Are you sure that the previous calculations were converged WITHOUT
ghostbands ?

Messages like:  "eigenvalues below  -9.0" are a strong indication of
ghostbands !!!

When does it occur ? Right in the first iteration (starting from
previous "good" run ?? or after many iterations (what was :DIS ?? ...)



Eventually, rm case.scf, *.bro*, restore "good-case" and start scf again.

jadhikari at clarku.edu schrieb:
> Dear WIEN2k users,
>
> The scf cycle converged well for 4 k-points in the IBZ and so did for 8
> k-points for Potassium Calcium Niobate, sp gp 11. But when moving to the
> next higher k-points in the IBZ, ie 15, it gets stuck in the LAPW2 step.
> Also for 24 and 44 k-points it never goes past the LAPW2 step of the first
> cycle.
>
> The message is just "Error in LAPW2" and in scf files it says many
> eigenvalues below  -9.0, the Emin choosen. In some of the calulations it
> says to check the IN1 and KGEN files. I am unable to figure out what is to
> be seen in these two files as I find no clue about the error. Is it about
> the memory of the system for larger number of k-points? or about the
> appropriate linearization of the energy? The calculations with 4 and 8
> k-points showed ghost bands which could be solved , but this time it shows
> no such QTL-B errors.
>
> I would be very grateful for any insights into this problem.
> Thanking you.
> Subin
> _______________________________________________
> Wien mailing list
> Wien at zeus.theochem.tuwien.ac.at
> http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien
_______________________________________________
Wien mailing list
Wien at zeus.theochem.tuwien.ac.at
http://zeus.theochem.tuwien.ac.at/mailman/listinfo/wien