Add property LeastMaxPeakObsY to FindPeaks

[Wenduo Zhou]

Add property LeastMaxPeakObsY to FindPeaks. This property will specify the minimum value of the maximum Y value of peak within specified range. By it, a peak with small maximum counts on Y can be excluded. It is useful for EventWorkspace with integer counts on each bin.

Propagate this new property to GetDetOffsetsMultiPeaks and CalibrateRectangularDetector.

[Peter Peterson]

That is a pretty bad name. How about "MinimumPeakHeightObs" since it is parallel to "MinimumPeakHeight".

[Wenduo Zhou]

Refs #10200. Added property MinimumPeakHeightObs.

To FindPeaks, GetDetOffsetsMultiPeak and CalibrateRectangularDetectors. On branch feature/10200_findpeaks_min_maxobsy Changes to be committed:

modified: ../Mantid/Framework/Algorithms/inc/MantidAlgorithms/FindPeaks.h modified: ../Mantid/Framework/Algorithms/inc/MantidAlgorithms/GetDetOffsetsMultiPeaks.h modified: ../Mantid/Framework/Algorithms/src/FindPeaks.cpp modified: ../Mantid/Framework/Algorithms/src/GetDetOffsetsMultiPeaks.cpp modified: ../Mantid/Framework/PythonInterface/plugins/algorithms/CalibrateRectangularDetectors.py

Changeset: 3d0dd68a3ec9b3efff298516ea95271fb5f089ab

[Wenduo Zhou]

For tester

The purpose of this ticket is to add one more criteria on finding peaks. With the new property MinimumPeakHeightOBs, the peaks, whose maximum observed intensities are lower than specified value, will not be considered. It is good to avoid to fit noise as peak.

You can download verify10200.py. Inside the script, a mockup workspace is created. Peaks are fit with and without the new option. You can tell the difference easily.

[Roman Tolchenov]

A couple of questions:

• FindPeaks with the new option leaves a row filled with zeros. Is it possible not to output this row at all?
• I ran FindPeaks without PeaksPositions. The new option doesn't make a difference. Is it expected?

[Wenduo Zhou]

Replying to Roman Tolchenov:

A couple of questions:

• FindPeaks with the new option leaves a row filled with zeros. Is it possible not to output this row at all?

FindPeaks is called intensively by GetDetOffsetsMultiPeaks, which specifies a list of peak positions to find. I used the zero-value row to indicate that the fit to a certain peak fails, i.e., no peak is found at that position. Such that it won't be considered. Without the zero-value row, it is difficult to have the mapping between input peaks positions and output fitted peak positions.

• I ran FindPeaks without PeaksPositions. The new option doesn't make a difference. Is it expected?

It is an expected behavior. If FindPeaks is run without PeakPositions, then an algorithm is used to search peak automatically. Usually it'd rather miss a peak than fit to tiny peak close to noise.

[Roman Tolchenov]

Merge remote-tracking branch 'origin/feature/10200_findpeaks_min_maxobsy'

Full changeset: ​280f1ba70bf8f9ea31e4210f4933144598242ed8

[Roman Tolchenov]

Merge remote-tracking branch 'origin/feature/10200_findpeaks_min_maxobsy'

Full changeset: ​280f1ba70bf8f9ea31e4210f4933144598242ed8

[Stuart Campbell]

This ticket has been transferred to github issue ​11042