An Update on Using QE Events to Estimate the Neutrino Flux and Some Preliminary Data/MC Comparisons for a QE Enriched Sample

1
An Update on Using QE Events to Estimate
theNeutrino Flux and Some Preliminary Data/MC
Comparisons for a QE Enriched Sample

• Review of motivation
• Update on QE sample selection
• Results for a high statistics MC set
• Data/MC comparisons

2
Motivation

• The x-sec for DIS events is fairly well known at
  high energies (20GeV) and it is easy to select a
  sample of DIS events at this energy. The DIS
  x-sec can be 'divided out' of such a sample to
  give an estimate of the neutrino flux at this
  energy.
• The shape of the QE x-sec is well known and is
  flat down to 1GeV but the normalization of this
  x-sec is not so well known. The DIS flux
  estimate can be used to 'pin' the normalization
  of the QE x-sec.
• Using the flat shape of the QE x-sec the neutrino
  flux can be estimated as a function of energy by
  again 'dividing out' the x-sec from samples of QE
  events in a series of reconstructed energy bins.
  

3
QE Sample Selection

• I am using a method of maximum likelihood in a
  series of reconstructed
• neutrino energy bins from 0-20GeV to identify
  QE events.
• The following plots briefly recap the variables
  that go into the ML analysis.
• They correspond to a MC sample that was
  generated with a flat energy
• spectrum

Shape for all events reflects total interaction
x-sec
Asymmetric binning to reflect energy resolution
and ensure adequate numbers of events in each
bin to make pdfs
4
QE Sample Selection

• The first two variables that go into the
  likelihood analysis are just the numbers of
  reconstructed tracks and showers generally
  events with no tracks are likely not to be QE and
  events with no showers are likely to be QE.
• I also use the reconstructed invariant mass
  squared

Large numbers of DIS events due to large flux
out to 20GeV and dominance of DIS x-sec
(blackQE, blueRES and redDIS)

• Now want to consider event topology
• and PH distributions near to the
• vertex to try to distinguish between
• QE (proton), RES (protonpion) and
• DIS (pions) events.

5
QE Sample Selection

• I remove the main muon track from an event (if
  there was one) as
• well as hits that occur more than 2m in z from
  the vertex (protons
• and pions will not travel further than this)
  and 'crosstalk-like' hits
• (defined as having PHlt1.5pe).
• I then define several variables

• The number of high PH hits (gt20pe) remaining
• The total PH of the remaining hits
• Total remaining PH as a fraction of total event
  PH (similar to y)

6
QE Sample Selection

• Black QE
• Blue RES
• Red DIS

• These two variables are highly
• correlated and so I combined
• them into a single 1D pdf using a toy
• principal components analysis.

7

• The final variable I am using for a pdf is
  obtained by performing a Hough
• 
  transform on the remaining hits
  
• 
  and taking the height of the
  peak.

• In each case the low energy events
• are the hardest to discriminate
• between.
• For all energy ranges the RES events are much
  harder to remove than the
• DIS events.

8
PID Results

• I then form a QE PID parameter for each energy
  bin based on the
• probabilities outputted from the ML analysis
  in that bin.
• Using the first half of the MC events to make
  the pdfs and running
• the second half through the analysis gives the
  following

9
Further Work

• There are several unfolding methods that could
  be used to get a flux
• estimate from a QE sample these have not
  been looked into fully
• yet.

Data/MC Comparisons for a QE Enriched Sample

• I have run samples of pME data and MC (R1.16)
  through the MLPID
• analysis in order to take a first look at some
  physics distributions for
• a QE enriched sample.
• In what follows all distributions have been
  normalized using POTs

• Data 1.21e18 POTs from May after 'good beam'
  cuts
• abs(hornI)gt0.1
• -2.0lthpos2lt0.0
• -1.0ltvpos2lt2.0
• closest spill lt2.0
• MC 0.90e18 POTs

10
pME MC v.s MC

• First I used half of the pME MC to construct my
  pdfs and then ran the
• remaining half through the MLPID analysis to
  see what sort of purities
• to expect

The resulting efficiencies (black) and purities
(red) look worse than those for my previous flat
energy spectrum sample am not sure yet why
this is the case.
Due to flux and x-sec I only had enough
statistics to look in the 2-10 GeV range

• The following sample of QE-like events using
  the MC for pdfs and running
• the data through the analysis should be 60
  QE events.

11
pME MC v.s Data

• Reconstructed neutrino energy

MC seems to be shifted by 0.5GeV above data.
Black data, Red MC
12
pME MC v.s Data

• Reconstructed muon energy

MC seems to be shifted by 0.5GeV above data.
Black data, Red MC
13
pME MC v.s Data

• Reconstructed shower energy

MC seems to be shifted by 0.1GeV below data.
Black data, Red MC
14
pME MC v.s Data

• Reconstructed y

MC seems to be shifted towards slightly lower y.
Black data, Red MC
15
pME MC v.s Data

• Reconstructed Q2

MC seems to be shifted towards slightly lower Q2.
Black data, Red MC