[IDA] FuryFit - Fixes To Passed Fitting Function

[Peter Parker]

• The fitting function passed thru for single exp is:

  name=LinearBackground,A0=0,A1=0,ties=(A1=0);
  name=UserFunction,Formula=Intensity*exp(-(x/Tau)),Intensity=0.976953,Tau=0.0289162;
  ties=(f1.Intensity=1-f0.A0)
  

  A single spectrum fit uses the fitted parameters for Intensity & Tau, bit not for A0 of Linear Background.

• For stretched exp it is:

  name=LinearBackground,A0=0,A1=0,ties=(A1=0);
  name=UserFunction,Formula=Intensity*exp(-(x/Tau)^Beta),Intensity=0.981297,Tau=0.0250165,Beta=0.841352;
  ties=(f1.Intensity=1-f0.A0)
  

Again A0 not passed through.

• There should also be a constraint on Beta so that 0 < Beta < 1.

[Nick Draper]

Moved at end of release 2.1

[Nick Draper]

Moved at the end of release 2.2

[Nick Draper]

Moved to milestone 2.4

[Nick Draper]

Moved at the code freeze for release 2.4

[Nick Draper]

Moved to r2.6 at the end of r2.5

[Nick Draper]

Moved to backlog at the code freeze for R2.6

[Samuel Jackson]

Adding constraint to Beta parameter.

Refs #5008

Changeset: 5fa677e7ba28f87258837a75a936443bd2d3db19

[Samuel Jackson]

Found better function for constaints.

Refs #5008

Changeset: e7acf689825ce6acc5482fe124fb30820303a463

[Samuel Jackson]

Fixed A0 parameter not being passed correctly.

Refs #5008

Changeset: 74746ee22a55d0f5ce5ccd10c50df27a9c7315ec

[Samuel Jackson]

To Test:

Run a FuryFit and make sure that the A0 parameter is set to something greater than 0. Check that this works with both a single and sequential fit and with a regular and stretched lorenzian.

Also check that a stretched lorenzian only takes beta value between 0 and 1.

[Arturs Bekasovs]

Merge remote-tracking branch 'origin/feature/5008_furfit_param_passing'

[Arturs Bekasovs]

Code changes are pretty straightforward and perfectly appropriate.

Testing a FuryFit on irs26176_graphite002_red.nxs showed that A0 is indeed set to the value greater than 0 using all the options of the dialog. In some cases, when starting parameters are not sensible, it can happen that A0 is less than 0 and fitted line is far from ideal, but repeating fit improves the situation.

During the testing Beta value was never out of the [0;1] interval.

[Stuart Campbell]

This ticket has been transferred to github issue ​5854