Room for performance improvements in "sequential fit" with Resolution function

[Michael Whitty]

A common use case for the indirect group is to do a convolution fit on their data, which is a lorentzian and a resolution function.

To facilitate this, I made changes to the resolution function to allow for a nexus file as input - so that it can be more easily generated and managed within Mantid.

When this fit function is used within the PlotPeakByLogValue algorithm, it will re-load the nexus file for each spectra.

Is there a way we can move some element of this up to the controlling algorithm so that the sub-algorithm Fit does not feel the need to do this?

Also - would be desirable to be able to have the resolution taken as being the same workspace index as the current input being fitted.

[Nick Draper]

Is this possible without breaking the architecture?

[Nick Draper]

Bulk move of tickets at the end of iteration 28

[Nick Draper]

Accepted and assigned tickets moved at iteration 29 code freeze

[Nick Draper]

Bulk move of tickets to iteration 31 at the iteration 30 code freeze

[Nick Draper]

Moved to iteration 33 at iteration 32 code freeze

[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

[Roman Tolchenov]

Re #2389. Added workspace index attribute to TabulatedFunction.

Changed it to be able to load any loadable file. Resolution inherits from TabulatedFunction.

Changeset: eeaa2288165d2eeaddee61ad0324aca2b245b98a

[Roman Tolchenov]

Re #2389. Correction to handling empty string attributes.

Changeset: 5c5c7e3a12d1abebe2f851db9670dd65fa5bb10f

[Roman Tolchenov]

Re #2389. Added StringEditorFactory to QtPropertyBrowser.

The created editors update properties only when user finishes editing.

Changeset: f9a0bf818613f1516220bf982c3e0637099375e2

[Roman Tolchenov]

Re #2389. Added parameter to pass workspace index to fitting function.

Changeset: 5b2bff6e368d945399931a40f2baf57b76266a6a

[Roman Tolchenov]

Re #2389. Added unit tests for new property of PlotPeakByLogValue.

Changeset: f5513c1a5a6b17d1a45f3cceb3bfd8d838cd96a1

[Roman Tolchenov]

Re #2389. Added a checkbox to sequential fit dialog to pass ws index.

Also some clean-up of Convolution code.

Changeset: 5a66f8566f7152fee36a8439ac791121f6d01b70

[Roman Tolchenov]

Re #2389. Changed Resolution class inheritance.

Changeset: b885c42fc799eb461a43b1f81f28936d058b28ba

[Roman Tolchenov]

To tester:

• Download the attached files and run the script.
• In the result table workspace each row contains parameters of the function fitted to the corresponding spectrum in data3.
• When PassWSIndexToFunction is True all parameters must be different. Changing WorkspaceIndex in Resolution doesn't change the result.
• When PassWSIndexToFunction is False parameters in the same column must be the same. Changing WorkspaceIndex in Resolution changes the result.
• The same functionality must be available via FitPropertyBrowser.

[Arturs Bekasovs]

All works as described when running using the script provided.

However, when trying to do the same thing using the FitPropertyBrowser - Mantid crashes. Happens when trying both Fit and Sequential fit, using parameters from the script.py.

[Roman Tolchenov]

Resolution mustn't call TabulatedFunction::functionDeriv1D method.

Re #2389

Changeset: d73b7c8f10688f2a3c0d12cafc979e8cd92234d3

[Samuel Jackson]

Merge remote-tracking branch 'origin/feature/2389_resolution_sequential_fitting'

Full changeset: ​9cc47b21f1553d5948b3887878fe95a6ffb42e30

[Stuart Campbell]

This ticket has been transferred to github issue ​3236