- Algorithm or Service name
- Projection Matching Algorithm
- Author or Maintainer
- Robert Sulej, Dorota Stefan
- firstname.lastname@example.org, Dorota.Stefan@cern.ch
- one line description
- Reconstructs structures of 3D tracks interconnected with vertices; the input is 2D clusters.
PMA is a multi-trajectory fit with some capabilities of pattern recognition. Algorithm can perform single track reconstruction, track stitching, EM shower direction reconstruction, vertex finding and reconstruction of particle hierarchy.
The input to PMA is a set of 2D clusters; particle tracks can be fragmented into a few clusters. Algorithm will search for matching clusters in all projections and it will iteratively build tracks. It is also possible to provide cluster associations calculated with another pattern recognition algorithm and use only the fitting functionality of PMA.
PMA underlying idea is to build and optimize objects in 3D by minimizing the cost function calculated simultaneously in all available 2D projections; the cost function consists of 2D distance of hits to optimized object projections, penalty of tracks curvature, and 3D distance of various feature points to optimized object (optional, used e.g. to deal with difficult wire-plane-parallel orientation of tracks).
The key features of PMA are:
- hit-hit associations are not needed to create 3D position; each 2D hit has its own position on the 3D trajectory;
- 3D objects and reconstruction of structures of objects is driven directly by 2D information;
- local information from multiple tracks is used in vertex optimization;
PMA works for 3- and 2-view geometries; data from sections where only one view is available are still useful (e.g. difficult track orientation, hit or cluster inefficiency, …
- location in code
- 3D hit reconstruction, other
- code analysis done
- improved code released