[Ifeffit] Some questions related to IFEFFIT_ ARTEMIS computer package program

[Matt Newville]

Hi Nick,

> On Tue, Mar 29, 2011 at 9:15 AM, Nicolae L. Aldea <naldea at itim-cj.ro> wrote:
> Dear users,
>
>
> My new questions are:
>
> 1) Should I introduce in GDS degen parameter for fitting beside amp, enot,
> delr and ss? What happened when I want to obtain coordination number (CN)
> associated to some coordination shell. I thank that there is a difference
> some time between CN and degen. In some cases they are the same. Many
> journals as you know report CN.   For a coordination shell I think that we
> have some times more paths, is it rue or false? Does dgen associate only
> for a path ?

Mathematically, the amplitude applied to a path is the product of the
degeneracy (Degen) from the the Feff calculation for that path with
adjustable Amp parameter.  Amp is often used to represent S02, but can
be used for more complicated analysis as well.  Generally speaking,
S02 does not depend on the path geometry, while the coordination
number does.

A common approach for varying the coordination number is to set Degen
to 1 (so that the degeneracy from the feff.dat file is ignored),
figure out what you think S02 should be set Amp to S02 * CN.

> 2) I could to view the feffXXXX.dat in according with your indication in
> own folder of computer package program c:\Document
> andSettings\USERNAME\Application Data\horae\stash, but can I save
> directly them? I did not find a direct bottom for saving them,   is it
> true or I did   not see it?

I'm sorry, but I don't understand "direct bottom for saving them".  Do
you want to save the feff.dat file for some other purpose?

> 3) The exafs signal is a real function from math. point of view. So,
> FT[chi(k)]= real_RDF_even(r)+i*RDF_odd(r). Then
> IFT[real_RDF_even(r)+i*RDF_odd(r)] = chi_real(k)+ i*zero, where
> i=sqrt(-1), is it true ? In affirmative case why you have in ARTEMIS
> plot region the   both components ? Is not the imaginary part zero ? from
> math p. v. the module of chi(r) is a symmetrical signal. Even, we take
> into account only a region [rmin:rmax] for RDF(r) we obtain for   IFT
> of RDF(r) only a real signal.

I believe that it is not correct to say that a band-limited inverse
Fourier transform will lead to a real function.

In addition to asserting that the measured chi(k) is a real function,
we also assert that chi(R) is zero below R=0.  I believe that,
strictly speaking, these are inconsistent, and that a real function
must have negative frequency components (perhaps someone can correct
be on this if it's wrong or simplistic).  Generally, we don't like to
think about distances smaller than 0.

Though the measured chi(k) is real, any theoretical description of
chi(k) will make it complex, and assert that we sample only part of it
(often referred to as the imaginary part of the theoretical chi(k),
leading to other ambiguities).

As a result of insisting that chi(R) = 0 for R<0, the inverse
transform (chi(q)) has both real and imaginary components, though
these are not independent.


--Matt