[Ifeffit] Re: Novice FEFF question--feel free to post

[Bruce Ravel]

Hi Ryan,

I am taking the liberty of CCing this to the Ifeffit mailing list.  

> I have a novice FEFF question for you guys.
>
>
>
> Assuming one has a good idea of the SO^2 value for a specific element
> either from published literature or analysis of a relevant standard
> reference phase (e.g. 0.85 for Co in Co(OH)2), what is the consensus on
> fixing the SO^2 value for the FEFF simulation and if one does fix the value
> in the feff.inp file (@ 0.85 for this thought experiment) would one still
> fix SO^2 at the same value during the fit (or fix at one)?

I'll assume that you are famliar with the normal caveats of the
assumption of the "chemical transferability" of S02 and that you are
aware that you introduce a source of systematic uncertainty into your
analysis when you make that assumption.  If that's not clear, please
ask so that I or someone else on the list can talk about that.

I can only answer you question by explaining what I prefer.  I cannot
presume to speak for the consensus -- different practitioners
reasonably come to their own conclusions.

I prefer to run feff with S02=1 and all sigma^2 values set to zero.  I
then prefer to use Ieffit (or Artemis or WinXAS or EXAFSPAK or
whatever) to change the parameters of the EXAFS equation from their
nominal values.

Of course, setting S02 to 0.85 in Feff and then setting it to 1 in
Ifeffit is equivalent to setting it to 1 in Feff and 0.85 in Ifeffit.
While multiplication is unambigously commutative, human brains are
usually notoriously poor at disentangling a sequence of arithmatic.

If you: (1) set S02 to something other than 1 in Feff, (2) need to
modify the value when doing analysis in Ifeffit, (3) go on vacation
for two weeks, and finally (4) return home to write a paper on your
analysis --- are you going to remember that you need to multiply your
two values of S02 together when you report on your analysis in the
paper?  Hmmmm.....

Another way of looking at this is to remember what Feff (that is,
Feff6 -- Feff8 adds a few steps) does:
  1. computes potentials and phases
  2. determines the path list
  3. constructs chi(k) for each path in the path list using the
     potentials and phases from part 1
  4. sums up all the paths to make the theoretical chi(k)

If you consider Ifeffit to be a much more interesting version of the
fourth part of Feff, then it should be clear that the better approach
is to use Ifeffit to deal with S02.

HTH,
B



-- 
 Bruce Ravel  ---------------------------------------------- bravel at anl.gov

 Molecular Environmental Science Group, Building 203, Room E-165
 MRCAT, Sector 10, Advanced Photon Source, Building 433, Room B007

 Argonne National Laboratory         phone and voice mail: (1) 630 252 5033
 Argonne IL 60439, USA                                fax: (1) 630 252 9793

 My homepage:    http://cars9.uchicago.edu/~ravel 
 EXAFS software: http://cars9.uchicago.edu/~ravel/software/exafs/