tcnt/ntrk: a counter for distinct tracks

[sam-grant]

tnct.ntrk: a counter for distinct tracks

Note: Not sure if this works consistently, sometimes you get unlucky.

The event-level tcnt branch is "MARKED FOR REMOVAL", but a counter for the number of distinct tracks (not hypotheses) would be very useful for signal-like event selection.

Example: An event containing a reflected track produced by a cosmic muon

Subrun 27992 event 280437 in the pure CRY MDC202aw dataset. This event has two distinct track "legs": an upstream muon fit and a downstream electron fit. The types of events are distinct from those which have multiple fit hypotheses to the same physical track.

Signal-like events do not look like this, they have a single track, so we need a way to filter these types of events.

This event has track PDGs [11, -11, 13, 13], which has a duplicate fit hypothesis.

Strategy: count unique PDGs

Each track gets four fit hypotheses: [e-, e+, mu-, mu+]:

trk.pdg: [-11, 11, -13, 13]

Each hypothesis is stored in EventNtuple if the fit converges, so if the mu+ fit fails:

trk.pdg: [-11, 11, -13]

If there are multiple distinct tracks per event:

trk.pdg: [-11, 11, 11, -11, -13, 13, 13, -13]

We can simply count the maximum number of instances of a unique track PDG.

Demo

1 track

-------------------------------------------------------------------------------------
evt.event: 13046
evt.subrun: 24888
tcnt.ntrk: 1
trk.trk.pdg: [-11, 11, -13, 13]
-------------------------------------------------------------------------------------

2 tracks

-------------------------------------------------------------------------------------
evt.event: 14399
evt.subrun: 24888
tcnt.ntrk: 2
trk.trk.pdg: [-11, 11, 11, -11, -13, 13, 13, -13]
-------------------------------------------------------------------------------------

3 tracks

-------------------------------------------------------------------------------------
evt.event: 2009
evt.subrun: 24939
tcnt.ntrk: 3
trk.trk.pdg: [11, 11, 11, -11, -11, 13, 13, -13, -13]
-------------------------------------------------------------------------------------

4 tracks

-------------------------------------------------------------------------------------
evt.event: 472302
evt.subrun: 24888
tcnt.ntrk: 4
trk.trk.pdg: [-11, -11, -11, 11, -11, 11, -13, -13, 13, 13]
-------------------------------------------------------------------------------------

5 tracks

-------------------------------------------------------------------------------------
evt.event: 230379
evt.subrun: 25843
tcnt.ntrk: 5
trk.trk.pdg: [-11, -11, 11, 11, 11, 11, -11, 11, -11, -13, -13, 13, -13, 13, 13, 13, -13, 13]
-------------------------------------------------------------------------------------

Edge case

You could get unlucky and have unique PDGs by accident, although I have never seen this happen. I'm also not sure if there is a way around this.

Track 1: [-11, 11, -13]  (mu+ failed)
Track 2: [13]  (only mu+ converged)
Result: [-11, 11, -13, 13] (all unique)