Create implementation of IFunction that uses workspace index

[Michael Wedel]

For fitting POLDI data an implementation of IFunction is required that computes function values while taking into account information associated to the spectrum it is applied to (each spectrum of a MatrixWorkspace is a domain).

I am not sure whether it should inherit from IFunction1D, even though it is in fact a kind of 1D-function. But I don't see any meaningful implementation of function1D and functionDeriv1D, since there is no way these methods can "know" the domain type.

On the other hand I am not sure if the fitting framework will be able to deal with a 1D-function that does not implement IFunction1D.

Any thoughts on this?

[Owen Arnold]

cc'd Roman so he can track what's going on here.

[Michael Wedel]

Refs #9445. Added PoldiInstrumentAdapter

This is needed in more than one place and bundling it in one place will make it easier to replace later.

Changeset: 51db44932ea77935e418f2729ecf399f3c80d388

[Michael Wedel]

Refs #9445. Added PoldiSpectrumDomainFunction class

Changeset: af76395d07bedf2364b7979dde57cdf43f3c572d

[Michael Wedel]

Refs #9445. Created 2D-spectrum calculation test

Changeset: 39124fbec5ac553ef0599708fe72c04163943120

[Michael Wedel]

Refs #9445. Added missing typedef

Missing typedef did not let the code compile

Changeset: 82623d2a11ae3b8b4da55dc51780477e447280e1

[Roman Tolchenov]

Re #9445. Corrected the multi-domain Fit call.

Changeset: ac83d9250ba56c8c89c255ff9391e9db1850c669

[Michael Wedel]

Refs #9445. Added PoldiSpectrumDomainTest.h to CMakeLists

Changeset: bb6981234acb8fcb57c5c124529636a48742f944

[Michael Wedel]

Refs #9445. PoldiPeakCollection now holds peak intensity type

A PoldiPeakCollection now stores information about the type of intensity information assigned to the contained peaks. Some parts need to work with the peak maxima, others need the area. This information is now put into the workspace logs and processed.

Changeset: 4f5f6eab4fc2141e1ce308fcad86c2fb4a8663d3

[Michael Wedel]

Refs #9445. PoldiCalculateSpectrum2D processes POLDI peaks

Instead of the fixed example peak, a peak table is now loaded and the corresponding functions are constructed.

Changeset: c7c8f284c93e34d908288a8befeaf4fcd9dfa783

[Michael Wedel]

Refs #9445. Corrected calculation of deltaT

Changeset: 4e1011f9eaf8ea4db8734d5b408bf4c15bcf090e

[Michael Wedel]

Refs #9445. Added clone method to PoldiPeak

Also added equality operator to MillerIndices, since the lack of it was failing the test for PoldiPeak::clone.

Changeset: 16fe42876f886528be123baaebc0349dda628305

[Michael Wedel]

Refs #9445. PoldiPeakCollection stores profile information

PoldiPeakCollection can now store information about which profile function has been used to fit the peak parameters. This is required for possible peak integration in PoldiCalculateSpectrum2D.

Changeset: 90e6ea498b8fafa051c47f701d1c1e1fe820ff94

[Michael Wedel]

Refs #9445. Added profile function check to PoldiPeakCollection

Changeset: 7ef6c6f72629551273da5dc194c2c1dda3b9d10f

[Michael Wedel]

Refs #9445. Extracted time conversion from PoldiSpectrumDomainFunction

Changeset: 2eda04a872490935e06e811f7ae63bae635d82ea

[Michael Wedel]

Refs #9445. Added PeakFunctionIntegrator

PeakFunctionIntegrator takes an IPeakFunction and integrates it numerically using GSL routines.

Changeset: 5f4146dace472d3501c163d95b6c0c36a7b57398

[Michael Wedel]

Refs #9445. Fixed NaN-problem in PoldiTimeTransformer

PoldiTimeTransformer::detectorElementIntensity produced NaN when a wavelength outside the spectrum boundaries was calculated, because that would be negative.

Changeset: efbd26b59be2de72e100bb4630c67a6e36eff1fa

[Michael Wedel]

Refs #9445. Added total intensity calculation to PoldiTimeTransformer

Changeset: 124e06c62755cc6e9cd8745cd70c879a2510a557

[Michael Wedel]

Refs #9445. PoldiCalculateSpectrum2D now integrates peak profiles

Changeset: c7798830e48ac6274d8fc778a728b340b77d3841

[Michael Wedel]

Refs #9445. PoldiFitPeaks1D stores profile information in table

Changeset: 2cb03b251a265da5616da9f5d3d99820bda5b0db

[Michael Wedel]

Refs #9445. Added Poldi2DFunction.

This function inherits from CompositeFunction to accumulate multiple functions and from IFunction1DSpectrum, so the correct domain creator is used in Fit.

Changeset: 92564e73ebc2ba1b2ebd90bd246a2be70ddb9a54

[Michael Wedel]

Refs #9445. Using Poldi2DFunction in spectrum calculation

Changeset: 2e7352799c8569d29e7b349233a4e65b58fed710

[Michael Wedel]

Refs #9445. Added null pointer check to PoldiTimeTransformer

Changeset: c1aec4c57658fa2de6f0ec2c4c9c2676616edeb2

[Michael Wedel]

Refs #9445. Cleaned up PoldiInstrumentAdapter.

Added unit tests for chopper speed extraction functors.

Changeset: 6767d05baa6970ffa3d6db66d1f92557592e0acd

[Michael Wedel]

Refs #9445. Cleaned up Poldi2DFunction.

Added documentation and unit tests.

Changeset: 12e90797e07d157baa4a244bd65267671303a94c

[Michael Wedel]

Re #9445. Integrated PoldiInstrumentAdapter into PoldiAutoCorrelation

Changeset: 951f296d2b1cbc5ba9e68c8f685a0d2fa05da99f

[Michael Wedel]

Refs #9445. Removed using namespace from PoldiMockInstrumentHelpers.h

Had to adjust two unit tests to use the namespaces instead.

Changeset: ec947a67cf1c0c612860af114a0b6f01695d69ad

[Michael Wedel]

Refs #9445. Added clone method to PoldiPeakCollection

Changeset: 0448cd9ec1d3b43d59c520b222eb0fc576f704cd

[Michael Wedel]

Refs #9445. Cleaned up PoldiSpectrumDomainFunction.

Changeset: b4dcfb25e256b096576345ea049dbf6ea22d9dc2

[Michael Wedel]

Refs #9445. Refactored PoldiCalculateSpectrum2D.

Most parts of the algorithm are tested by the unit tests now. The actual spectrum calculation will be done in a system test, as instrument definitions etc. are required for this.

Changeset: 33232934c13174d2dd52f79739fcc947a84a6178

[Michael Wedel]

Refs #9445. Fixed comments in PoldiCalculateSpectrum2D

Changeset: 70314183c31f27c6bb4d3fcd7670b07495d541a9

[Michael Wedel]

Refs #9445. Added documentation with usage example.

Changeset: c0ac991de289d577684ccdf5fcc790ae1122d9af

[Michael Wedel]

Refs #9445. Integrate detector efficiency in IDF

Changeset: d051660674af7967e8d20126bee312aacc4cf5cd

[Michael Wedel]

Refs #9445. Added optional profile function parameter for integration.

Changeset: fc88a1214af52e068381c690c2281461e01cf51b

[Michael Wedel]

Refs #9445. Added system test for PoldiCalculateSpectrum2D

The test uses a spectrum calculated with the original analysis software.

Changeset: dc5e7fbcb1933d14cb6e0bce630e085ebb1935fc

[Michael Wedel]

Refs #9445. Casting GSL enums to int in test

Changeset: 46647d4239229c14532e94797a22bf486db05d22

[Michael Wedel]

Refs #9445. Adding export symbol to wrapper function

Changeset: 19fb420a99e71c67cf97e3a501357f9448b91f71

[Michael Wedel]

Testing information:

This ticket looks quite big, but essentially it adds only one algorithm: PoldiCalculateSpectrum2D. How to use it is documented in the wiki-page. Running the usage example should result in a spectrum as shown there (with "color fill plot").

There is a system test making sure that the calculated values do not deviate too much from the results produced by the existing analysis software.

Remark/question:

There is one class in PoldiUtilities which could potentially be useful for others as well, PeakFunctionIntegrator. As the name indicates, it integrates an IPeakFunction. If I had to guess, I would put it into CurveFitting.

[Michael Wedel]

Refs #9445. Fixing some style problems

Changeset: dc06bbb7a3b2fc74d50dcaa5f224d45e26f4965d

[Michael Wedel]

Refs #9445. Fixing more style problems.

Changeset: 7a7daa5d0accabe65a4022cb4afcc7f12291d793

[Michael Wedel]

Refs #9445. Peak function now uses time units instead of d.

Changeset: 368923935c0a619eff46dfc990d0e33b592f1734

[Michael Wedel]

Revert "Refs #9445. Peak function now uses time units instead of d."

This reverts commit 368923935c0a619eff46dfc990d0e33b592f1734.

Changeset: 485a9890ad16750c97225687da8e9684cce0d398

[Michael Wedel]

Sorry about the confusion with the last two commits. I thought the change to time-based parameters was necessary, which turned out to be wrong after some detailed testing.

[Michael Wedel]

I am removing the block from #9497, because in further going tests when actually performing a fit (which is the intention in the end), this did not seem to cause problems.

[Roman Tolchenov]

Merge remote-tracking branch 'origin/feature/9445_poldi_spectrum_domain_function'

Full changeset: ​97eb107c9250af5b12e54a7b9b4222f15eeabefc

[Michael Wedel]

Merge branch 'feature/9445_systemtest_for_poldi_2d_spectrum' into feature/10074_merge_poldi_system_tests

Full changeset: ​ad13d60fcd3acf298fae5b2ff6a43918a489120d

[Martyn Gigg]

Merge branch 'feature/9445_systemtest_for_poldi_2d_spectrum'

Full changeset: ​bacee637856abc704ed9c088ab43b0d788eb0dca

[Stuart Campbell]

This ticket has been transferred to github issue ​10288