GitLab

Apparently a tiny change (10^-18) in the calculated one_over_wl (k) value causes the IndexPeaks tests to fail.

The value can already be calculated from a Peak object so
these changes simple ensure it is displayed in the table
representation.

Monte Carlo simulation to calculate a quantity called absorption
weighted path length that is used in extinction correction calculations
downstream from Mantid eg in Jana. Requested by WISH. Specific changes:

- new AddAbsorptionWeightedPathLengths algorithm that uses the existing MCAbsorptionStrategy class to do most of leg work
- adjust the MCAbsorptionStrategy class to make the code more reusable for calculation involving a single wavelength such as the absorption weighted path length for a peak. In particular change the Calculate function signature to replace the HistoramData and Spectrum parameters with simple vectors and move the interpolation functionality back into the MonteCarloAbsorption algorithm
- extend the Peak class to add in an extra column called tbar for the absorption weighted path length + get\set functions. Also expose the get\set in the Python interface
- update SaveReflections python algorithm to write the actual tbar value for each peak in a peaksworkspace out to Jana file instead of zero
- add the absorption weighted path length to the nexus file and adjust save\load code

Giovanni Di Siena authored Mar 31, 2020 and Gigg, Martyn Anthony committed Apr 07, 2020

In places other substitutions have been made, e.g
Clang does not yet specialize std::shared_ptr for T[]. Vector
has been used instead. The operator[] methods were incorrectly
marked const but returning a non-const reference - this has been fixed.
Refs #25842

Runs clang-tidy pass by ref across the codebase to fix-up various
warnings we are seeing in cppcheck

Also updated class_maker.py

For some peaks not on a detector the intersection code
can given false positives that not intersection is
possible due to precision issues. In our case we only
need the distance to any intersection point on the surface
regardless of whether it is an entry/exit/undefined type
of track.
Refs #26690

Refs #26829

Improves readability, code reuse and style.

Refs #26829

Also use std::round over boost::iround in Peak
Refs #26829

Refs #26829

It's possible that rounding errors introduce a small
difference that throws the check in Track::reset.
Refs #26690

Refs #26690

- Throw in a few places where the user tries to acquire an angle from a
null vector
- Fix discovered issues
- More noexcept, constexpr

Re #24860

Adding the new column in the middle is a breaking change to already existing functionality - tests can be adapted, but user scripts might be affected