scispace - formally typeset


A measurement of muon neutrino disappearance with the MINOS detectors and NuMI beam

01 Aug 2008-

Abstractv Chapter

Topics: MINOS (78%), NuMI (69%), Neutrino detector (60%), Neutrino (59%), Muon neutrino (57%)

Summary (13 min read)

Jump to: [A measurement of muon neutrino disappearance with][1.1 A brief history of neutrinos][1.2 Theory of neutrino oscillations][1.3 Experimental evidence for neutrino oscillations][1.3.1 Electron neutrinos from the sun][1.3.2 Electron anti-neutrinos from nuclear reactors][1.3.3 Atmospheric neutrinos][1.3.4 Man-made muon neutrino beams][1.4 Summary of neutrino oscillations][1.5 Outline of the dissertation][2.2 NuMI facility][2.3 MINOS detectors][2.4 Detector calibration][3.1 Event reconstruction][3.2 Charged-current muon neutrino interactions][3.3 Selection of QES, RES, and DIS events][3.4 Energy reconstruction in QES, RES, and DIS events][3.5 Monte-Carlo simulation][3.6 Study of near detector hits in the data and MC][3.7 Summary][4.1 Reconstructed muon and non-muon tracks][4.2 Improving sensitivity to muon tracks][4.3 Track based muon identification variables][4.3.1 Number of track scintillator planes][4.3.2 Mean pulse height of track hits][4.3.3 Signal fluctuation][4.3.4 Transverse track profile][4.4 Event classification with the k-nearest neighbor al-][4.5 Identification of muon neutrino charged-current events][4.6 A study of event selection for several beam config-][4.7 Comparison of muon tracks for data and MC][4.8 Summary][5.1 Charge-sign measurement using fitting algorithm][5.2 New method for charge-sign measurement][5.3 Charge-sign for non-muon tracks][5.4 Selecting muon neutrino and anti-neutrino events][5.5 Summary][6.1 Calibration of the near detector magnetic field][6.2 Selecting near detector events][6.3 Near detector data][6.4 Summary][7.2 Parameterization of the cross-section model][7.3 Fit description and χ2 function][7.4 Kernel method and χ2 function][7.5 Tuning results][7.6 Study of different fit approaches][7.7 Summary][8.1 Extrapolation from near to far detectors][8.2 Selecting far detector events][8.3 Far detector data][8.4 Summary][9.1 Description of the fit method][9.2 Measurement of neutrino oscillations parameters][9.3 Statistically independent pseudo-experiments][9.4 Treatment of the systematic errors][9.5 MINOS results] and [9.6 Conclusions]

A measurement of muon neutrino disappearance with

  • The MINOS detectors and NuMI beam Publication No. Rustem Ospanov, Ph.D. The University of Texas at Austin, 2008 Supervisor: Karol Lang MINOS is a long-baseline two-detector neutrino oscillation experiment that uses a high intensity muon neutrino beam to investigate the phenomena of neutrino oscillations.
  • The neutrino beam is produced by the NuMI facility at Fermilab, Batavia, Illinois, and is observed at near and far detectors placed 734 km apart.
  • The neutrino interactions in the near detector are used to measure the initial muon neutrino flux.
  • The vast majority of neutrinos travel through the near detector and Earth matter without interactions.
  • This thesis presents a measurement of the muon neutrino oscillation parameters in the framework of the two-neutrino oscillation hypothesis.

1.1 A brief history of neutrinos

  • In 1914, James Chadwick showed that a spectrum of electrons emitted in the β-decays is continuous [1]: N → X + e−, where N and X stand for the initial and final nucleus states.
  • The β-decays are now used to set a limit on the mass of the emitted neutrino by accurately measuring electron momentum and nuclear recoil energy.
  • In the 1960s, Davis used this reaction for the first time to detect neutrinos produced in the sun [9].
  • In 1964, Murray Gell-Mann and George Zweig proposed the idea that this family of particles4, which included pions and protons but not leptons, is composed of even more elementary constituents called quarks.
  • The theory predicted the existence of the heavy gauge bosons, W± and Z0, which mediate charged-current (CC) and neutral-current (NC) interactions.

1.2 Theory of neutrino oscillations

  • The Standard Model describes all existing data on the weak, electro- magnetic and strong interactions [5].
  • There are no experimental measurements to suggest a lepton number violation for the charged leptons [5].
  • 6See Reference [23] for a comprehensive review of different scenarios.
  • Experimental data suggest two very different scales for the mass differences squared, ∆m232 ∆m212.

1.3 Experimental evidence for neutrino oscillations

  • Neutrino oscillations are investigated using an intense source of neutri- nos and a large neutrino detector.
  • Neutrino oscillation experiments compare a number of neutrino interactions in the detector with an expected interaction rate, assuming that neutrinos travel directly from the source to the detector without oscillations.
  • A detailed understanding of the neutrino source is required to compute the expected interaction rate.
  • For neutrinos produced by the sun or cosmic rays, computation of the predicted rate presents a complex experimental and theoretical problem.
  • Man-made sources of neutrinos are 12 easier to understand, but they generate less intense neutrino fluxes.

1.3.1 Electron neutrinos from the sun

  • A first hint of neutrino oscillations came from the radiochemical exper- iment at Homestake Gold Mine, South Dakota, led by Davis.
  • Through this period the experiment measured a consistent rate of argon capture, and the final result for the production of argon atoms per day was 2.56±0.16(stat)±0.16(syst) SNU (1 SNU = 1 Solar Neutrino Unit = 1 capture per second and per 1,036 target atoms).
  • The sun generates energy via the nuclear conversion reaction, which schematically is expressed as: 4p →4 He + 2e+ + 2νe (1.14) Two neutrino are produced in each fusion reaction.
  • New experiments were constructed to study the electron neutrinos pro- duced by the sun.
  • The Kamiokande detector was built in the late 1980s.

1.3.2 Electron anti-neutrinos from nuclear reactors

  • A remarkable confirmation of the solar neutrino oscillation hypothesis was carried out by the Kamioka Liquid-scintillator Anti-Neutrino Detector .
  • This experiment observes interactions of electron anti-neutrinos through the inverse β-decay reaction: ν̄e + p → e+ + n (1.16) The positron deposits energy in the scintillator and then annihilates.
  • The average distance between KamLAND and the reactors is 180 km.
  • In addition, Bugey [37], Palo Verde [38], and CHOOZ [39] collabora- 20 tions searched for neutrino oscillations using reactor nuclear plants as a source of electron anti-neutrinos.

1.3.3 Atmospheric neutrinos

  • In the 1990s, several deep underground neutrino experiments measured fluxes of atmospheric neutrinos.
  • The underground detectors measured the ratio of muon neutrinos to electron neutrinos, R = Nνµ+Nν̄µ Nνe+Nν̄e , for neutrino energies of few GeV.
  • The distance L is a reconstructed distance from the detector to the point in the Earth’s atmosphere where the muon neutrino was produced.
  • The recent MINOS result [46] disfavors these two models compared to the neutrino oscillation hypothesis.

1.3.4 Man-made muon neutrino beams

  • The first man-made, neutrino beam was constructed at Brookhaven, Upton, New York.
  • Neutrino beam facilities were later constructed at CERN, Fermilab, KEK, Los Alamos, and Serpukhov, where they were utilized to study weak interactions.
  • In the early 1990s, atmospheric neutrino experiments provided the first hints for the muon neutrino oscillations with L/E around 400 km/GeV (∆m2 ≈ 10−3eV2).
  • A chief advantage of the man-made neutrino beams is the significant reduction of systematic errors in the measurement of the neutrino oscillation parameters.
  • A smaller (near) detector is placed very close to the neutrino source, and a larger (far) detector is placed a few hundred km away from the neutrino source.

1.4 Summary of neutrino oscillations

  • Experimental data provide compelling evidence that neutrinos do os- cillate between three mass states.
  • The values of the mass differences among these states are illustrated in Figure 1.8.
  • These experiments do not tell us which of these two oscillation scenarios is realized in Nature.
  • The atmospheric, accelerator, and reactor neutrino data are consistent with pure νµ → ντ oscillations with |∆m2| = 2.44 × 10−3eV2.
  • The MINOS experiment provides the most precise current measurement of |∆m2| [46].

1.5 Outline of the dissertation

  • This dissertation presents a measurement of muon neutrino disappear- ance using the MINOS detectors and NuMI neutrino beam.
  • The detectors and beamline facility are briefly discussed in Chapter 2.
  • This near detector analysis computes corrections to neutrino flux and cross-section models.
  • These corrections are used in Chapter 8 to compute a predicted event rate at the far detector assuming there are no oscillations.
  • 29 Chapter 2 NuMI facility and detectors The Main Injector Neutrino Oscillation Search experiment [52] is a two-detector neutrino experiment that uses an intense muon neutrino beam to study neutrino oscillations.

2.2 NuMI facility

  • The NuMI is a neutrino beamline facility [53,57,58] constructed at the Fermi National Accelerator Laboratory in Batavia, Illinois.
  • The horns are pulsed simultaneously with a proton pulse reaching a 200 kA peak current and generating a maximum 30 kG magnetic field.
  • 34 35 The transport line from the Main Injector to the target is instrumented with beam position monitors and beam profile monitors.
  • The accepted window was defined to select proton pulses with charatestics as similar as possible to the NuMI design specifications.
  • Figure 2.5 shows the contributions to the predicted MC neutrino spec- trum at the near detector from the pion and kaon decays for different beam configurations.

2.3 MINOS detectors

  • MINOS consists of two iron and scintillator detectors designed as hadronic sampling calorimeters with a muon tracking capability.
  • Fully instrumented planes contain 96 scintillator strips connected to two PMTs.
  • This detector was exposed to test beams at CERN, and the results of the tests showed that a calibrated response of the near and far detector readout systems are equal [65].
  • The near detector electronics [66] continuously record signals from 44 all detector channels during a neutrino beam spill.
  • 45 The near detector DAQ permanently records all detector hits within a spill window without any preconditions.

2.4 Detector calibration

  • A precise measurement of the oscillation parameters requires accurate calibration of two spatially separated detectors.
  • In the far detector, the attenution corrections vary by approximately 30% for an 8 m scintillator strip (for a sum of signals from both strip-ends).
  • 52 Chapter 3 Neutrino interactions in the MINOS detectors.
  • These detectors hits are examined using reconstruction algorithms for patterns that match muon tracks and hadronic cascades .
  • In Section 3.6, detector hits from the data and MC simulation are examined to validate the MC simulation.

3.1 Event reconstruction

  • The event reconstruction procedure uses the topology and timing of de- tector hits to identify neutrino interactions inside the detector or surrounding rock.
  • The uncertainty of the muon momentum measurement via curvature is estimated using the stopping muon tracks by comparing the muon momentum via the range with the muon momentum via curvature.
  • Reconstructed showers and tracks are assigned 56 to a reconstructed event based on the start position of the track and shower and the timing of the track and shower hits.
  • The vertex positions of the selected νµ charged-current events for the data and MC simulation are shown in Figures 3.2 and 3.3 for the near and far detectors, respectively.
  • In the far detector, a demultiplexing algorithm identifies which of the eight possible scintillator strips are associated with each far detector hit [77].

3.2 Charged-current muon neutrino interactions

  • The total νµ charged-current cross-section includes contributions from three processes2: quasielastic scattering (QES), resonance production (RES), 1100 ADC ≈ 1 photoelectron.
  • Figure 3.9 shows a typi- cal reconstructed QES event.
  • The absence of the visible hadronic energy is an efficient signature of the QES events in the MINOS detectors [80].
  • RES interactions produce muons and baryon resonances.
  • These interactions can change the multiplicity and the particle type of the hadrons produced in neutrino interactions.

3.3 Selection of QES, RES, and DIS events

  • A method to select QES, RES, and DIS events is devel- oped, using events from the Monte-Carlo simulation.
  • The QES and RES events can be used to measure precisely the shape of the neutrino energy spectrum, as discussed in Chapter 9.
  • The protons and pions, with momenta around 1 GeV, were observed to produce only a few detector hits [72].
  • The three categories of events are defined using reconstructed variables; each category is enriched with either QES, RES, or DIS events.
  • Approximately 40% of the QES events have no reconstructed hadronic energy.

3.4 Energy reconstruction in QES, RES, and DIS events

  • The reconstructed event energy is examined for QES, RES, and DIS events, using events from the Monte-Carlo simulation.
  • The true values are generated by the MC simulation.
  • 79 Figure 3.18 shows the energy resolution computed for the true QES, RES, and DIS events .
  • If the nuclear effects are ignored, the QES interaction is a two-body scattering process; this is illustrated in Figure 3.9.
  • The standard method estimates the neutrino energy as the sum of hadronic shower energy and muon energy.

3.5 Monte-Carlo simulation

  • The MC simulation of the MINOS experiment is used to develop anal- ysis methods and to compute a predicted event rate the far detector.
  • The neutrino interactions in the MINOS detectors are modeled using the NEUGEN neutrino event generator [78].
  • Final state leptons are allowed to leave the target nucleus without addi- tional interactions and are passed directly to the detector simulation.
  • The secondary particles produced in these interactions are traced through the realistic model of the MINOS detectors, 85 including the effects of the magnetic field.
  • The energy depositions within the active detector elements (scintillator strips) are passed to the MC simulation of the detector response.

3.6 Study of near detector hits in the data and MC

  • This section compares a number of neutrino induced hits in the near detector data and the MC simulation.
  • This effect was first observed in 2005 when it was noticed that the near detector channels for a few percent of the hits record an additional hit following the earlier hit in the same channel.
  • These two figures do not include hits from the channels that have 8 or more hits in a single spill; the hits from these channels are shown in Figures 3.22b and 3.22d.
  • The number of high signal hits agrees with the MC simulation.

3.7 Summary

  • This chapter opened with the description of the event reconstruction and followed with the discussion of the νµ charged-current interactions in the MINOS detectors.
  • The absolute QES and RES cross-sections are constrained by the previous neutrino experiments, but the normalization uncertainties are greater than the statistical precision of the MINOS near detector data.
  • A simple method was developed to select QES, RES, and DIS events.
  • The MC simulation is used to develop analysis techniques and to compute a prediction for the νµ charged-current event rate at the far detector without oscillations.
  • These events are identified by the observation of a muon track.

4.1 Reconstructed muon and non-muon tracks

  • This section provides a framework for the subsequent analysis of recon- structed muon and non-muon tracks in the MINOS detectors.
  • In the MC simulation, each energy deposit is tagged with a unique code for the particle associated with that energy deposit.
  • First, only tracks that have at least 5 scintillator plane hits in each detector view were considered.
  • Qualitative observations of the true νµ charged- 100 current events that fail this requirement indicate that these muon tracks are obscured by the associated hadronic shower.
  • The second preselection requires that a track vertex (see Section 3.1 for a definition) is contained within the fiducial volume of the detector.

4.2 Improving sensitivity to muon tracks

  • In these interactions, the muons are produced in association with the hadronic shower.
  • Figure 4.1 shows a νµ charged-current event and neutralcurrent events that contain a reconstructed track.
  • Figure 4.2 shows the mean pulse height of track segments plotted as a function of the track length.
  • Removing the hadronic shower hits closest to the track vertex might improve a muon identification algorithm.
  • All available parameters were varied in this fashion to minimize the figure of merit for the three variables described in Section 4.3.

4.3 Track based muon identification variables

  • Four variables were constructed to distinguish muon tracks from non- muon tracks.
  • The pattern of the muon energy loss in the scintillator strips provides another muon track signature.
  • These four figures are discussed in the following four subsections.
  • A variable sensitive to this signature was developed [95].
  • All histograms were normalized to the unit area.

4.3.1 Number of track scintillator planes

  • The number of track scintillator planes is plotted in Figure 4.4.
  • This variable is proportional to the length of the muon track within the detector.
  • First, a multidimensional likelihood method (described in Section 4.4) is employed for an event classification; this method compares reconstructed tracks of a similar length.
  • Second, a muon carries only a fraction of the total reconstructed neutrino energy, where the fraction is determined planes is used.

4.3.2 Mean pulse height of track hits

  • The mean pulse height of track hits measures the average energy loss in the MINOS scintillator strips.
  • This variable is computed using hits located away from the track vertex.
  • The above-mentioned optimization procedure was used to determine the fraction of the excluded planes.
  • The mean pulse height variable is computed as follows: Exclude 30% of the track scintillator planes closest to the track vertex; Compute the mean pulse height of the track hits in the remaining scin- tillator planes.
  • Events with large energy depositions by muons are infrequent; thus, this variable has a narrow distribution for the muon tracks.

4.3.3 Signal fluctuation

  • The third variable, Rf , measures fluctuations in the energy deposited in the MINOS scintillator strips.
  • The signal fluctuation variable is computed as follows: Exclude 30% of the track scintillator planes closest to the track vertex; .
  • Sort the selected hits in ascending order by pulse height.
  • The fraction where the two parts are divided is a tunable parameter used for the sensitivity optimization procedure; Compute the mean of the low pulse height hits and the mean of the high pulse height hits.
  • The peak of the muon distribution is located on the right of the peak of the non-muon distribution.

4.3.4 Transverse track profile

  • The fourth variable is the transverse track profile, Rt. A scintillator plane is made from closely packed 4.1 cm wide scintillator strips.
  • A typical hadronic shower has a transverse profile a few strips wide, so it deposits energy in multiple 111 112 strips within one scintillator plane [72].
  • The following three adjustable parameters determine this variable: (a) the fraction of the excluded planes; (b) the number of strips in the window around a track; (c) the time span of the window around a track.
  • In the remaining scintillator planes with track hits, select all detector hits, Salli , that fall within a 4 strip window and within a 37.36 ns time window around the track hits, Strackj , including hits that belong to the track.
  • The transverse track profile variable is the ratio of the track signal over the signal of all the selected hits: Rt = ∑M j=1 S track j ∑N i=1 S all i , (4.2) where Strackj is the pulse height of the track hits, and S all i is the pulse height of all the selected hits (including track hits).

4.4 Event classification with the k-nearest neighbor al-

  • Gorithm A typical classification problem has two classes: signal and background.
  • The knn algorithm uses a training set to estimate a density for the sig- nal and background events in a small neighborhood around the query event.
  • A value of k determines the average size of the neighborhood over which probability density functions are evaluated.
  • For their analysis, the number of the muon tracks included within the training set was reduced to match the number of non-muon tracks for the following two reasons.
  • This search algorithm requires 3The code developed for the knn algorithm was contributed to the TMVA project [97].

4.5 Identification of muon neutrino charged-current events

  • Reconstructed events in the MINOS detectors are classified using in- formation from a reconstructed track.
  • Both νµ charged-current and ν̄µ charged-current interactions are included into one category since the track classification variables do not distinguish between µ+ and µ− tracks.
  • For the far detector, the training set contains 163,838 muon tracks and an equal number of non-muon tracks.
  • The discriminant variable is shown in Figure 4.10 for the near detector and in Figure 8.6 for the far detector.
  • Figure 4.11 shows the selection efficiency and the background rejection; these two quantities do not depend on k, for k = 20, 60, 100.

4.6 A study of event selection for several beam config-

  • The energy dependence of the muon selection efficiency is examined.
  • The MC events used to create a training set are generated using the MC simulation of the low energy beam configuration.
  • The classification algorithm includes the number of planes variable, which depends on muon momentum.
  • Figure 4.12 shows the efficiency and purity (defined in Section 3.3) for the three beam configurations.
  • The purity for the high energy beam is lower, because the number of neutral-current background events from the high energy tail is increased.

4.7 Comparison of muon tracks for data and MC

  • The excess of low pulse height hits in the near detector data (see Sec- tion 3.6) implies differences between the data and MC simulation in topology and pulse height of detector hits.
  • This section evaluates a systematic error resulting from the excess of low pulse height hits.
  • This algorithm uses hits with pulse heights greater than 2 photoelectrons to create track clusters; this requirement significantly reduces the dependence of the track finding algorithm on low pulse height hits.
  • The number of the selected νµ chargedcurrent data events is greater than the number of MC events, consistent with the results of tuning muon neutrino flux and cross-sections models in Chapter 7.
  • A fraction of events that pass the original method and fail the modified method, and vice versa, is consistent with both the data and the MC simu- 128 129 130 lation.

4.8 Summary

  • A new method to the select νµ charged-current and ν̄µ charged-current events in the MINOS detectors was introduced.
  • This method uses the k-nearest neighbor algorithm and the four track based variables.
  • The 131 selected νµ charged-current events at the far detector include the estimated 0.6% background contamination from the neutral-current interactions [46].
  • The near and far detectors have a similar response to physical detector hits, so these results are appliacable to far detector analysis.
  • It is shown that this variable is sensitive to the muon charge sign and that it reduces a number of the background νµ charged-current events, among the selected ν̄µ charged-current events.

5.1 Charge-sign measurement using fitting algorithm

  • The MINOS detectors are muon tracking detectors; muons leave visi- ble tracks in the detectors, as illustrated in Figure 4.1.
  • The track finding and track fitting algorithms are described in Section 3.1.
  • The first two variables are computed together as the q/p ratio.
  • For some tracks, the q/p ratio cannot be unambiguously computed; such tracks are assigned a “failed fit” 134 135 flag.
  • The fitting algorithm correctly determines the muon charge sign for a majority of reconstructed tracks.

5.2 New method for charge-sign measurement

  • The selection of the ν̄µ charged-current events requires an efficient mea- surement of the muon charge sign.
  • An improved method for the measurement of the muon charge sign is described.
  • The majority of the µ− tracks move toward the coil, where many of these tracks stop.
  • The muon direction at the track vertex is determined by the track fitting algorithm.
  • These two variables are plotted in Figure 5.6.

5.3 Charge-sign for non-muon tracks

  • It is interesting to note that the curvature of the non-muon tracks does not depend on the magnetic field.
  • Figure 5.9 shows the azimuthal angle φ and the polar radius for the muon and non-muon tracks, and Figure 5.9 shows these variables for the π− and π+ tracks.
  • The non-muon tracks do not experience significant deflections in the magnetic field.
  • For these tracks, the 143 deflections are independent of the charge sign.
  • Figure 4.4 shows a number of track planes for the non-muon tracks; these tracks are typically very short and do not have a sufficient number of hits to measure the track curvature.

5.4 Selecting muon neutrino and anti-neutrino events

  • In this section a new charge-sign method is examined using the MC simulation.
  • The standard charge-sign selection uses the q/p variable computed by the track fitting algorithm for the reconstructed track in the event.
  • It was implemented to remove tracks with errors measuring the q/p value.
  • Figure 5.11b shows the true ν̄µ charged-current events.
  • The conclusion is that the new charge-sign method does not affect the selection of the νµ charged-current events.

5.5 Summary

  • A new and improved method for the measurement of the muon charge sign in the MINOS detectors was introduced.
  • This selection can be further improved by requiring that the tracks contain at least 20 hits in each detector view.
  • This work resulted in an improved understanding of the magnetic field in the MINOS detectors and reduced the systematic error for the measurement of the muon momentum from the curvature.
  • Initial observations show substantial disagreements between the data and MC simulation.
  • In Section 6.3, the νµ charged-current events in the near detector are examined.

6.1 Calibration of the near detector magnetic field

  • The near detector magnetic field is calibrated using muon tracks.
  • The procedure takes into account effects from 1The track fitting algorithm used in this section was developed by Sergei Avvakumov and it is based on techniques described in [102].
  • A comparison of the muon momentum from the range, Prange, with the muon momentum from the curvature, Pcurv, provides a tool for the calibration of the magnetic field strength in the near detector.
  • For the old magnetic field map, there is an approximately 5% shift between the data and MC simulation.
  • This observation prompted an investigation of the magnetic field properties in the near detector and far detectors; this work produced the new magnetic field maps, discussed earlier.

6.2 Selecting near detector events

  • The reconstructed events are required to contain at least one recon- structed track to be selected for their analysis.
  • This fiducial volume was adopted to minimize reconstruction errors and improve the measurement of event energy [103].
  • The νµ charged-current events can be further subdivided into the QES, RES, and DIS events using the technique developed in Section 3.3.
  • This se- lection method can be altered by changing the selection requirements listed in Table 6.1.
  • A complete analysis of the near detector data is performed for each of the nine additional sets of events.

6.3 Near detector data

  • The authors analysis uses all the near detector data recorded between May 2005 and July 2007.
  • These figures show events from the Run I and Run II data, recorded in the low energy beam configuration.
  • This fact suggests that in this energy region the neutrino cross-sections are not correctly modeled.
  • The agreement between the data and the MC simulation significantly improved after the tuning, especially for the invariant mass squared variable that is used to select the RES events.
  • The reconstructed energy spectra for the selected QES, RES, and DIS events are shown in Figures 6.11 and 6.13.

6.4 Summary

  • This chapter examined the near detector νµ charged-current events.
  • The near detector data are utilized to compute this prediction.
  • This usage of the MINOS near detector data results in a partial cancellation of systematic errors from the incorrect modeling of the detector, neutrino flux, and cross-sections, as well as the mismeasurement of event observables [58].
  • The present analysis uses a tuning method.
  • These events are separated into three categories, using the technique described in Chapter 3.

7.2 Parameterization of the cross-section model

  • The neutrino interaction rate in the detector is a function of the product of the neutrino flux and cross-section.
  • The ν̄µ charged-current events are dominated by DIS events.
  • The three separate scale factors for the QES, RES, and DIS events measure the energy scale uncertainty for different regions of neutrino energy.
  • The 6 tuning parameters are summarized in Table 7.1, where the last column lists the uncertainty of these parameters.
  • In addition, their results indicate that simple scale factors for the normalization of the QES and RES events are sufficient to describe the near detector data.

7.3 Fit description and χ2 function

  • This section briefly describes the fit procedure (fit) used to minimize the differences between the data and MC simulation.
  • The χ2 function depends on the flux and cross-section parameters via the MC event importance weights.
  • The weights of data events are always set to unity.
  • The fit is an iterative procedure consisting of these steps: Step 1 Compute the MC event weights using the current tuning parameters.
  • Fill the histograms with the MC and data events.

7.4 Kernel method and χ2 function

  • The χ2 function is a different function of the energy scale and all other parameters, and this difference affects a minimization procedure.
  • The energy scale parameters change reconstructed event energy.
  • A discontinuous χ2 function presents a problem for the MINUIT program because MINUIT searches for a minimum by following a gradient descent.
  • This method introduces a finite event width so that one event can contribute some weight to multiple histogram bins.
  • The kernel smoothing method uses the “kernel” function.

7.5 Tuning results

  • This section presents results of the near detector fits.
  • These tables show the χ2 values and the number of events before (default MC) and after tuning (tuned MC).
  • The number of true RES events is scaled down by approximately 13%.
  • The additional systematic errors for these two parameters are discussed in Section 7.6.
  • Figures 6.11 and 6.13 show the reconstructed energy for the selected QES, RES, and DIS events for the data and tuned MC simulation.

7.6 Study of different fit approaches

  • The fit parameters listed in Table 7.4 are the corrections for the MC simulation.
  • The average interaction probability of the secondary pions with the nucleus was increased by approximately 8% [110, 111].
  • For the previous MC version, the QES scale factor is 1.48 compared to 1.23 for the current MC version.
  • The MC simulation also uses the magnetic field map to simulate the passage of muons through the detector.
  • The Eν formula (defined in Equation 3.1) and EQES formula (defined in Equation 3.5) are used to reconstruct the energy of the selected QES events in the data and MC simulation.

7.7 Summary

  • The systematic errors from the nuclear effects were estimated: approximately ±0.24 for the QES events, and approximately ±0.11 for the RES events.
  • The detector energy scale was adjusted by approximately 2%, within the estimated 5% uncertainty.
  • This chapter discusses the differences in the neutrino interaction rate between the near and far detectors.
  • The selected νµ charged-current events in the far detector are examined by comparing these events with the expected event distributions.

8.1 Extrapolation from near to far detectors

  • The far detector is located 734.3 km away from the near detector.
  • The distance and the direction from the neutrino source to the detector determine the rate of neutrino interactions and predict differences between the two detectors.
  • Tuning methods use the near detector data to adjust the parameters of the MC simulation.
  • The present analysis employs the tuning method, described in Chapter 7.
  • The errors on the predicted far detector event rate were determined to be less than 3%, if the near detector data were used to compute this predicted rate [101].

8.2 Selecting far detector events

  • The far detector events are selected using the expected arrival time of the neutrino beam spills (see Section 2.3).
  • The far detector events, within a 50 µs timing window of the neutrino spill time, are subject to additional analysis cuts.
  • The direction of the reconstructed track at the vertex is required to be within 53.1◦ angle with the detector Z axis.
  • The two methods predicted less than 0.5 background events for the Run I data [58].

8.3 Far detector data

  • The MINOS collaboration follows a blind analysis policy for the far detector events produced by the NuMI neutrino beam.
  • This policy was implemented to minimize a potential physicist bias.
  • Then, simple event distributions were examined for the entire data set to check for any potential problems [46, 52, 58].
  • The present analysis uses the subset of the νµ chargedcurrent events2 selected by the collaboration for the recent measurement of the muon neutrino disappearance [46].

8.4 Summary

  • The event selection procedure for the far detector events was described.
  • The selection requirements are similar to the requirements for the near detector events, with three exceptions.
  • The selected far detector data events are accurately described using the oscillation hypothesis.
  • This exposure includes only the data from the low energy beam configuration.
  • Assuming there are no neutrino oscillations, the expected number of νµ charged-current events is 910.

9.1 Description of the fit method

  • The data events are analyzed using a simple two-neutrino oscillation hypothesis.
  • The procedure to select νµ charged-current events removes many ντ charged-current and νe charged-current events because these interactions look similar to background neutral-current events.
  • The computation of the χ2 function defined in Equation 9.2 depends on histogram binning for the reconstructed neutrino energy.
  • This fit does not separate the DIS events from the QES and RES events.

9.2 Measurement of neutrino oscillations parameters

  • The two oscillation parameters, |∆m2| and sin2 2θ, are derived from a fit to the selected νµ charged-current events.
  • Figures 9.4 and 9.6 show the ratios of the data spectra and the os- cillated MC spectra to the MC spectra without oscillations as a function of the reconstructed neutrino energy.
  • Figure 9.8 shows the allowed 68% C.L. regions for the primary and secondary fit methods.
  • The kinematic formula reduces the energy reconstruction bias for the selected QES events.
  • This comparison of the two approaches suggests a presence of systematic effect possiblly not modeled by the MC simulation.

9.3 Statistically independent pseudo-experiments

  • The fit procedure is tested using MC pseudo-experiments.
  • The pseudo-experiments are created following this procedure: 1. Apply the corrections derived in Chapter 7 to all the MC events.
  • These oscillation parameters are listed in Table 9.1.
  • These pseudo experiments are fitted using the primary and secondary fit methods.

9.4 Treatment of the systematic errors

  • The systematic uncertainties can be broadly separated into two cate- gories.
  • These 1200 pseudo-experiments are compared to the 100 pseudo-experiments without the systematic uncertainties.
  • This set is generated by requiring simultaneous ± shifts for the 6 systematic uncertainties, described above.
  • The systematic error of measuring the oscillation parameters is com- puted using differences between two best fit parameters.

9.5 MINOS results

  • The oscillation results obtained in this dissertation are compared with the latest published result by the MINOS collaboration [46].
  • For the MINOS result, Figure 9.21 shows the reconstructed energy spectrum for the far detector data, the MC prediction without oscillations, and the MC spectrum weighted by the best fit oscillation parameters.
  • Two analyses are compatible and have measured the same values for the oscillation parameters.
  • The systematic errors measured with the present analysis are larger than the errors for the MINOS analysis.

9.6 Conclusions

  • This thesis presented a complete analysis of the MINOS near and far detector data.
  • The study of the neutrino interactions in the near detector led to the development of a simple method to identify quasi-elastic, resonance production, and deep inelastic scattering νµ charged-current events.
  • The fit to the far detector data was performed using the oscillation hypothesis for the two massive neutrinos.
  • The fit gives an excellent description of the data.

Did you find this useful? Give us your feedback more

Content maybe subject to copyright    Report

Rustem Ospanov

The Dissertation Committee for Rustem Ospanov
certifies that this is the approved version of the following dissertation:
A measurement of muon neutrino disappearance with
the MINOS detectors and NuMI beam
Karol Lang, Supervisor
Duane Dicus
Karl Gebhardt
Sacha Kopp
Jack Ritchie

A measurement of muon neutrino disappearance with
the MINOS detectors and NuMI beam
Rustem Ospanov, B.S., M.A.
Presented to the Faculty of the Graduate School of
The University of Texas at Austin
in Partial Fulfillment
of the Requirements
for the Degree of
December 2008

Dedicated to my family.

A measurement of muon neutrino disappearance with
the MINOS detectors and NuMI beam
Publication No.
Rustem Ospanov, Ph.D.
The University of Texas at Austin, 2008
Supervisor: Karol Lang
MINOS is a long-baseline two-detector neutrino oscillation experiment
that uses a high intensity muon neutrino beam to investigate the phenomena
of neutrino oscillations. The neutrino beam is produced by the NuMI facility
at Fermilab, Batavia, Illinois, and is observed at near and far detectors placed
734 km apart. The neutrino interactions in the near detector are used to
measure the initial muon neutrino flux. The vast majority of neutrinos travel
through the near detector and Earth matter without interactions. A fraction
of muon neutrinos oscillate into other flavors resulting in the disappearance of
muon neutrinos at the far detector. This thesis presents a measurement of the
muon neutrino oscillation parameters in the framework of the two-neutrino
oscillation hypothesis.

Figures (146)
More filters

Journal Article

125 citations

01 Jan 2000

119 citations

Journal ArticleDOI
P. Adamson1, C. Andreopoulos2, K. E. Arms3, R. Armstrong4  +165 moreInstitutions (33)
Abstract: The energy dependence of the neutrino-iron and antineutrino-iron inclusive charged-current cross sections and their ratio have been measured using a high-statistics sample with the MINOS near detector exposed to the NuMI beam from the main injector at Fermilab. Neutrino and antineutrino fluxes were determined using a low hadronic energy subsample of charged-current events. We report measurements of ν-Fe (ν -Fe) cross section in the energy range 3–50 GeV (5–50 GeV) with precision of 2%–8% (3%–9%) and their ratio which is measured with precision 2%–8%. The data set spans the region from low energy, where accurate measurements are sparse, up to the high-energy scaling region where the cross section is well understood.

105 citations

Cites background from "A measurement of muon neutrino disa..."

  • ...8: An illustration defining the Relative Angle variable[52]....


  • ...8 An illustration defining the Relative Angle variable[52]....


Posted Content
01 Jan 2008
Abstract: The OPERA neutrino detector at the underground Gran Sasso Laboratory (LNGS) was designed to perform the first detection of neutrino oscillations in appearance mode, through the study of nu_mu to nu_tau oscillations. The apparatus consists of a lead/emulsion-film target complemented by electronic detectors. It is placed in the high-energy, long-baseline CERN to LNGS beam (CNGS) 730 km away from the neutrino source. In August 2006 a first run with CNGS neutrinos was successfully conducted. A first sample of neutrino events was collected, statistically consistent with the integrated beam intensity. After a brief description of the beam and of the various sub-detectors, we report on the achievement of this milestone, presenting the first data and some analysis results.

82 citations

Journal ArticleDOI
Abstract: The MINOS experiment has used the world’s most powerful neutrino beam to make precision neutrino oscillation measurements. By observing the disappearance of muon neutrinos, MINOS has made the world’s most precise measurement of the larger neutrino mass splitting and has measured the neutrino mixing angle . Using a dedicated antineutrino beam, MINOS has made the first direct precision measurements of the corresponding antineutrino parameters. A search for and appearance has enabled a measurement of the mixing angle . A measurement of the neutral-current interaction rate has confirmed oscillation between three active neutrino flavours. MINOS will continue as MINOS+ in an upgraded beam with higher energy and intensity, allowing precision tests of the three-flavour neutrino oscillation picture, in particular a very sensitive search for the existence of sterile neutrinos.

40 citations

Cites methods from "A measurement of muon neutrino disa..."

  • ...These variables are input into a k-nearest-neighbour algorithm, which calculates a single discrimination variable, shown in Figure 6 [49]....


More filters

Journal ArticleDOI
Abstract: (2004). The Elements of Statistical Learning: Data Mining, Inference, and Prediction. Journal of the American Statistical Association: Vol. 99, No. 466, pp. 567-567.

7,980 citations

Journal ArticleDOI
TL;DR: The multidimensional binary search tree (or k-d tree) as a data structure for storage of information to be retrieved by associative searches is developed and it is shown to be quite efficient in its storage requirements.
Abstract: This paper develops the multidimensional binary search tree (or k-d tree, where k is the dimensionality of the search space) as a data structure for storage of information to be retrieved by associative searches. The k-d tree is defined and examples are given. It is shown to be quite efficient in its storage requirements. A significant advantage of this structure is that a single data structure can handle many types of queries very efficiently. Various utility algorithms are developed; their proven average running times in an n record file are: insertion, O(log n); deletion of the root, O(n(k-1)/k); deletion of a random node, O(log n); and optimization (guarantees logarithmic performance of searches), O(n log n). Search algorithms are given for partial match queries with t keys specified [proven maximum running time of O(n(k-t)/k)] and for nearest neighbor queries [empirically observed average running time of O(log n).] These performances far surpass the best currently known algorithms for these tasks. An algorithm is presented to handle any general intersection query. The main focus of this paper is theoretical. It is felt, however, that k-d trees could be quite useful in many applications, and examples of potential uses are given.

6,444 citations

"A measurement of muon neutrino disa..." refers methods in this paper

  • ...A significant improvement in the performance of the knn algorithm was attained by using the kd-tree data structure [98, 99]....


Journal ArticleDOI

4,813 citations

"A measurement of muon neutrino disa..." refers background or methods in this paper

  • ...The discontinuity problem is solved using a kernel smoothing method [109]....


  • ...This function is typically a narrow Gaussian distribution or a similarly shaped function [109]....


Journal ArticleDOI
Abstract: The effect of coherent forward scattering must be taken into account when considering the oscillations of neutrinos traveling through matter. In particular, for the case of massless neutrinos for which vacuum oscillations cannot occur, oscillations can occur in matter if the neutral current has an off-diagonal piece connecting different neutrino types. Applications discussed are solar neutrinos and a proposed experiment involving transmission of neutrinos through 1000 km of rock.

2,875 citations

"A measurement of muon neutrino disa..." refers methods in this paper

  • ...The original theory was modified by Lincoln Wolfenstein, Stanislav Mikheyev, and Alexei Smirnov [35,35] to account for resonant effects introduced by the interaction of neutrinos with the matter inside the sun....


Journal ArticleDOI
TL;DR: An algorithm and data structure are presented for searching a file containing N records, each described by k real valued keys, for the m closest matches or nearest neighbors to a given query record.
Abstract: An algorithm and data structure are presented for searching a file containing N records, each described by k real valued keys, for the m closest matches or nearest neighbors to a given query record. The computation required to organize the file is proportional to kNlogN. The expected number of records examined in each search is independent of the file size. The expected computation to perform each search is proportional to logN. Empirical evidence suggests that except for very small files, this algorithm is considerably faster than other methods.

2,791 citations

"A measurement of muon neutrino disa..." refers methods in this paper

  • ...A significant improvement in the performance of the knn algorithm was attained by using the kd-tree data structure [98, 99]....


Related Papers (5)
Y. Ashie, J. Hosaka, K. Ishihara, Yoshitaka Itow, J. Kameda, Yusuke Koshio, A. Minamino, C. Mitsuda, M. Miura, Shigetaka Moriyama, Masayuki Nakahata, Toshio Namba, R. Nambu, Y. Obayashi, Masato Shiozawa, Yasunari Suzuki, Y. Takeuchi, K. Taki, Shinya Yamada, Masaki Ishitsuka, Takaaki Kajita, K. Kaneyuki, Shoei Nakayama, A. Okada, Ko Okumura, T. Ooyabu, C. Saji, Y. Takenaga, Shantanu Desai, E. Kearns, S. Likhoded, J. L. Stone, L. R. Sulak, C. W. Walter, W. Wang, M. Goldhaber, David William Casper, J. P. Cravens, W. Gajewski, W. R. Kropp, D. W. Liu, S. Mine, Michael B. Smy, H. W. Sobel, C. W. Sterner, Mark R. Vagins, K. S. Ganezer, John Hill, W. E. Keig, J. S. Jang, J. Y. Kim, I. T. Lim, R. W. Ellsworth, S. Tasaka, G. Guillian, A. Kibayashi, John G. Learned, S. Matsuno, D. Takemori, M. D. Messier, Y. Hayato, A. K. Ichikawa, T. Ishida, T. Ishii, T. Iwashita, T. Kobayashi, Tomoyuki Maruyama, K. Nakamura, K. Nitta, Yuichi Oyama, Makoto Sakuda, Y. Totsuka, Atsumu Suzuki, Masaya Hasegawa, K. Hayashi, T. Inagaki, I. Kato, H. Maesaka, Taichi Morita, Tsuyoshi Nakaya, K. Nishikawa, T. Sasaki, S. Ueda, Shoji Yamamoto, Todd Haines, Todd Haines, S. Dazeley, S. Hatakeyama, R. Svoboda, E. Blaufuss, J. A. Goodman, G. W. Sullivan, D. Turcan, Kate Scholberg, Alec Habig, Y. Fukuda, C. K. Jung, T. Kato, Katsuhiro Kobayashi, Magdalena Malek, C. Mauger, C. McGrew, A. Sarrat, E. Sharkey, C. Yanagisawa, T. Toshito, Kazumasa Miyano, N. Tamura, J. Ishii, Y. Kuno, Y. Nagashima, M. Takita, Minoru Yoshida, S. B. Kim, J. Yoo, H. Okazawa, T. Ishizuka, Y. Choi, H. Seo, Y. Gando, Takehisa Hasegawa, Kunio Inoue, J. Shirai, A. Suzuki, Masatoshi Koshiba, Y. Nakajima, Kyoshi Nishijima, T. Harada, Hirokazu Ishino, R. Nishimura, Y. Watanabe, D. Kielczewska, D. Kielczewska, J. Zalipska, H. G. Berns, R. Gran, K. K. Shiraishi, A. L. Stachyra, K. Washburn, R. J. Wilkes