Create a 2D rebinning algorithm

[Martyn Gigg]

(In [12639]) Add a Vertex and ConvexPolygon class to aid in 2D rebinning. Re #3197

[Martyn Gigg]

(In [12641]) Fix for exception include. Why does MSVC let this compile? Re #3197

[Martyn Gigg]

(In [12647]) Add methods for area determination to ConvexPolygon. Refs #3197

[Martyn Gigg]

(In [12648]) Correction for include (again). Refs #3197

[Martyn Gigg]

(In [12664]) Change the implementation of the area method for convex polygon so that it is not as easy to get one of the internal triangles vertex order mixed up. Refs #3197

[Martyn Gigg]

(In [12719]) Convex polygon intersection code should be working correctly now. Maybe not as efficient as it can be but tests are there and it works. Re #3197

[Martyn Gigg]

(In [12723]) Extract some code regarding bin creation and put it in NumericAxis. Re #3197

[Martyn Gigg]

(In [12724]) Checkpoint for Rebin2D. Not included in CMakeLists yet so doesn't build. Re #3197"

[Martyn Gigg]

(In [12782]) The performance can be improved but Rebin2D should give the correct values with the exception of masked bins at the moment. Re #3197

[Martyn Gigg]

(In [12786]) I forgot I changed something in Gometry as well. Re #3197

[Martyn Gigg]

(In [12788]) Imrpove the performance of Rebin2D with a couple of simple checks. Re #3197

[Martyn Gigg]

(In [12791]) Add a performance test for the polygon intersection code. Re #3197

[Martyn Gigg]

(In [12795]) Refactor some code in Rebin2D to make handling distibutions easier and clearer. Re #3197

[Martyn Gigg]

It's ready to test. A workspace with a spectra axis will need ConvertSpectrumAxis running first. It should probably be done as part of the algorithm but it is horrendously slow at the moment so I chose not to for the time being.

[Stuart Campbell]

There is no wiki page - created ticket #3268

[Stuart Campbell]

It seems to work, but it is not the most intuitive algorithm at the moment! :-)

[Martyn Gigg]

Jon Taylor has demonstrated some oddities to me that mean that this isn't quite right yet so lets not broadcast it for the release.

[Stuart Campbell]

Just a suggestion; but should we close this one under iteration 29 (as this was where the work has been done) and create new tickets for the problems Jon has seen under the next iteration?

[Martyn Gigg]

(In [12998]) Improve the polygon intersection algorithm from O Rourke by allowing more than integer precision when computing the cross products. Re #3197

[Martyn Gigg]

(In [13022]) Revert the changes for precision as they are actually not reliable. Re #3197

[Martyn Gigg]

Following advice from Michael Reuter I'm going to implement the implementation of polygon intersection by M. Laszlo to avoid the precisison problems with the implementation by O Rourke.

I'll close this one and I have created a new ticket, #3341, for this.

[Owen Arnold]

Tested using combintation of CreateWorkspace and Rebin2D. No issues. Worth going back to original tester before signing-off.

[Stuart Campbell]

Looks good to me. I basically read in some (BASIS) data, converted to EnergyTransfer, Rebinned, ConvertedSpectrum Axis to Q, then ran Rebin2D...

[Stuart Campbell]

This ticket has been transferred to github issue ​4044