Project

General

Profile

2020 Convergence and modernization suggestion list

What items inherited from the previously distinct nue and numu analyses could be replaced, improved, and/or synchronized better between the analyses?
Please copy and paste format from previous items.

Kalman and RemID in numu

Use case: muon track identification; containment cuts
Issue(s): no Kalman/remid expert --> unsupported
Potential solution(s): replace with prongs, BreakPointFitter, CVN-prong (as muon selector)
Roadblocks:
Possible knock-on effects from changes: {{strikethrough|nue signal extrapolation}}(ND numu selections), nue BEN decomposition (uncontained numu sample)
Modules that depend directly on input from Kalman/RemID:

MEFinder/MEFinder.fcl
  Track input, should be an easy change to BPF

NumuEnergy/
  Basically all the fcls uses kalman as the track input and remid as the PID input. NumuEAlg and NumuEAna use the cosrej.kal variables
  Not an issue if we change energy estimators.

CVN/art/RegCVNMapper.fcl
  2d and 3d track input

CVN/art/TFRegCVNEvaluator.fcl
  3d track input

CosRej/
  There are a number of "kal" variables that are set. Also remid is used as PID when setting kal variables. 

QEEventFinder/
  uses kalman as the track input and remid for the PID input/

ShowerLID/Recluster.fcl
  kalman is the input track label

2020 urgency (range 1-5):
prod5 urgency (range 1-5):

ShowerLID in nue

Use case: nue pT/p cut
Issue(s): ShowerLID PID training data is not maintained
Potential solution(s): replace "shower" pT/p (nuecosrej.ParticleShowerTransMom() --> srnuecosrej.partptp) with "prong" pT/p (nuecosrej.ParticleShowerTransMom() --> srnuecosrej.partptp). Bonus: use prongs w/ good EM score for prong CVN only?
Roadblocks: Needs study to ensure similar performance. Calculated in cosrej::MakeNueCosRej (ART level). Would need updates to be usable w/ prong CVN.
Possible knock-on effects from changes: {strikethrough|none?} nue energy estimator( prong.shwlid.calE)
2020 urgency (range 1-5): 3
prod5 urgency (range 1-5): depends on whether prong CVN to be used or not

Quantiles in nue Extrapolation

Use case: Extrapolation
Issue(s): The hadronic energy quantiles are used for numu extrapolation, but not for nue extrapolation.
Potential solution(s): Use hadronic energy quantiles in nue extrapolation. Benefit of also eliminated/seriously reducing the near-far differences systematic category.
Roadblocks: A new extrapolation will require significant study. Mapping nues to numus is (much?) more complicated than matching numu quantiles to each other in the two detectors.
Possible knock-on effects from changes:
2020 urgency (range 1-5): 2
prod5 urgency (range 1-5): 1

Unified Systematic Treatment, Studies

Use case: Creating prediction files, examining systematics, improving systematic studies
Issue(s): Historically, the numu and nue analysis evaluated systematics using different systems, and some systematics were parameterized differently.
Potential solution(s): Some effort has already gone into unifying the treatment of systematics. I would suggest continuing to push on this unification. For example, and numu and nue prediction files be all made simultaneously if they are not already? Similarly, the "big PDFs" that are made should be able to be made for numu and nue simultaneously by the same person. All the better if the "big pdf" can be supplemented by a website which makes all the information linked/browsable.
Roadblocks: This code is complicated, and so any work with it can be challenging. At this stage, we should probably avoid total rewrites and instead focus on adiabatic improvements.
Possible knock-on effects from changes:
2020 urgency (range 1-5): 3
prod5 urgency (range 1-5): 1