ePIC Jet Reco. Task Force
Discussion Notes
Derek Anderson
02.14.2023

(Apologies for any ommissions or mistakes!)

Meeting start: 3:03 PM ET

 - Discussing jet edm object & algorithm prototypes:
   - edm object can be found here:
     - https://github.com/ruse-traveler/JJetFactory/blob/main/Jet.yaml
   - algorithm can be found here:
     - https://github.com/jizhongling/EIC-JJetFactory/blob/main/JetReconstruction.cc

   - How would background be incorporated into this
     algorithm/object? If we use rho, how would this
     be stored/accessed?
     - We could calculate rho for the event, and then
       store it as an attribute of each jet?
     - That would wasteful, though. It should be kept
       with the other event-wise quantities.
     - Maybe off-axis cones: background energy density
       will vary strongly with rapdidity in e+p/e+A 
       collisions.
       - OA Cones could be added to jet edm object
       - We should think about this before pursuing
         any one strategy...
     - Note that calculating rho for each event/jet
       would be very, very slow
       - The acceptance of ePIC is also *very* large.
         Rho  calculations are already slow at STAR,
         which (while having a large acceptance) is
         smaller in acceptance than ePIC.
       - Not to mention, storing rho for each jet
         would entail rho being recalculated *every*
         time a jet is created.

 - How do we deal with multiple inputs/outputs? What
   is the common practice in EICrecon/JANA for using
   a generic algorithm with multiple inputs?
   - Would we create a separate factory for each
     input collection we would want to process? e.g.
     a factory for calorimeter jets, a factory for
     track jets, etc.?
     - Yes: that's how JANA deals with this. Consider
       how clustering is handled in the calorimeters.
   - Right now we have a reconstructed charged
     particle (using tracking info) class, and
     eventually we'll have reconstructed particle
     (using tracking + calorimeter info) class.
     - This is probably how we should approach jets
       to begin: have charged jets (& particle-level
       jets)
     - Combining detectors gets complicated *fast*
       - Adding in calorimeters means we have
         to consider hadronic subtractions.
       - Also if we're adding in both ECals and
         HCals, we have to consider how to weight
         each one.
       - While we want flexibility, we also want
         our output to be physically meaningful
         and able to be understood.
   ==> Let's keep it simple to begin with: we
       will produce 2 output collections
       (i)  Jets made out of reconstructed charged
            particles
       (ii) Jets made out of MC (truth) particles
            (e.g. the Pythia8 event record or thrown
            particles)

 - Homework:
   - Zhongling will start work on jet factories
     for reconstructed charged particles and
     MC particles
   - Derek will issue a PR to add jet edm object
   - Derek will get an "official" fork associated
     with issue #470
     - (That way development is publically visible.)

Meeting end: 3:47 pm ET