[Ifeffit] Results Reporting

[Bruce Ravel]

On Thursday, April 05, 2012 12:59:55 am John Farell wrote:
> Hi..
> 
> Can you please explain me how to report the errors of EXAFS parameters like
> bond length, Co - ordination number, disorder etc. (Or how to obtain those
> errors ??)

This topic is discussed at length in the standard online resources for
EXAFS and for our software.

Ifeffit's error reporting is reasonably well explained in its
documentation.  See http://cars9.uchicago.edu/ifeffit/Documentation

Artemis reports all of Ifeffit's error reporting in the log file of
each fit.

Ifeffit uses a Levenberg-Marquardt, non-linear, least-squares
minimization.  Error bars and correlations are taken in the standard
manner form the covariance matirx, with the caveat that the error bars
are scaled assuming that the fit was good and that the measurement
uncertainty, epsilon, was incorrectly evaluated.  Thus, if you
actually like the fit, the error bars are scaled to the correct size.

I talk about all this at some length in my "Advanced Topics in EXAFS
Analysis" talk, atea.pdf at
https://github.com/bruceravel/XAS-Education/downloads


I am not really certain what you mean by "how to report" them.  I
usually say something like "<best fit> +/- <uncertainty>" for any
number I report.  If you use a lot of constraints in a fit, you will
need to do error propagation by hand -- neither Ifeffit nor Artemis
attempt to do error propagation.

B


-- 

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

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

 My homepage:    http://xafs.org/BruceRavel
 EXAFS software: http://cars9.uchicago.edu/ifeffit/Demeter