vision fitting of the elastic line

[Jose Borreguero]

The problem: for every pixel in a tube, the elastic line peaks at a slightly different time-of-flight. The lack of superimposition poses a problem. If we simply add the intensities of all the pixels in the tube, we’ll end up with a smeared elastic line, and it will be difficult to extricate the inelastic signal from the total spectrum.

The origin of the shift in the peak center with pixel ID is mainly due of a geometrical effect of the disposition of the analyzer crystals. the finite dimensions of the single flat analyzer crystal produce a distribution of Bragg angles, θ1 to θ2. The particular [θ1, θ2] range is different for every crystal, hence the slight difference in tof for each crystal.

We will fit the peak found in the spectrum of each analyzer crystal with the sum of two Gaussians and a background. However, the intensities of the two gaussians are constrained to obey a quadratic equation across pixel ID i in a single tube (remember there are 128 pixels, and a fraction of them are masked)

Fitting strategy: We fit a whole tube, made up of at most 128 pixels. Each pixel may have its own FunctionDomain1D because the spectrum of each pixel may have different values for the tof bins. If we do rebinning then the same FunctionDomain1D could be used by all pixels (t-coordinate). However, the current implementation of MultiDomainFunction requires different domains, not a common one.

We will create a Mantid algorithm with a MultiDomainFunction as its core engine for fitting. the CompositeFunction will be made up of as many functions as spectra to fit.

Details in the attached file

[Jose Borreguero]

details of the fitting

[Jose Borreguero]

Refs #5770 files for singlefunction

Changeset: b01cd5d9c3eda41a6a6239b29ca4bb81a8c92002

[Jose Borreguero]

Refs #5770 small changes

Changeset: acb245ca4617f7efb0e417606e6990ee63a10b12

[Russell Taylor]

Re #5770. Clear compiler warnings.

Changeset: b98b08455c860ae31beab7d14e3fddd17e0c9d73

[Jose Borreguero]

Refs #5770 Removed the C++ code and wrote a Python script instead

modified: CMakeLists.txt deleted: VisionCalibrationMultiplePeak.h modified: BackToBackExponential.cpp deleted: VisionCalibrationMultiplePeak.cpp

Changeset: 4c2873444b358799d6e260c7289994e46ad36df8

[Jose Borreguero]

Refs #5770 files for singlefunction

Changeset: b01cd5d9c3eda41a6a6239b29ca4bb81a8c92002

[Jose Borreguero]

Refs #5770 small changes

Changeset: acb245ca4617f7efb0e417606e6990ee63a10b12

[Jose Borreguero]

I decided to do a python script, since this fitting must be customized to the particular sample used in the experiment

[Russell Taylor]

Re #5770. Clear compiler warnings.

Changeset: b98b08455c860ae31beab7d14e3fddd17e0c9d73

[Jose Borreguero]

Refs #5770 Removed the C++ code and wrote a Python script instead

modified: CMakeLists.txt deleted: VisionCalibrationMultiplePeak.h modified: BackToBackExponential.cpp deleted: VisionCalibrationMultiplePeak.cpp

Changeset: 4c2873444b358799d6e260c7289994e46ad36df8

[Stuart Campbell]

This ticket has been transferred to github issue ​6616