On Friday, June 08, 2012 11:18:25 AM shbone@berkeley.edu wrote:
I've tried using the spline function in Sixpack (and Athena), but I don't have enough options to modify the spline. It seems to me that I need to use a very rigid spline (for instance, one with only 3 or so knots), and that the options in Sixpack and Athena don't allow for this.
Sharon, That is not the problem that the spline in Ifeffit (and therefore in Athena and Sixpack) is designed to solve. The purpose of Ifeffit's spline is to isolate the chi(k) from mu(E) and it only aims to do so reliably above 2 or so inverse Angstroms. There are reasons for that: 1. There are information theoretic limits to how much freedom the chi(k) isolation spline should have. In general, it cannot have enough freedom to reliably go down the edge -- which is the fastest changing part of the data. 2. We typically do our EXAFS analysis starting a few inv. Ang above k=0, so in practice Ifeffit's spline does what we want. 3. The theory for EXAFS analysis is probably unreliable below 2 or 3 inv Ang, but quite reliable above that. So if the spline cannot isolate chi(k) below 2 or 3 and the theory cannot be trusted below 2 or 3, nothing is lost for EXAFS analysis. So, I think your problem is that you are trying to use the wrong tool for the job. You want to treat your XANES data as a generic peak fitting problem. There are many, many options for that. You don't really need to rely upon software written explicitly for XAS data, although many of them do some kind of peak fitting. Athena does, for example, using thinsg like arctangents and gaussians, although it is perhaps not the strongest part of that program. A general peak fitting problem can be dealt with in something like Matlab or Mathematica. It can probably be done reasonably well in Excel. Whatever.... Step-like functions (like an arctangent) are often used to model the edge. I suppose you could use a spline for that, but it may be challenging to give the spline enough freedom to follow the step-like shape but not enough freedom that it wants to follow the peaks in the egde. That's the nice thing about an arctangent or an error function -- purely a broadened step. HTH, B -- Bruce Ravel ------------------------------------ bravel@bnl.gov National Institute of Standards and Technology Synchrotron Methods Group at NSLS --- Beamlines U7A, X24A, X23A2 Building 535A Upton NY, 11973 Homepage: http://xafs.org/BruceRavel Software: https://github.com/bruceravel