Introduce Pawley function

[Michael Wedel]

One mode of analysing POLDI data with the original analysis software is to perform a Pawley fit. In this approach, peak profiles are fitted to the data. Unlike in "unconstrained" peak fitting, the peak positions are not fitted individually, instead they are calculated from unit cell parameters.

While this functionality exists in other powder diffraction analysis software, it has to be implemented in Mantid in order to be able to use it for POLDI data. On the other hand, it should be made general enough so that it can be used in connection with other instruments as well.

In this ticket I would like to introduce an implementation example. A design document has been written already, but some points are still open for discussion.

[Michael Wedel]

A design document was added to the documents-repository here:

​https://github.com/mantidproject/documents/blob/3f825dce48120efc1802cfd17fdc1a8a26ec3daa/Design/Design-PawleyFunction.md

[Michael Wedel]

Refs #11043. Added PawleyFunction skeleton

Changeset: 4e79d47ebe56ed0a47ac343518e2afbfcf3d5249

[Michael Wedel]

Refs #11043. Renaming function, adding unit cell parameters

Changeset: 1428773c4a35503a3ba52efc17b857cb52c00071

[Michael Wedel]

Refs #11043. Adding actual PawleyFunction

Changeset: 95581851a106ac9db694364434c9a78ead086c8c

[Michael Wedel]

Refs #11043. Correcting parameter behavior, adding peaks.

Changeset: 492aff6d156bc1f4a38c8bffef85609c2996ad29

[Michael Wedel]

Refs #11043. Implementing function

Changeset: 4ad731913a3c89af32bfbeca7b7215022b754231

[Michael Wedel]

Refs #11043. PawleyFit-skeleton

Changeset: c7cc0cea5bbb0eefbfb35ac957e50a83eed77759

[Michael Wedel]

Refs #11043. Extending peak handling in PawleyFunction

Changeset: 75d80b49011baa5d313258fc15010537c9ac01e7

[Michael Wedel]

Refs #11043. First working version of PawleyFit

Changeset: 65a71b5abf27fe86e417b66bfe2ac8d495da9bfa

[Michael Wedel]

Peaks of FCC lattice with a = 3.6 in d-range [0.7...2.5]

[Michael Wedel]

The algorithm can now be run with the following script:

Load(Filename='PawleyFcc3_6.nxs', OutputWorkspace='PawleyFcc3_6')
PawleyFit(InputWorkspace='PawleyFcc3_6', CrystalSystem='Cubic', InitialCell='3.58 3.58 3.58', MillerIndices='[1,1,1];[2,0,0];[2,2,0];[2,2,2];[3,1,1];[3,3,1];[4,0,0];[4,2,0];[4,2,2]', OutputWorkspace='output')

After this, output contains the input/calculated/difference curves.

[Michael Wedel]

Refs #11043. Additional option.

Changeset: d4330d2487d522d11269d68a7327f9789bf4b5e9

[Michael Wedel]

Refs #11043. Fixing parameter at right place, profile function name

Changeset: 0628fc061434e13c44cc811c391eb99f9854b60e

[Michael Wedel]

Refs #11043. Add setWorkspace to FunctionParameterDecorator

The setWorkspace and setMatrixWorkspace methods were not implemented in FunctionParameterDecorator

Changeset: e8676f9b5480f969a1fa2cfffe2ffc3a1d82d453

[Michael Wedel]

Refs #11043. Adapting PoldiPeakSearch to work with different units.

Changeset: 5db3ba1563a1ef036f674744fdf5e0199c9aaef0

[Michael Wedel]

Refs #11043. Adding more unit tests.

Changeset: d02719ab0f55af1d4251ceff5deff23b7cf0abaf

[Michael Wedel]

Refs #11043. Get reflections from table

Changeset: 7151e2cc576d2898816843813d7cb0545e3e2e65

[Michael Wedel]

Refs #11043. Generating output tables

Changeset: 9452f4e3fa28be1f94d8236fb1bd8e51615c1c30

[Michael Wedel]

Refs #11043. Adding unit tests for PawleyFunction

Changeset: c230d44e7edcc0229c1efd3b2c9e87b8b229b2ee

[Michael Wedel]

Refs #11043. Documentation for PawleyFunction

Changeset: 581e3808f275a58e6036ffb8595ed470293f8fa3

[Michael Wedel]

Refs #11043. PawleyFunction uses Unit

Changeset: f1f22ee4e5f1bf9f4bbc36092971d17c1cc2c6e5

[Michael Wedel]

Refs #11043. PawleyFit uses Unit

Changeset: 77d662a993610c047ff4cf64542cf31a432ecbde

[Michael Wedel]

Refs #11043. Fixing validation problem in PoldiIndexKnownCompound

This was a very annoying problem which was caused by too strict validation.

Changeset: 307851998d525ee5af922077773db6fcf5650558

[Michael Wedel]

Refs #11043. Make unit transformation in PoldiPeakSearch more clear

Changeset: ca17e372098ba1fbdd86c7b9b8ab7e4fb259555f

[Michael Wedel]

Refs #11043. Unit conversion in PawleyFunction more robust

Changeset: 42323ede4b9c5b54d20564fb0921d16e55532088

[Michael Wedel]

Refs #11043. Fixing CodeConventions system test

Changeset: 8bae0e8486334d844ad137afabf1a52cd6b290fa

[Michael Wedel]

Refs #11043. Making PawleyFit more flexible regarding table columns

Changeset: 840401c283320f44cdc5498e7c3e1fa6da782614

[Michael Wedel]

Refs #11043. Adding background to PawleyFit

Changeset: 0d34aeab8084d732dfb1a7aac1256c21b4f0656e

[Michael Wedel]

Refs #11043. Add tests for PawleyFit

Changeset: 2cde6a4b81b782678d839791b5683b6358bb6d2b

[Michael Wedel]

Refs #11043. Checkpointing work

Changeset: f44b9989c650a2267cb7e1d66a35629056dc55bd

[Michael Wedel]

Refs #11043. Adding data file for doc test

Changeset: 097a4b14e18a7e9a00e95f9bcebc25f80790de29

[Michael Wedel]

Refs #11043. Complete docs and doc-test

Changeset: 9c952a5335c0c81ad7e4e765956bc3c3b621b1f3

[Michael Wedel]

Refs #11043. Adding some logging

Changeset: 8d3907a29a0b3e152bf53ce109b3c0b02989e24d

[Michael Wedel]

This is being verified as ​pull request #439.

[Michael Wedel]

Refs #11043. Forgot to add export symbol

Changeset: fb68d0d8a851492e0f5f5fd91f43cfbc23dc3cae

[Michael Wedel]

Refs #11043. Fix lattice parameter typo in doctest

I accidentally inserted an additional 2 into the decimals of the lattice parameter of Si.

Changeset: e0a263c7ab7c7fdd8b8f020ac5c6eaa8aa8552c5

[Michael Wedel]

Jenkins retest this please

[Michael Wedel]

Jenkins retest this please

[Michael Wedel]

Testing information Code review. There is a [design document](​https://github.com/mantidproject/documents/blob/9e0539b5d8c93614de17a7e252d078414a9db861/Design/Design-PawleyFunction.md) that explains the implementation a bit, especially the "Update" section.

In the unit tests for PawleyFunction and PawleyFit there are some basic examples for what the algorithm can do. In the documentation for PawleyFit there is a usage example. The algorithm should work on any MatrixWorkspace which contains a spectrum in terms of d-spacing or momentum transfer - if you know a good example from a powder diffraction beamline that would be something to try.

[Michael Wedel]

The Test SystemTests.ISISDirectInelastic.MERLINReduction failed, but it don't know how that could be caused by these changes. I tried running it locally, but I don't have 16 GB of free memory to run it. I will try again on the build server.

[Michael Wedel]

Jenkins retest this please

[Michael Wedel]

The same test failed again and it failed on other pull request builds as well, so it seems that this has nothing to do with the changes.

[Federico Montesino Pouzols]

Jenkins, retest this please

[Wenduo Zhou]

In testing script, you used a property named 'MillerIndicies' to input peaks of algorithm PawleyFit. But it does not exist in the final code. Did you removed it later?

[Michael Wedel]

Yes, sorry for not updating the script, I forgot about that. It turned out that some reasonable guesses for peak height/fwhm were required, so the list of HKLs was not enough to provide in some cases. That's why I made the TableWorkspace mandatory and removed the MillerIndices property.

So in case one does not want to use the POLDI algorithms, a TableWorkspace would have to be obtained some other way, I guess it could be done through Python.

[Wenduo Zhou]

The structure of the codes is clear. All tests are passed. Failed pull request is due to other irrelevant reason.

I tried to fit the attached silicon data and the result is very good.

The ticket is then closed.

[Michael Wedel]

Refs #11043. Added reduced chi square to output

Added this to the algorithms documentation.

Changeset: 08d3845eaeeadde29389e917ec309812bbd00131

[Michael Wedel]

I've thought about running PoldiPeakSearch inside the algorithm when only calculated peaks are supplied, but the algorithm also has some parameters and the default ones don't always work, so I think for now it's better to keep this out of the algorithm.

[Wenduo Zhou]

Everything looks find. Ticket is thus closed.

[Wenduo Zhou]

Merge pull request #439 from mantidproject/11043_introduce_pawley_function

Introduce Pawley function and algorithm for using it

Full changeset: ​3a622047da3cc95a538cc5bbe869a845ab293c0f

[Stuart Campbell]

This ticket has been transferred to github issue ​11881