7.8.1 Including Background Refinement

It is often desirable in XAFS analysis to understand how the background absorption function effects or is affected by the structural fitting parameters. Trationally, this has been difficult to do, as background removal and parameter fitting have been completely separated. Though IFEFFIT still separates these procedures, the feffit() command can be used to modify background-like parameters for $\chi$(k) at the same it modifies structural parameters. This option is very easy to add to a fit: simply add the argument do_bkg = true to feffit():

  Iff> feffit(1, chi= data.chi, group = fit, do_bkg=true)

This will automatically add several variables to define a smoothly varying spline $\mu_{0}^{}$(k) (now in k-space instead of E-space) that will be added to the model $\chi$(k) to match your data. The miminum R value used in the fit will be set to 0.0. The number of spline parameters will be determined by rmin, which now takes the role of rbkg in the spline() command) so that the spline will only be ``free enough'' to easily match the low-R portion (ie, those parts below rmin) of the spectra.

The main purpose of this feature is to investigate how the background parameters are correlated with the structural (or rather, traditional) fitting parameters. The additional outputs from using this switch are a set of fitting variables bkg01_01 ...bkg01_NN for NN background variables. The uncertainties in these variables and the correlations between these and the traditional fitting parameters will be available as normal. Note that this feature can greatly increase the number of fitting variables and therefore slow down the fit. In addition to adding the bkg01_NN fitting variables, the do_bkg switch will also cause the output of additional array, fit.kbkg which will contain the additional background function, $\mu_{0}^{}$(k) added to the model in order to match the data.