[Wien] error in runafm_lapw with Pratt

[Peter Blaha]

> Always fix the first error first ...
> I looked at it for just a second, line 362 of SRC_clmcopy/clmcopy.f is
>             READ(MARGN,710) JATOM
>             710  FORMAT(4x,i2)
>
>
> It seems (look at the lines around this one) that the first line of the
> case.scf2up is read into the character-variable MARGN.  Then, as this
> line should contain an atom number, the integer JATOM is read from it as
> the fifth and sixth position.
>
> My guess :
> Since atom numbers are now being treated differently in the
> 100-atom-notation (they need three positions), this read statement is
> probably not good anymore.
> Check the scf2up-file, and see what is needed (probably the
> format-statement 710 should read (3x,i3) or (4x,i3) or something like
> that); change it in clmcopy.f, recompile, see whether it works now, and
> let us know ...


format-statement 710 should read  (4x,i3)
You are right, the scf file created by clmcopy can't be correct.

Checking clmcopy reveals several of those wrong formats. All 710-770 must
be changed, also some ranges in corresponding write(...

A new version of clmcopy is on the web. It has never been tested though!!!

When testing it, please compare scf2up with scf2dn, where the latter has
been created by the new clmcopy.

> 	With a helful suggestion from Kevin Jorissen, I have gotten
> runafm_lapw to work properly in its Wien2k_03 implementation, but only
> with Broyden mixing.  If I try to use Pratt mixing, it crashes in the
> third iteration with a formatting error in clmcopy.  Bravely continuing

At least the errors in clmcopy cannot not correlate with BROYD or PRATT ....



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