[Ifeffit] Scanning of "Potential Surfaces" in EXAFS

[Bruce Ravel]

On Tuesday 08 July 2003 07:24 am, Norbert Weiher wrote:
> I have been guessing for a long time of doing an algorithm of doing some 
kind 
> of potential surface scanning when doing an EXAFS fit. This procedure has 
> been known in e.g. ab initio codes like GAUSSIAN for a long time and can be 
> used to check if you are really in a global minimum on the potential 
surface. 
> As EXAFS analysis is the ultimate search for a global minimum in the 
> parameter space, but you never know if you really end up there, I was 
> planning to do such kind of investigations.
> 
> However, before I start off with wild coding :) I want to have more opinions 
> on this topic. Here are my main points speaking for this kind of algorithm:
> 
> 1) Computer power is quite fast now - and ifeffit is also really fast in 
> computing the fit quality if you do not guess any variable (which you don't 
> need in this case as you vary the parameters by your own).
> 
> 2) In cases where you would expect large correlations between certain 
> variables (e.g. when you have overlapping shells at nearly the same 
> distance), one could systematically investigate the influence of small 
> changes in the parameter space on the fit.

This is an excellent idea and, in fact, is among my long-range plans
for Artemis.

You might take a look at Biochemistry 35 (1996) pp. 9014-9023 and
other papers by the same authors for an example of what I have in
mind.  In that example, they raster through a plane of two variables
and fit the remaining variables.  The end product is a contour plot of
chi-square in the plane of the two rastered variables.

Another possibility would be to start at Ifeffit's best fit and raster
by hand through as many variables as you like, saving a matrix where
the abscissae are the parameters and the values of the matrix elements
are chi-square.  You could then take cuts through this matrix to
explore the multi-variate parameter space.

In any case, it would not be a difficult programming chore to use
perl, python, or C to walk through the parameters and organize the
output and to use ifeffit to do the heavy lifting.

Let us know how it works out for you, Norbert.
B



-- 
 Bruce Ravel  ----------------------------------- ravel at phys.washington.edu
 Code 6134, Building 3, Room 222
 Naval Research Laboratory                          phone: (1) 202 767 5947
 Washington DC 20375, USA                             fax: (1) 202 767 1697

 NRL Synchrotron Radiation Consortium (NRL-SRC)
 Beamlines X11a, X11b, X23b, X24c, U4b
 National Synchrotron Light Source
 Brookhaven National Laboratory, Upton, NY 11973

 My homepage:    http://feff.phys.washington.edu/~ravel 
 EXAFS software: http://feff.phys.washington.edu/~ravel/software/exafs/