SCD Bragg peak integration algorithms

[sf1919]

Summary
A project to provide a suite of integration algorithms and visualisation tools that allows users to select the optimal integration method for a particular sample. At present Mantid only supplies 2 choices.

Intended outcome
Integrate some samples/use-cases which currently cannot do without a lot of manual adjustment (e.g. overlapping domains) - more efficient and better user experience

Improve accuracy of integration – this leads to better refinements and results.

There is a reputational benefit to ISIS.

Acceptance criteria

• Quantitative benchmarking of different algorithms types (e.g. fitting vs summing) to check the different algorithms produce consistent refinements for standard samples for which the algorithms are equally suitable

• Track usage data for MVP to show that new algorithms are starting to be used

• Qualitative positive feedback on visualisation/mock-ups

How will it work?

Features in Scope:

• Algorithm to fit peak profiles in 1D independently (in TOF/d-spacing) and sum intensity from fits in neighboring detectors. Ensuring 1D performs well is a prerequisite to fitting in 3D.
• Algorithm to fit peaks in 3D (TOF & detector XY or Qx, Qy, Qz)
• Algorithm to integrate by summation and adaptive box in detector XY and TOF/d-spacing
• Learnt-profile to use strong nearby reflections to adjust profile parameter/integration box in these algorithms
• Handling overlapping peaks (choice of integrate together, or flag to ignore)
• Another summing algorithm (e.g. shoebox in TOF and detector XY)
• Visualisation tools for each new algorithm to judge quality of integration (help adjust parameters and compare methods)
• Visualisation tool for current 3D (Qx,Qy,Qz)-space integration algorithm (IntegratePeaksMD)
• Quantitative validation/comparison of different methods – check refinements are consistent

Features out of scope:

• Scaling/corrections to integrated data