[Wien] Question on internal parameter optimization

[Peter Blaha]

Hi,

I checked your Bi case and could verify the problems.

Actually, the problem is not so much in clmaddsub, but in SRC_dstart.

The variable PI was not defined properly (needed ONLY for R cases), thus
the list of fourier coefficients is wrong.
This does NOT matter for a regular scf calculation, but leads
to a crash with the new charge-extrapolation.

You can replace SRC_dstart/init.f with the attached file.

PS: As mentioned by L.Marks, pairhess does NOT support R lattices.
However, there's no such check in w2web and it will run pairhess anyway.
Thus "PORT" may start a bit strange in this case, but to my understanding
it should "correct" itself (its hessian).


>      > >> (mini) arrived at end -> exit
>      > Fallback to compatibility mode with "old" save_lapw
>      > broyden files deleted, clm*, dmat*, vorb*, scf and struct files
>     saved under
>      > bismuth_1
>      >
>      > DSTART ENDS
>      > 0.852u 0.012s 0:01.23 69.9% 0+0k 0+0io 0pf+0w
>      > forrtl: severe (24): end-of-file during read, unit 13, file
>      > /home/s010/jacob/bismuth/old_super.clmsum
>      > Image PC Routine Line Source
>      >
>      > clmaddsub 0809ABFB Unknown Unknown Unknown
>

-- 

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