Michael's Notes

Notes on to do items:

teaching TrackCleanUpAlg about pions and protons

This is Kanika's suggestion. We should teach TrackCleanUpAlg about how to treat pions and protons so we can run it over more "types" of track (like the other BPF tracking assumptions.)
  • She suggested to me that this is as simple as importing a single number (the MIP dE/dx for each particle type) - we just need to get that number
  • I thought that maybe we could get that number out of the dE/dx LL histos already generated for the BPF muon PID
  • we would then need to figure out how to adjust TrackCleanUpAlg to take in a "particle type"
  • but this is tricky because weighted prong energy already (sort of) accounts for this...

cleaning up the BPF CAF branches

  • add a "nbpf" to the fuzzyk.png branch (or wherever is the appropriate place...) - AND set it to fill properly (see how this number is set for fuzzyk and elasticArms...)
  • set the variables properly for the bpf tracks - are we putting in an entry for all 3 track assumptions regardless of muon or not?
  • set the ntracks correctly for bpf in the trk branch - OR even better: just remove the trk.bpf branch since that is all located in the fuzzyk.png branch

other tracker items on Mark's wish list:

(from an old e-mail from Mark)

Hi Michael,

cc-ing Erica because she and I also talked about some of these metrics:

A few things I’d like to see, esp. data vs. MC comparisons for are:

- position chi^2 / dof, maybe as a ratio, maybe as a scatter plot
One hopes data and MC agree on this and that the P-values are more-or-less uniformly distributed for good tracks, but maybe there are differences due to alignment or mis-modeling of the geometry data vs. mc. Maybe we haven’t tuned the per cell uncertainties correctly. Maybe we don’t do the rotations into track-based coordinate frames correctly.

- scattering chi^2 / dof, ditto on ratio and scatter plot
Again, one hopes to see uniform P-values here if things are working as they should

- scattering chi^2 vs. number of scattering planes from the end of the track
This one speaks to how well we are doing with the dEdx and multiple scattering calculations. Very dependent on how well we are modeling the radiation length of the detector. Instead of #scattering planes, it could be reco particle momentum at the scattering plane on the x axis.

I think these can be made from the debugging ntuples in the break point fitter module although for memory reasons you’d probably want to make them histograms rather than output all the gory details break point fitter is capable of putting into TTrees.


Studies to be done:

looking into the overlap/intersection of CVN/BPF/remid event selection

First round of this study is done (see docDB 19266.)

Requested follow up plots:
  • plot by true variables
  • make same plots for the FD
  • make same plots for the ND but cut out the muCat
  • plot CVN scores for events in each bucket (is it labeling the events in bucket 2 as cosmics?)
  • add in cosrej and see what happens (do we lose the events in bucket 2 in the FD anyway?)

true vertex study

  • Make sure to do this study with and without my recent changes to the BPF dE/dx tables!
  • replace the EArms vertex with the true vertex in FuzzyK (this is just a fcl switch), rerun FuzzyK and everything downstream, remake my muon Eres plots
  • NOTE: this study will inform how hard we need to push on getting better vertexing (currently Arist's domain...)

BPF containment cuts:

  • develop a robust set of BPF and prong based containment cuts
  • use the normal remid-selected muon track cuts applied to the best BPF track
  • add in some cuts for all other prongs (to be developed...)
  • maybe look to nue for their prong containment cuts
  • calculate numbers of events (maybe broken down into true E categories) kept by the old cuts and the new.

tracker comparison follow ups

  • regenerate my muon Eres plots using CVN-prong for muon selection and BPF muon energy reconstruction
  • regenerate all plots from docDB 18729 using a tighter CVN-event cut (0.8??? - confirm with Fernanda)
  • finer binning on the theta diff plots
  • plot theta diff vs. true muon theta
  • code in my tracking efficiency plots into the CAFAna framework
  • investigate the bump at 0.5 in completeness (really just validate that that is a view-matching problem...)
  • split some of the plots I've already made into categories of true lepE???

correct the BPF muCat energy

  • (done to first order...)

muCat Notes

General To Do list:

  • generate some kind of official recommendation as to which tracker/PID combo we should use for this summer and going forward for numu
  • possibly code up (some of) my CAFAna vars used in my tracker studies in BPFVars.h
  • help push the vertex improvements forward - depending on the results of my true vertex study

retired thoughts