[Ifeffit] Artemis and FEFF interaction in fit process

[Bruce Ravel]

Artemis has a parameter, feff-->executable, which identifies the Feff
executable.  Artemis runs this by a system call to create the
potentials and phase shifts given the input structure.

Artemis then runs its own version of the pathfinder.

The results of those two calculations are stored in the project file.

In an instance of Artemis, the Feff executable is run as needed to
make the feffNNNN.dat files.  That is, the first time you make a plot
or a fit (there are a few other actions that will also trigger this),
Artemis will run the Feff executable to make the path that is needed.
Thus each path plotted or used in a fit requires a system call.

The advantage of this is that project files remain relatively small
because no data is saved for any path, only the metadata required to
reproduce the path.  This works because making a path is fast enough,
certainly it is on the time scale that one spends staring at Artemis.

All of that works with Feff6L.  Given the limited amount of time that
I can possibly spend on Demeter and the large number of things that
need doing, making all that work with all other versions of Feff is
not a priority for me.

You could try pointing Demeter at whatever version of Feff you are
using and seeing how it goes.  I probably won't be in a position to do
much if/when things don't go as hoped.

That said, I am committed to making Artemis work with this version of
Feff 8.5: https://github.com/xraypy/feff85exafs.  In fact, I've been
working on that for the last few weeks.  There's a lot to do, so I
don't have an ETA for when that version of Feff will be integrated
into Artemis.

If you have a compelling need to use some version of Feff other than
the one we get to redistribute, then Artemis may not be the right tool
for you.

Of course  there are things that you or anyone could do to help this
along.  We would welcome help with development of Feff 8.5 for EXAFS.
I would welcome contributions to Demeter.  Matt would be happy to see
more people working with him on Larch.

B


On 08/19/2014 02:07 PM, Nate Turner wrote:
> Hello All,
>
> I have a question about Artemis' interaction with FEFF.
>
> I know you can either run FEFF6L in Artemis to populate a paths list or
> externally create the FEFFnnnn.dat files in FEFF and import them into
> Artemis. We currently do the later using FEFF9, the use of the SCF card
> gives a better Eo initial guess and our compound has fractional
> occupancy/disorder that requires us to do some altering of the atoms
> list after its populated in ATOMS, which we find best to do outside of
> Artemis.
>
> My question is once these external FEFF9 paths are brought in does
> Artemis use FEFF6L in the fit process or is FEFF in Artemis only to
> populate the paths list?
>
> If Artemis doesn't use FEFF in the fit process then there's no issue but
> while I've read the mailing list and seen that many do not believe there
> is any improvement in the EXAFS results from FEFF6L to FEFF9, I'd rather
> be consistent in which version of FEFF we use throughout the analysis. I
> would have Artemis calling to FEFF9 instead of FEFF6L if in fact Artemis
> does use FEFF in the fit.
>
> The reason I haven't gone ahead with having Artemis call to FEFF9 making
> this a non issue is I'm not sure how. I know you can direct Artemis to
> whatever version of FEFF you want under preferences, but then I'm not
> sure what file I need to direct to specifically for FEFF9.
>
> So to summarize,
>
> Does Artemis use FEFF in the fitting process or is its only function in
> Artemis to create the paths for the fit?
> If it does use FEFF in the fitting, how do you/is it possible to have
> Artemis call to FEFF9 instead of FEFF6L?
>
> Thanks all for your time,
>
> Nate Turner
>
>
> _______________________________________________
> Ifeffit mailing list
> Ifeffit at millenia.cars.aps.anl.gov
> http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit
>


-- 
  Bruce Ravel  ------------------------------------ bravel at bnl.gov

  National Institute of Standards and Technology
  Synchrotron Science Group at NSLS --- Beamlines U7A, X24A, X23A2
  Building 535A
  Upton NY, 11973

  Homepage:    http://bruceravel.github.io/home/
  Software:    https://github.com/bruceravel
  Demeter:     http://bruceravel.github.io/demeter/