As part of the Mantid parsing we also offer limited support for Mantid's MultiFit functionality.
Here we outline how to use Mantid's MultiFit with FitBenchmarking, in which some options differ from the standard native
.
Warning
Due to the way Mantid uses ties (a central feature of MultiFit), MultiFit problems can only be used with Mantid minimizers.
In this format, data is separated from the function. This allows running the same dataset against multiple different models to assess which is the most appropriate.
An example of a multifit problem is:
../../../../examples/benchmark_problems/MultiFit/MUSR62260.txt
Below we outline the differences between this and the native
.
- software
Must be Mantid.
- name
As in
native
.- description
As in
native
.- input_file
As in
native
, but you must pass in a list of data files (see above example).- function
As in
native
.When fitting, this function will be used for each of the
input_files
given simultaneously.- ties
This entry is used to define global variables by tieing a variable across input files.
Each string in the list should reference a parameter in the function using Mantid's convention of
f<i>.<name>
wherei
is the position of the function in the function string, andname
is the global parameter.For example to run a fit which has a shared background and peak height, the function and ties fields might look like:
function='name=LinearBackground, A0=0, A1=0; name=Gaussian, Height=0.01, PeakCentre=0.00037, Sigma=1e-05' ties=['f0.A0', 'f0.A1', 'f1.Height']
- fit_ranges
As in
native
.