CC Meeting, March 9, 2005

Flux via CC events

Eventual goal is to use high energy CC events and relatively well known high energy total cross section to predict high energy flux. Perhaps try to use QE events and relatively well known QE cross section to predict shape.

Motivated by MDC needs.

Selection Criteria for CC-like events

fiducial: Constructed to match outline of near partial planes, but shrunk a bit. Taken (shamelessly) from Pitt group. For event vertex I require (in meters): (0.6 < z < 3.56), (0.3 < u <1.8), (-1.8 < v < -0.3), (x < 2.4) and 0.8 m from coil hole.

Figure 1: True and reconstructed (x,y) vertex positions for true nu-CC events passing the fiducial cut.

pass track: Tracker passes, chi2/ndf < 10, track U/V vertex z position within six planes of each other.

Figure 2: Track chi2/ndof (left) and difference in planes between the track beginning in the U and V views (right).

fit momentum error: fractional error less than 20% for muons exiting the detector. No cut for stoppers.

Figure 3: The fit momentum divided by its error for neutrinos and anti-neutrinos vs. the reconstructed muon energy. The figure is in "logz" to make bins with a small number of entries more visible. Plot is for events passing the cut.

cleaning: no other event with a vertex time within 50 ns. More than 90% of non-track energy in fully instrumented planes.

Figure 4:The difference, in time, between the event under consideration and the nearest event in time. The shaded distribution was made for those cases in which both events were derived from the same neutrino (e.g. split events).

Cut Accounting
cut	# events	# events / 10¹³POT	A
true fiducial CC	43017	0.14746	1
reco fiducial	53752	0.184259	1.24955
pass track	37285	0.127811	0.86675
fit momentum error	30361	0.104076	0.705791
cleaning	28649	0.0982072	0.665993
reco track energy > 0.5 GeV	27529	0.0943679	0.639956

A is defined as #(events passing selection)/#(true fiducial CC). It can be larger than one (e.g. NC passing cuts).

Figure 5: The number of events reconstruced as CC (i.e. passing my cuts) divided by the number of CC events generated in the fiducial volume as a function of energy. The reconstruced CC events include events which are actually NC.

Figure 6: true energy spectrum for CC events (truth) and CC-like events (reconstructed).

Figure 7a:Effect of quality cuts on reconstructed muon energy. True CC events passing the fiducial and track pass stage are in color. True CC events passing all cuts are shown as boxes.

Figure 7b:Effect of quality cuts on reconstructed neutrino energy. True CC events passing the fiducial and track pass stage are in color. True CC events passing all cuts are shown as boxes.

nu-CC energy distributions

I identify neutrino (as opposed to anti-neutrino) induced CC events by requiring (q/p)<0. The (small) anti-neutrino background comes from events with a poor fit. Generally, these are events with low energy tracks ranging out in the detector.

Figure 8: The fit momentum divided by its error for neutrino and anti-neutrino CC events. The peak at zero is due to muons which range out in the detector. I measure the momentum via the range for those events and do not cut on the error in the fit momentum. Plot is for events passing the cut.

Full CC sample

Figure 9a: True and reco energies for events selected as nu-CC.

Figure 10a: Reconstructed energy for MC and Mock-data for events selected as nu-CC.

Correction Factors

Tests

Figure X: Test of the method. Top right: a made up truth distribution. Top left: a made up acceptance function. Lower left: the bin by bin correction factors. Lower right: the true, reconstructed and corrected distributions. No background was included in the test and the energy smearing was 0.5/sqrt(E).

Figure 11a: Expected (MC truth) and observed (mock data) nu-CC energies. A bin-by-bin correction was applied to the observed histogram to account for efficiency and smearing. The mock data events were selected as detailed above.

DIS enriched

Figure 9b: True and reco energies for events selected as nu-CC. Events are required to have emu>1.0 GeV and eshw>1.0 GeV ("DIS-enriched").

Figure 10b: Reconstructed energy for MC and Mock-data for events selected as nu-CC. Events are required to have emu>1.0 GeV and eshw>1.0 GeV ("DIS-enriched").

Figure 11b: Expected (MC truth) and observed (mock data) nu-CC energies. A bin-by-bin correction was applied to the observed histogram to account for efficiency and smearing. The mock data events were selected as detailed above and were also required to have emu>1.0 GeV and eshw>1.0 GeV ("DIS-enriched").

QE enriched

Figure 9c: True and reco energies for events selected as nu-CC. Events are required to have emu>0.5 GeV and eshw<0.5 GeV ("QE-enriched").

Figure 10c: Reconstructed energy for MC and Mock-data for events selected as nu-CC. Events are required to have emu>0.5 GeV and eshw<0.5 GeV ("QE-enriched").

Figure 11c: Expected (MC truth) and observed (mock data) nu-CC energies. A bin-by-bin correction was applied to the observed histogram to account for efficiency and smearing. The mock data events were selected as detailed above and were also required to have emu>0.5 GeV and eshw<0.5 GeV ("QE-enriched").

No shower

Figure 9d: True and reco energies for events selected as nu-CC. Events are required to have emu>0.5 GeV and eshw<0.05 GeV ("zero-shower").

Figure 10d: Reconstructed energy for MC and Mock-data for events selected as nu-CC. Events are required to have emu>0.5 GeV and eshw<0.05 GeV ("zero-shower").

Figure 11d: Expected (MC truth) and observed (mock data) nu-CC energies. A bin-by-bin correction was applied to the observed histogram to account for efficiency and smearing. The mock data events were selected as detailed above and were also required to have emu>0.5 GeV and eshw<0.05 GeV ("zero-shower").

Conclusions

Still a work in progress.
I understand the data+reconstruction somewhat better now than before and can select a useful, reasonably well reconstructed sample.
No large MC / Mock-data differences seen in reco energy distribution. Possible this isn't the right place to look.
Also, # of MC events not sufficient to find few % differences.
Simple correction factors applied to predict true energy spectrum from reconstructed spectrum. A number of more sophisticated methods exist to do this.
Need to apply cross section.

Figure Y: Test of the method in which the underlying truth distribution differs by ~10% from the one used to construct the correction factors.