Segment Point Matching

 

The segment finding analysis works in three steps:

• a match is made between reconstructed segments found by the segment finding algorithm and MC segments,
• the quality of the matched segments are assessed and those passing a contamination cut are assigned to MC ID's, and
• assigned segments are further examined to determine their quality.

Each hit that is generated has stored with it the MC number of its parent segment. Due to the left-right ambiguity in the drift direction, two points are produced for each MC hit. The MC ID is therefore signed to indicate which sense of the ambiguity is the true one. When a hit is place on a reconstructed segment it is actually one of the points that is being used. It is a genuine points if the correct sense of the ambiguous drift direction is used and a ghost points if the wrong sense of the ambiguous drift direction is used. A point has only been correctly placed on a segment if the hit used has the same MC ID as the true segment.

If all reconstruction were perfect then the matching and assignment of reconstructed segments to MC truth would be trivial. However, inefficiencies and contamination of segments with incorrect points complicates the matching and assignments. In the analysis we have taken the point matching\ approach. Point matching is based on comparing entries in segment grids. Grids are two dimensional arrays where the points placed on a reconstructed segment are cross-referenced with the MC segment from which they came. For each reconstructed segment in turn the maximum entry in a grid containing the genuine points is found, this is called the leading entry and the segment is considered matched to the leading entry. The contamination is then defined as the total number of points placed on the segment but not in the leading entry. The reconstructed segment is considered an assigned segment to a particular MC ID if the ratio of the contamination to leading entry is less than 0.2 (an input parameter).

The following classification of MC segments was found to be useful in studying the segment finding performance. If a MC segment had a reconstructed segment assigned to it then it was said to be found, else it was lost. If a MC segment had several segments assigned to it then it was split. A reconstructed segment that is not assigned to any MC segment is a spurious segment.

In making assignments fiducial cuts are sometimes placed on the sample of MC segments that are to be considered. These fiducial cuts ensure that only MC segments in a reasonable kinematic range are counted against the segment finding efficiency. The cuts are made on the and tangent of the polar angle, , of the track giving the segment. Each MC segment failing the fiducial cuts is considered a nasty segment. If a nasty segment is reconstructed then neither found nor spurious segment counters are incremented, while if a nasty segment is not found then no action is taken. Nasty segments could create a problem because they create a background of segments that are difficult to exclude from the spurious category.