That has been my experience too. Or, in french: Cela a été mon expérience aussi. Anatoly -----Original Message----- From: ifeffit-bounces@millenia.cars.aps.anl.gov [mailto:ifeffit-bounces@millenia.cars.aps.anl.gov]On Behalf Of Matt Newville Sent: Thursday, February 12, 2009 1:11 PM To: bounif; Sam Webb; XAFS Analysis using Ifeffit Subject: Re: [Ifeffit] SIXPACK Hi Bounif, I'm actually not responsible for the PCA analysis part of SixPACK, Sam Webb (the author of SixPACK!) is. And I have to admit I don't know all the details beyond what is documented at: http://www-ssrl.slac.stanford.edu/~swebb/spdocs/sixpack_documentation.htm#PC... I'll try to answer your questions, and hopefully Sam or someone on this list will correct me:
My question is the following one: When I calculate the principals components from the data matrix ( sampls ) I must make a target transformation to obtain the réels compoents which constitutes mixtures. 1-a reels components are obtained by the transformationof the principals components or de target components?
I believe it uses the components found from PCA as the pieces with which to build a spectra that matches the target.
2-How to calculate the concentration of the reels constituents ?
I am not sure how to get those numbers directly. If the target
transformation works to reproduce your spectra, you could always then
try linear combinations to get the sample constituents. I'm not sure
that this is guaranteed to work or give a unique answer (say, if all
of your samples were truly mixtures of multiple components
themselves), but I think that it does normally work in practice with a
small number of components and some of them "nearly" endmembers.
Cheers,
--Matt Newville <newville at cars.uchicago.edu>
2009/2/11 bounif
Dear Newville, With big enjoyment which I writes you,I am a student in paris 12 université,I work on the decomposition of the spectre EXAFS of mixtures at several pure constituents,I test various methods, Among these methods the principals components analisis(PCA).For this, I use the SIXPACK software which you are the author. My question is the following one: When I calculate the principals components from the data matrix ( sampls ) I must make a target transformation to obtain the réels compoents which constitutes mixtures. 1-a reels components are obtained by the transformationof the principals components or de target components? 2-How to calculate the concentration of the reels constituents ? Very cordially
cher Newville, avec grande joie que je vous écrit,je suis un doctorant à luniversité de paris 12,je travail sur la décomposition des spectre EXAFS des mélanges en plusieurs composantes pures,je test différentes méthodes,parmi ces méthodes l'analyse en composantes principale.pour cela j'utilise le logiciel SIXPACK dont vous êtes l'auteur. Ma question est la suivante:quand je calcule les composantes principales (components) à partir de la matrice des donnés (sampls) je dois effectuer un transformation pour obtenir les composés réels qui constitue les mélanges. 1-les composés réel sont obtenus par la transformation des composantes principales où les composantes cible (target ) ? 2-comment calculer la concentration des composantes réel? bien cordialement
Bounif Mohamed
Institut de Chimie et des Matériaux Paris Est (ICMPE) UMR CNRS-Université Paris 12 n° 7182 Equipe SAX 2 à 8 Rue Henri Dunant, 94320 THIAIS France Tel : +33.1.49.78.13.38 Fax : +33.1.49.78.12.03
_______________________________________________ Ifeffit mailing list Ifeffit@millenia.cars.aps.anl.gov http://millenia.cars.aps.anl.gov/mailman/listinfo/ifeffit