Re: no xmu.dat file created when using cfaverage card in feff8.2

[Benjamin Gilbert <bgilbert@eps.berkeley.edu>]

dear Anna,

	I also found this problem, which Alex Ankudinov solved. There 
is a bug in feff8.2, which can be solved in the source code and 
recompiling. I'm pasting the details below. With this fix, the 
CFAVERAGE card works as detailed in the manual.
	If you do not have a compiler for the computer you are using, 
it's possible one of the earlier compiled versions of Feff may work.

	Note that the use of the CFAVERAGE card with XANES 
calculations is potentially inaccurate, because FEFF does not 
recalculate the scattering potentials at each absorber site. 
Consequently, Alex discourages use of the CFAVERAGE card with XANES. 
You should consider calculating each XANES individually, and 
combining the results (including the normalization factors). This 
rapidly becomes tedious for large clusters, and ideally, FEFF would 
have an option to recalculate the potentials at each site within the 
CFAVERAGE loop. The CFAVERAGE card may work for your calculations, 
but you should check the results.

regards

ben

The source code fixes:

From: Alex Ankudinov <alex@leonardo.phys.washington.edu>
> > Thank you for reporting the bug in feff8.2.
> > I found a quick fix: simply comment out or remove the line
> >      iabs = 1
> > in subroutine ffmod6.


	After this change, the chi file (EXAFS calculations) or xmu 
file (XANES calculations) will be produced when using the CFAVERAGE 
card.

	There's a related bug associated with the use of the potential indexes.

From: Alex Ankudinov <alex@leonardo.phys.washington.edu>
> >   I also found a fix for this: in subroutine rdinp change lines
> >
> > c     No gaps allowed in unique pots.  Make sure we have enough
> > c     to overlap all unique pots 0 to nph.
> >       do 340  iph = 0, nph
> >
> >   to
> >
> > c     No gaps allowed in unique pots.  Make sure we have enough
> > c     to overlap all unique pots 0 to nph.
> >       if (iphabs.gt.0 .and. iatph(0).le.0)   iatph(0) = iatph(iphabs)
> >       do 340  iph = 0, nph
> >
> > I.e. you need to insert just one line.


> Hi,
>
> I am using feff8.2 (executable downloaded from the web)
> to calculate the XAFS spectra of a cluster and would like to
> average over all Ni atoms of the cluster. In the calculation i am also
> using the XANES and DEBYE cards.
> When using the cfaverage card with ipot 0 for one of the Ni and ipot
> 1 for the rest of the Ni atoms, i get the program running but no output
> xmu.dat file is created.
> If i use ipot 1 or ipot 0 for all the Ni atoms then i get the complains:
>
> XANES:
> no atoms or overlap cards for unic pot 0
> can not calculate potentials, etc.
> (See attached file: feff.inp)
> RDINP-
>
> or
>
> XANES:
> more than one absorbing atom (potential 0)
> only one absorbing atom allowed
> RDINP
>
>
> Could you please tell me how i should use the cfaverage card in the feff8.2
> version?
> I attach the input file.
>
> thanks,
>
> Anna Puig
> Content-type: application/octet-stream;
> 	name="feff.inp"
> Content-Disposition: attachment; filename="feff.inp"
>
> Attachment converted: Talc:feff.inp 7 (????/----) (00057230)


--
Dr. B Gilbert

Visiting Postdoctoral Researcher
Department of Earth and Planetary Sciences
University of California - Berkeley
371 McCone Hall
CA 94720-4767

+1 510 643-9120 (tel:office)
+1 510 643-9980 (fax)