Intelligent Fitting¶

Fitting is the modelling of data where parameters of the model are allowed to vary during a fitting process until the model and data agree better according to some cost function.

In summary the Mantid fitting provides

General fitting capabilities
Neutron and Muon intelligent fitting tools, which makes use of additional information about the data, such as instrument geometry and log value information
Easy expandability

Here a more advanced aspect of Mantid fitting is presented. We will fit an asymmetric peak from a GEM data set with the Ikeda-Carpenter function on a LinearBackground.

Let’s load GEM63437_focussed.nxs file and plot GEM63437_focussed_3 workspace. Zoom into the region 4300 - 4900 microseconds.

Try to fit it with a Gaussian (plus LinearBackground):

Not a very good job. The Ikeda-Carpenter function is a better choice here. But this is a very difficult function to work with. It requires very good initial parameter values for the fit to converge. The Mantid approach is to use the pre-set values which are defined on a per-instrument basis. For example, when the Ikeda-Carpenter is used with GEM data the fitting tools automatically find and set the appropriate initial values. This results in a good fit.