WAVELET TRANSFORM OF XAFS SPECTRA
MUÑOZ Manuel - FARGES François - ARGOUL Pierre
Université de Marne la Vallée - CNRS FRE 2455 - Ecole Nationale Supérieure
des Ponts et Chaussées
(Polytechnicum de Marne la Vallée, France)
FAQ (foire aux questions)
:
A new approach of EXAFS spectra analysis
was developped: the Continuous Cauchy Wavelet Transform (CCWT). The
method provides an accurate "time-frequency" decomposition
of the analyzed signal. Applied to EXAFS spectroscopy,
the signal is decomposed both, in the wavevector (k) and in the distance
(uncorrected for phase-shifts; R')
dimensions :

Qualitative interpretations of the EXAFS spectra are directly
based on the CCWT modulus, and may help, among others, next-nearest neighbors
identification.
References:
Details concerning the numerical
description of CCWT and illustrations of the method applied to EXAFS spectra
are provided
in two papers :
• Muñoz M., Argoul P. and Farges F. (2003) Continuous Cauchy wavelet transform analyses of EXAFS spectra: a qualitative approach. American Mineralogist vol. 88, pp. 694-700 [pdf]
• Muñoz M., Farges F.and Argoul P. (2003) Continuous Cauchy wavelet transform analyses of XANES and EXAFS spectra. Physica Scripta (proceeding #5220 of the XAFS-12 meeting, Malmö-Lund, Sweden, in press) [pdf]
Source code:
The Matlab/Octave script file
can be freely downloaded for free here.
The script was positively tested on Matlab 5.2/MacOS9, Matlab 5.3/Win2k and Matlab 6.5/MacOSX (under both X11 or OroborOSX).
Clever non-Matlab users: OCTAVE (under the terms of the GNU General Public License ( GPL ) as published by the Free Software Foundation) is your free alternative to run this script (+ gnuplot).
The computed *.mat files can be reloaded in plot mode (mode = 2) to save CPU time.
Altough binaries, *.mat files are cross-platform compatible (tested Win2K —> MacOSX and MacOS9 —> MacOSX).
Contact :
Any question, remark or suggestion
? contact us at nospam@univ-mlv.fr (BUT
replace "nospam" by either "munoz" OR "farges" ;
this is done to short-cut scanners for spam mail - sorry for this !).