Update on track finding in FDC

From GlueXWiki
Jump to: navigation, search

The segment-based track-finding algorithm has been refined:

  • I no longer rely on the refitting to determine the charge of the particle unless the individual segment information is ambiguous for those segments linked into candidates.
  • The dip angle and vertex position are determined from either the most upstream segment or the result of the refit depending on which gives a vertex position closest to the center of the target.
  • The track-matching cut from segment to segment is now momentum-dependent: \Delta r<2.79+2.88/p^{2}.

Segment matching cut.gif

These fixes appear to solve the 15 degree inefficiency problem for all values of momentum studied except for very low momentum. Shown below is the efficiency plot for single pions. The ratio of hits matched on a candidate track after swimming to the hits matched on a thrown track after swimming is required to be greater than 50%. A minimum of 5 FDC hits on the thrown track is required.

Track finding eff 1pion.gif

Here is the efficiency/track for 3 pion events generated with genr8:

Track eff 3pion.gif