DetectorEfficiencyVariation Diagnotics Algorithm (2)

[Nick Draper]

The algorithm should perform an integral over range divided by solid angle and the average counts per steradian.

Code on how the fortan code does this but I suggest.

1. Run SolidAngle (SubAlgorithm)
2. Intergrate data over the range (SubAlgorithm)
3. divide intergral by solid angle (SubAlgorithm)
4. calculate average integral value
5. for each spectra divide 3/4
6. Apply cut off values to create mask list

[Steve Williams]

I understand the main functionality. My time estimate is only that approximately half the time could be spent looking at the fortran and half the time on the implementation. It depends how hard it is to find the relevent fortran subroutines and how complicated they are.

[Steve Williams]

(In [2629]) New algorithm FindProblemDetectors, should we change the name (FindBadDetectors?)?. The algorithm isn't tested yet. From a white beam spectrum it identifies detectors that have unusually high and low count rates (detector efficiency). SolidAngle now throws when it is passed a workspace that doesn't have geometry information. re #598

[Steve Williams]

(In [2630]) Fixing Linux build failures, forgotten <climits> re #598

[Steve Williams]

(In [2652]) Tests for FindProblemDetectors, no tests for the solid angle part of the calculation yet and the algorithm doesn't deal with masking re #598

[Steve Williams]

(In [2705]) WBVMedianTest is a more descriptive name and so it replaces FindProblemDetectors detectors. Here are some bug fixes and tests. If this check in works I'll close the ticket re #598

[Steve Williams]

More work may be needed because currently this algorthim doesn't differentiate between spectra with high and low counts when writing the ouptut workspace. I did this to maintain compatiablity with methods of masking detectors, which I have not investigated.

[Stuart Campbell]

This ticket has been transferred to github issue ​1446