Lepton-Nucleus Scattering Measurements for

Neutrino Interactions and Oscillations

by

Afroditi Papadopoulou
B.S., National Kapodistrian University of Athens (2016)
Submitted to the Department of Physics
arXiv:2301.03114v1 [hep-ex] 8 Jan 2023

in partial fulfillment of the requirements for the degree of

Doctor of Philosophy
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
May 2022
© Massachusetts Institute of Technology 2022. All rights reserved.

Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Department of Physics
May 12, 2022

Certified by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Or Hen
Associate Professor
Thesis Supervisor

Accepted by . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Deepto Chakrabarty
Associate Department Head of Physics, MIT
2
 Lepton-Nucleus Scattering Measurements for
Neutrino Interactions and Oscillations
by
Afroditi Papadopoulou

B.S., National Kapodistrian University of Athens (2016)

Submitted to the Department of Physics

on May 12, 2022, in partial fulfillment of the
requirements for the degree of
Doctor of Philosophy

Abstract
Currently running and forthcoming precision neutrino oscillation experiments aim
to unambiguously determine the neutrino mass ordering, the charge-parity violating
phase in the lepton sector and the possible existence of physics Beyond the Standard
Model. To have an understanding of all the effects necessary for the success of these
experiments, lepton-nucleus interactions must be modeled in unprecedented detail.
With this thesis, expertise in both neutrino and electron cross-section modeling and
analysis was leveraged in order to make fundamental and critical improvements to
our understanding of these interactions. The outlined work takes a significant step
towards this high-precision measurement era with three complementary approaches.
Cross sections are reported using neutrino data sets from the MicroBooNE liquid
argon time projection chamber detector at Fermi National Laboratory, as well as
electron scattering data from the CLAS detector at Thomas Jefferson National Labo-
ratory. Furthermore, the modeling development of the commonly used GENIE event
generator is presented.

3
Thesis Supervisor: Or Hen
Title: Associate Professor

4
Acknowledgments
First and foremost, my eternal gratitude goes to my advisor, Prof. Or Hen. Or has
provided me with all the resources necessary to proceed in my academic career. I am
also grateful to Prof. Lawrence Weinstein. Their guidance has been crucial in my
formation as a physicist.
I’m grateful to all the members of our group for their invaluable support and to
all the MIT graduate students in my year. Many thanks to all the members of the
MicroBooNE, “Electrons-For-Neutrinos”, CLAS, GENIE, and GlueX collaborations
for all their input and guidance. I’m also grateful to my entire family and all my
friends for their continuous support and encouragement.

5
6
Contents

1 Introduction 11
1.1 Neutrinos In The Standard Model . . . . . . . . . . . . . . . . . . . . 11
1.2 Neutrino Oscillations . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3 Long-Baseline Accelerator-Based Neutrino
Oscillation Experiments . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.4 Neutrino-Nucleus Interactions . . . . . . . . . . . . . . . . . . . . . . 24
1.5 Connections To Electron Scattering . . . . . . . . . . . . . . . . . . . 28
1.6 Thesis Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31

2 The MicroBooNE Experiment At Fermi National Laboratory 33

2.1 Fermilab Neutrino Beamlines . . . . . . . . . . . . . . . . . . . . . . 33
2.2 The MicroBooNE Detector . . . . . . . . . . . . . . . . . . . . . . . . 36
2.3 Liquid Argon Time Projection Chambers . . . . . . . . . . . . . . . . 38
2.4 Scintillation Light . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
2.5 Hardware And Software Triggers . . . . . . . . . . . . . . . . . . . . 44
2.6 Optical Event Reconstruction . . . . . . . . . . . . . . . . . . . . . . 46
2.7 TPC Signal Processing . . . . . . . . . . . . . . . . . . . . . . . . . . 47
2.8 Pandora Reconstruction Framework . . . . . . . . . . . . . . . . . . . 48
2.9 Cosmic Overlay Simulation . . . . . . . . . . . . . . . . . . . . . . . . 52

3 MicroBooNE Quasielastic-like Cross-Section Results

[Phys. Rev. Lett. 125, 201803 (2020)] 55

7
 3.1 First Measurement of Differential Charged Current Quasielastic-like
Scattering Cross Sections . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.1.1 Quasielastic-like Neutrino Data Analysis . . . . . . . . . . . . 55
3.1.2 Quasielastic-like Cross-Section Results . . . . . . . . . . . . . 63
3.1.3 Quasielastic-like Cross-Section Analysis Conclusions . . . . . . 68
3.2 First Multidimensional Measurement Of Kinematic Imbalance Cross
Sections On Argon . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
3.2.1 Kinematic Imbalance Neutrino Data Analysis . . . . . . . . . 68
3.2.2 Cross-Section Extraction Technique . . . . . . . . . . . . . . . 74
3.2.3 Event Generator Modeling And Configurations . . . . . . . . . 78
3.2.4 Kinematic Imbalance Differential Cross-Section Results . . . . 79
3.2.5 Kinematic Imbalance Cross-Section Analysis Conclusions . . . 87
3.3 Prospects With Future Neutrino Experiments . . . . . . . . . . . . . 87

4 Inclusive Electron Scattering And The GENIE Event Generator

[Phys. Rev. D 103, 113003 (2021)] 91
4.1 Electron-Nucleus Modeling Development . . . . . . . . . . . . . . . . 91
4.2 Inclusive Electron Scattering Data Comparisons . . . . . . . . . . . . 101
4.3 Implications For Neutrinos . . . . . . . . . . . . . . . . . . . . . . . . 109

5 The CLAS Electrons-For-Neutrinos Experiment At Thomas Jeffer-

son National Laboratory 115
5.1 Continuous Electron Beam Accelerator Facility . . . . . . . . . . . . . 115
5.2 The CEBAF Large Acceptance Spectrometer . . . . . . . . . . . . . . 116
5.3 Toroidal Magnet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.4 Drift Chambers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
5.5 Electromagnetic Calorimeter . . . . . . . . . . . . . . . . . . . . . . . 119
5.6 Cherenkov Counters . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
5.7 Time-Of-Flight Detector . . . . . . . . . . . . . . . . . . . . . . . . . 121
5.8 E2a Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122

8
6 Electrons-For-Neutrinos Results
[Nature 599, 565–570 (2021)] 125
6.1 Electron Data Mining Analysis . . . . . . . . . . . . . . . . . . . . . 125
6.2 Incident Energy Reconstruction Results . . . . . . . . . . . . . . . . . 139
6.3 Kinematic Imbalance Results . . . . . . . . . . . . . . . . . . . . . . 144
6.4 Electrons-For-Neutrinos Conclusions . . . . . . . . . . . . . . . . . . 153
6.5 Prospects With CLAS12 . . . . . . . . . . . . . . . . . . . . . . . . . 154

7 Summary 157

8 Appendices 159
8.1 Total Struck Nucleon Momentum Derivation . . . . . . . . . . . . . . 159
8.2 Wiener SVD Regularization Technique . . . . . . . . . . . . . . . . . 163
8.3 Electrons-For-Neutrinos Fiducials . . . . . . . . . . . . . . . . . . . . 171
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200

9
10
Chapter 1

Introduction

1.1 Neutrinos In The Standard Model

The Standard Model (SM) of particle physics was developed to describe the parti-
cles that are considered to be fundamental and their interactions [1]. Though an
extremely accurate theory, we are already aware of particles exhibiting behaviors not
predicted within the scope of the SM. These peculiar particles are called neutrinos,
which travel enormous distances before they interact with matter, exclusively via
weak interactions. The recent realization of neutrino oscillations was a historic dis-
covery [2, 3]. The SM had been incredibly successful, yet it requires neutrinos to
be massless. The new observations clearly showed that neutrinos oscillate between
different identities. This behavior is driven by their non-zero masses and is indicative
of new physics Beyond the Standard Model.
Though very challenging to detect them, the study of neutrinos is a promising
venue towards the generalisation of the SM. The latter, initially developed more than
40 years ago, introduced the notion of fundamental particles. Those particles are
either the building blocks of matter referred to as fermions with semi-integer spins, or
mediators of the interactions referred to as bosons with integer spins. The collection
of the fundamental particles and the carriers is shown in figure 1-1.
Prior to the SM, the electroweak theory incorporated neutrinos (𝜈𝑒 , 𝜈𝜇 , 𝜈𝜏 ) as
left-handed partners of the three families of charged leptons (e, 𝜇, 𝜏 ) and the corre-

11
Figure 1-1: The Standard Model of particle physics illustrating the three generations
of fermions, the gauge bosons, and the scalar Higgs boson. Figure adapted from
ScienceAlert [4].

sponding anti-neutrinos as right-handed partners, which is illustrated in figure 1-2.

The experimental verification that neutrinos (anti-neutrinos) are left-handed (right-
handed) was established with the Goldhaber experiment [5].

Figure 1-2: Graphic illustrating the left-handed (right-handed) nature of neutrinos

(anti-neutrinos) via the orientation of the momentum-spin vectors. Figure adapted
from [6].

In the context of the SM, neutrinos are assumed to be massless and their individ-
ual lepton number is conserved. Under the assumption that neutrinos are massless,
only the left-handed component of the Dirac spinor interacts via the weak force and

12
right-handed components are completely absent. Prior to the discovery of neutrino
oscillations, there was no experimental evidence of violation of the individual lepton
number. A typical example of such searches is the muon- and electron-number violat-
ing decay 𝜇± → 𝑒± + 𝛾, where only upper limits have been placed on the branching
ratio [7]. Yet, the experimental discovery of neutrino oscillations established that
the conservation of individual lepton number is not universal. That raises a wealth
of questions related to the possible right-handed nature and the non-zero mass of
neutrinos.
If neutrinos are massive, helicity is not exactly conserved. A massive neutrino
can have its spin and momentum anti-aligned, and that would correspond to a left-
handed neutrino. However, in the reference frame of an observer travelling faster than
the neutrino itself, the neutrino spin and momentum can be aligned, and that would
correspond to a right-handed particle. On the other hand, if neutrinos are massless,
they must travel at the speed of light. Since there exists no reference frame travelling
faster than that, the neutrino helicity cannot change.
At the same time, the neutrino masses are significantly smaller compared to other
particles. The latest measurements by KATRIN indicate that electrons are 𝒪(106 )
more massive than neutrinos [8]. Massless particles require no extra terms to be
added in the SM Lagrangian. On the other hand, the tiny neutrino masses constitute
a riddle that requires fine tuning of the additional Lagrangian terms. Instead, many
theorists argue in favor of a fundamental reason why the neutrino masses are so small.
In order to accommodate a non-zero neutrino mass, there are two potential ap-
proaches. The first one includes new particles, namely Dirac neutrinos, and the second
introduces a new particle type, namely Majorana neutrinos [9].
A Dirac neutrino can acquire mass via its coupling to the Higgs field. Those parti-
cles that interact with the Higgs field change helicity and, thus, left-handed particles
become right-handed and vice versa. Experimental results to-date indicate that in-
teracting neutrinos are left-handed [5]. A potential extension of the SM includes such
right-handed neutrinos that obtain mass via the Higgs field. However, these neutri-
nos do not have electroweak charge, thus they interact only via the mixing with the

13
left-handed counterpart. If indeed neutrinos are Dirac particles and obtain their mass
due to the coupling with the Higgs field, their masses should be comparable to the
other fermions predicted by the SM. However, the neutrino interaction coupling can
be tuned to retrieve masses that are comparable with the experimental observations.
Another possibility would be the introduction of Majorana particles. With this
approach, there is no distinction between neutrinos and anti-neutrinos, with the latter
being plausible since neutrinos are electrically neutral. Massive neutrinos can be
accommodated in this extension of the SM. Earlier the argument that an observer
traveling at the speed of light might observe a flip of the neutrino helicity was used.
However, if neutrinos are their own anti-particles, there is no helicity change within
this massive neutrino hypothesis. The simplest SM extension with Majorana-like
particles is the type 1 seesaw mechanism [10]. In this framework, a left-handed
neutrino interacts with the Higgs boson and a really heavy right-handed neutrino is
briefly produced. The latter further interacts with the Higgs field to produce a light
left-handed Majorana neutrino.

1.2 Neutrino Oscillations

Neutrinos are produced in specific flavor eigenstates, namely 𝜈𝑒 , 𝜈𝜇 or 𝜈𝜏 . The
neutrino mass eigenstates can be expressed as a superposition of flavor eigenstates
and vice versa. The flavor eigenstates evolve as a function of the distance that neu-
trinos travel, the neutrino energy and the neutrino mass differences squared. Such
an evolution is referred to as “neutrino oscillation” and is a natural consequence of
the fact that a neutrino flavor state is composed of multiple mass eigestates. For
simplicity, the derivation of the formalism for the two-neutrino oscillation case with
masses 𝑚𝑖 for i ∈ {1,2} is outlined here. The neutrino wavefunction is treated as a
quantum-mechanical plane wave 𝜓 that evolves in time and space as:

𝜓(𝐿) = 𝜓(0) · 𝑒𝑖·𝑝·𝑥 , (1.1)

with 𝑝·𝑥 = 𝐸𝑡− 𝑝⃗ ·⃗𝑥 corresponding to the Lorentz-invariant phase and with (𝐸,⃗𝑝 )

14
being the energy and the three-momentum, respectively. For a particle of mass 𝑚,
its momentum 𝑝 can be obtained in the highly relativistic limit where 𝐸 ≫ 𝑚 via a
Taylor expansion as:

√
𝑝= 𝐸 2 − 𝑚2 ≈ 𝐸 − 𝑚2 /2𝐸. (1.2)

If a neutrino travels some distance 𝐿 in vacuum, a phase shift is introduced in its

wavefunction given by equation 1.1, and equation 1.2 becomes:

2
𝜈𝑖 (𝐿) = 𝜈𝑖 (0) · 𝑒−𝑖·𝑚𝑖 ·𝐿/2𝐸 (1.3)

where all the constants have been absorbed in the global phase 𝜈𝑖 (0). It is impor-
tant to stress the fact that these expressions are accurate for plane waves propagat-
ing in vacuum. Additional phase shifts have to be introduced when neutrinos travel
through a high density material.

In the two-neutrino case, the flavor eigenstates 𝜈𝑒 and 𝜈𝜇 can be expressed as a

linear superposition of the mass eigenstates 𝜈1 and 𝜈2 :

𝜈𝑒 = 𝑐𝑜𝑠𝜃 · 𝜈1 + 𝑠𝑖𝑛𝜃 · 𝜈2 ,
(1.4)
𝜈𝜇 = −𝑠𝑖𝑛𝜃 · 𝜈1 + 𝑐𝑜𝑠𝜃 · 𝜈2 ,

where 𝜃 is the mixing angle between the two states. Consider now a pure beam
of electron neutrinos produced at the source, effectively at a distance of 𝐿 = 0. The
wavefunction evolves following the formalism in equation 1.3,

2 2
𝜈𝑒 (𝐿) = 𝑐𝑜𝑠𝜃 · 𝑒−𝑖·𝑚1 ·𝐿/2𝐸 · 𝜈1 (0) + 𝑠𝑖𝑛𝜃 · 𝑒−𝑖·𝑚2 ·𝐿/2𝐸 · 𝜈2 (0). (1.5)

Given that in our detectors the products of weak interactions are reconstructed

15
based on the neutrino flavor, equation 1.5 needs to be rewritten using the flavor basis,

2 2
𝜈𝑒 (𝐿) = [𝑐𝑜𝑠2 𝜃 · 𝑒−𝑖·𝑚1 ·𝐿/2𝐸 + 𝑠𝑖𝑛2 𝜃 · 𝑒−𝑖·𝑚2 ·𝐿/2𝐸 ] · 𝜈𝑒 (0)
(1.6)
𝑖·𝑚21 ·𝐿/2𝐸 𝑖·𝑚22 ·𝐿/2𝐸
−𝑠𝑖𝑛𝜃 · 𝑐𝑜𝑠𝜃 · [𝑒 −𝑒 ] · 𝜈𝜇 (0)

The probability of detecting a neutrino of a given flavor is obtained via the square
of the amplitude,

∆𝑚2 𝐿
𝑃𝜈𝑒 →𝜈𝑒 = ⟨𝜈𝑒 (𝐿)|𝜈𝑒 (𝐿)⟩ = 1 − 𝑠𝑖𝑛2 (2𝜃)𝑠𝑖𝑛2 ( ), (1.7)
4𝐸

∆𝑚2 𝐿
𝑃𝜈𝑒 →𝜈𝜇 = ⟨𝜈𝑒 (𝐿)|𝜈𝜇 (𝐿)⟩ = 𝑠𝑖𝑛2 (2𝜃)𝑠𝑖𝑛2 ( ), (1.8)
4𝐸

where ∆𝑚2 = 𝑚22 − 𝑚21 is the neutrino mass difference squared. Provided that
𝜃 ̸= 0 or 𝜋/2 and ∆𝑚2 ̸= 0 for oscillations to take place, the neutrino beam evolves
as a function of 𝐿/𝐸. The amplitude of this oscillation is given by 𝑠𝑖𝑛2 (2𝜃). The
wavelength, expressed in commonly used units, is obtained by

∆𝑚2 [𝑒𝑉 2 ]𝐿 [𝑘𝑚]

1.27 . (1.9)
𝐸 [𝐺𝑒𝑉 ]

The experimental observation of neutrino oscillations serves as proof of evidence

that the initial state indeed undergoes a phase shift. The explanation for such a shift is
that at least some neutrinos have non-zero mass and that the transformation between
the mass and the flavor eigenstates involves a non-zero mixing angle. Equation 1.8
gives the neutrino oscillation probability for the simplified case of only two types of
neutrinos. The full three-flavor probability for 𝜈𝜇 → 𝜈𝑒 oscillation in vacuum is given
by

∆𝑚213 𝐿
𝑃𝜈𝜇 →𝜈𝑒 (𝐸, 𝐿) ≈ 𝐴 sin2 (1.10)
(︂4𝐸 2
∆𝑚213 𝐿
)︂
∆𝑚13 𝐿
−𝐵 cos + 𝛿𝐶𝑃 sin ,
4𝐸 4𝐸

16
 where ∆𝑚213 = 𝑚21 − 𝑚23 is the neutrino mass difference squared that determines
the oscillation wavelength as a function of 𝐿/𝐸 and 𝛿𝐶𝑃 is the charge-parity (CP)
symmetry violating phase [11–13]. The coefficients 𝐴 and 𝐵 depend primarily on the
neutrino oscillation mixing angles,

sin 2𝜃12 sin 2𝜃23 ∆𝑚221 𝐿 2

𝐴 = sin2 𝜃23 sin2 2𝜃13 , 𝐵 = − sin sin 2𝜃13 . (1.11)
2 sin 𝜃13 4𝐸

The different flavor neutrinos (labelled 𝜈𝑒 , 𝜈𝜇 and 𝜈𝜏 ) are linear combinations of

the different mass neutrinos labelled 1, 2, 3.
The outlined neutrino mixing can be generalized to N neutrino eigenstates. The
probability of finding an 𝛼 flavor eigenstate neutrino in a flavor eigenstate 𝛽, after
traveling some distance 𝐿, is given by:

∆𝑚2𝑗𝑖 𝐿
(︂[︂ ]︂ )︂
∑︁
* * 1.27 𝐺𝑒𝑉
𝑃𝛼→𝛽 = 𝛿𝛼𝛽 − 4 𝑅𝑒[𝑈𝛼𝑖 𝑈𝛽𝑖 𝑈𝛼𝑖 𝑈𝛽𝑗 ]𝑠𝑖𝑛2
𝑗>𝑖
𝑒𝑉 2 𝑘𝑚 𝐸
(1.12)
∆𝑚2𝑗𝑖 𝐿
(︂[︂ ]︂ )︂
∑︁
* * 2.54 𝐺𝑒𝑉
+2 𝐼𝑚[𝑈𝛼𝑖 𝑈𝛽𝑖 𝑈𝛼𝑖 𝑈𝛽𝑗 ]𝑠𝑖𝑛
𝑗>𝑖
𝑒𝑉 2 𝑘𝑚 𝐸

where U is the N×N unitary neutrino mixing matrix and ∆𝑚2𝑗𝑖 = 𝑚2𝑗 − 𝑚2𝑖 [14].
The corresponding antineutrino oscillation probability can be obtained by replacing
𝑈 → 𝑈 †.
Within the widely accepted neutrino model, there exist three active neutrinos [15],
resulting into two squared mass splittings, ∆𝑚221 and ∆𝑚232 . The corresponding 3 × 3
mixing matrix, referred to as the Pontecorvo-Maki-Nakagawa-Sakata (PNMS) matrix,
relates the mass eigenstates (𝜈1 , 𝜈2 , 𝜈3 ) to the flavor eigenstates (𝜈𝑒 , 𝜈𝜇 , 𝜈𝜏 ):

⎛ ⎞ ⎛ ⎞ ⎛ ⎞
⎜ 𝜈𝑒 ⎟ ⎜ 𝑈𝑒1 𝑈𝑒2 𝑈𝑒3 ⎟ ⎜𝜈1 ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
(1.13)
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎜ 𝜇 ⎟ ⎜ 𝜇1 𝑈𝜇2 𝑈𝜇3 ⎟ · ⎜𝜈2 ⎟
⎜𝜈 ⎟ = ⎜𝑈 ⎟ ⎜ ⎟
⎜ ⎟ ⎜ ⎟ ⎜ ⎟
⎝ ⎠ ⎝ ⎠ ⎝ ⎠
𝜈𝜏 𝑈𝜏 1 𝑈𝜏 2 𝑈𝜏 3 𝜈3

17
 The mixing of the mass-flavor eigenstates is parametrised with three mixing angles
(𝜃12 , 𝜃23 , 𝜃13 ) and the 𝛿𝐶𝑃 violating phase. The best measured values of the these
angular parameters to date are obtained from global fits as documented in the latest
Review of Particle Physics edition [1] and listed in table 1.1.

Table 1.1: Summary table of the 3𝜈 oscillation parameters. The 1𝜎 intervals for both
the case normal are inverse ordering are shown. Table adapted from [1].

Normal Ordering Inverted Ordering

𝑠𝑖𝑛2 𝜃12 0.320+0.020
−0.016 0.320+0.020
−0.016
𝑠𝑖𝑛2 𝜃23 0.547+0.020
−0.030 0.551+0.018
−0.030
𝑠𝑖𝑛2 𝜃13 (2.160+0.083
−0.069 ) · 10
−2
(2.220+0.074
−0.076 ) · 10
−2

∆𝑚221 (7.55+0.02
−0.16 ) · 10
−5
𝑒𝑉 2 (7.55+0.02
−0.16 ) · 10
−5
𝑒𝑉 2
∆𝑚232 (2.42+0.03
−0.03 ) · 10
−3
𝑒𝑉 2 (−2.50+0.04
−0.03 ) · 10
−3
𝑒𝑉 2

The 𝛿𝐶𝑃 phase and the neutrino mass hierarchy (𝑚2 < 𝑚3 or the other way
around) are still not determined. A potential way to resolve these open questions is
via the Mikheyev-Smirnov-Wolfenstein (MSW) effect [16]. In the presence of matter,
the neutrino wavefunction propagation is modified. When a neutrino travels through
a dense material, the energy-momentum relationship is affected via coherent interac-
tions with matter particles. Given that regular matter on Earth contains electrons,
but not muons or taus, charged current interactions with the medium only affect the
electron neutrino propagation. This interaction flavor dependence results in mea-
surable changes in the contribution of the different flavors. Thus, the MSW effect
is exploited to perform long-baseline neutrino experiments. The Deep Underground
Neutrino Experiment (DUNE) is a forthcoming multi-billion-dollar international ex-
periment aiming to resolve the aforementioned open questions. To successfully achieve
that goal, DUNE will utilize an intense muon beam, with a near detector located at
Fermi National Laboratory IL and a far detector ≈ 1300 km away from the neutrino
source at Sanford Underground Research Facility SD.

18
 Before DUNE starts taking data, a number of other neutrino experiments will
have already provided constraints for 𝛿𝐶𝑃 and the mass hierarchy [17]. For these
results to be obtained, different neutrino oscillation experiments utilize a number of
neutrino sources covering a wide range of energies.
Reactor neutrino experiments exploit the large electron anti-neutrino fluxes pro-
duced in nuclear reactors by 𝛽 decays of heavy nuclei (nuclear fissions of 235
𝑈, 238
𝑈,
239
𝑃 𝑢, 241
𝑃 𝑢). The typical energy scale of reactor 𝜈 𝑒 ’s is a few MeVs.
Atmospheric neutrino experiments take advantage of neutrinos produced by cos-
mic rays. The latter interact with the upper layers of the atmosphere producing
a large flux of pions and kaons. These decay in the atmosphere into muons and
muon neutrinos. These muons might decay into electrons, electron anti-neutrinos,
and muon neutrinos before they reach the Earth. Atmospheric neutrino experiments
aim to detect these muon neutrinos. The energy of these neutrinos spans a really
wide range up to 𝒪(100) GeV. In some cases, like in IceCube [18], even neutrinos
with PeV energies can be detected. Such neutrinos cover distances between 𝒪(10) km
- for neutrinos produced in the upper layers of the atmosphere directly above the de-
tector - to 𝒪(104 ) km - for netrinos that are produced on the other side of the Earth
and travel through the core. Atmospheric experiments provide the current best limit
of ∆𝑚232 = 2.56+0.13
−0.11 × 10 𝑒𝑉 [1].
−3 2

Solar neutrino experiments detect the electron neutrinos produced in the Sun core
due to nuclear fusion processes. Solar neutrino experiments are designed to detect
the 𝜈𝑒 ’s produced via these processes. The currently best limit corresponds to an
extremely small value of ∆𝑚221 = 7.37+0.59 −5
−0.44 × 10 𝑒𝑉
2
[1], much smaller than the
∆𝑚232 splitting mentioned above.
Accelerator-based experiments use muon neutrino beams produced via the decay of
primarily pions produced when a proton beam hits a heavy target. Such experiments
are further classified into appearance and disappearance experiments. The former
search for electron neutrinos oscillated from the initial muon neutrino beam. The
latter look for the reduction of muon neutrino interactions due to oscillations.
Different experiments are designed to be sensitive to different values of ∆𝑚2 by

19
choosing the appropriate 𝐿/𝐸 ratio. Building on equation 1.9, the value of ∆𝑚2 for
which

∆𝑚2 𝐿
≃1 (1.14)
2𝐸

corresponds to the ∆𝑚2 sensitivity of a given experiment.

Neutrino oscillation experiments are further classified depending on the average

value of this 𝐿/𝐸 ratio. Using equation 1.14, short baseline experiments with 𝐿/𝐸 . 1
km/GeV are sensitive to ∆𝑚2 & 1 𝑒𝑉 2 . Long baseline experiments where 𝐿/𝐸 . 103
km/GeV are sensitive to ∆𝑚2 & 10−3 𝑒𝑉 2 .

1.3 Long-Baseline Accelerator-Based Neutrino

Oscillation Experiments

Figure 1-3: The experimental setup of an accelerator-based long-baseline neutrino

oscillation experiment.

Long-baseline accelerator-based experiments consist of a near detector (ND) posi-

tioned close to the neutrino source, and a far detector (FD), which is positioned close
to the oscillation maximum„ as shown in figure 1-3. The number of interactions of

20
flavor 𝛼 in the near detector (𝑁𝑁𝛼 𝐷 ) is obtained as

∫︁
𝑁𝛼𝑁 𝐷 (𝐸𝑟𝑒𝑐𝑜 ) = Φ𝑁 𝐷
𝛼 (𝐸𝑡𝑟𝑢𝑒 )𝜎𝛼 (𝐸𝑡𝑟𝑢𝑒 )𝜖𝛼 (𝐸𝑡𝑟𝑢𝑒 )𝑓
𝑁𝐷
(𝐸𝑡𝑟𝑢𝑒 , 𝐸𝑟𝑒𝑐𝑜 )𝑑𝐸𝑡𝑟𝑢𝑒 (1.15)

where Φ𝑁
𝛼 (𝐸𝑡𝑟𝑢𝑒 ) is the neutrino flux of flavor 𝛼 close to the source, 𝜎𝛼 (𝐸𝑡𝑟𝑢𝑒 )
𝐷

is the cross section for a given flavor 𝛼, 𝜖𝛼 (𝐸𝑡𝑟𝑢𝑒 ) is the reconstruction efficiency for
flavor 𝛼, and 𝑓 𝑁 𝐷 (𝐸𝑡𝑟𝑢𝑒 , 𝐸𝑟𝑒𝑐𝑜 ) is the detector response function, describing how 𝐸𝑡𝑟𝑢𝑒
is mapped to 𝐸𝑟𝑒𝑐𝑜 in the near detector. For 𝜈𝛼 → 𝜈𝛽 oscillations, the number of
events in the far detector can be obtained as

∫︁
𝐹𝐷
𝑁𝛼→𝛽 (𝐸𝑟𝑒𝑐𝑜 ) = Φ𝐹𝛼 𝐷 (𝐸𝑡𝑟𝑢𝑒 )𝑃𝛼→𝛽 (𝐸𝑡𝑟𝑢𝑒 )𝜎𝛽 (𝐸𝑡𝑟𝑢𝑒 )𝜖𝛽 (𝐸𝑡𝑟𝑢𝑒 )𝑓 𝐹 𝐷 (𝐸𝑡𝑟𝑢𝑒 , 𝐸𝑟𝑒𝑐𝑜 )𝑑𝐸𝑡𝑟𝑢𝑒
(1.16)
where 𝑁𝛼→𝛽
𝐹𝐷
(𝐸𝑟𝑒𝑐𝑜 ) is the number of 𝛽 flavor interactions, Φ𝐹𝛼 𝐷 (𝐸𝑡𝑟𝑢𝑒 ) is the neu-
trino flux of flavor 𝛼, 𝑃𝛼→𝛽 (𝐸𝑡𝑟𝑢𝑒 ) is the oscillation probability for 𝛼 → 𝛽, 𝜎𝛽 (𝐸𝑡𝑟𝑢𝑒 )
is the cross section for flavor 𝛽, 𝜖𝛽 (𝐸𝑡𝑟𝑢𝑒 ) is the reconstruction efficiency for flavor
𝛽, and 𝑓 𝐹 𝐷 (𝐸𝑡𝑟𝑢𝑒 , 𝐸𝑟𝑒𝑐𝑜 ) is the detector response function, describing how 𝐸𝑡𝑟𝑢𝑒 is
reconstructed in 𝐸𝑟𝑒𝑐𝑜 .
Accelerator-produced neutrino beams predominantly contain muon neutrinos [19].
Therefore, long-baseline accelerator-based neutrino oscillation experiments focus on
muon neutrino disappearance and electron neutrino appearance studies. The former
are sensitive to the oscillation parameters 𝜃23 and ∆𝑚223 , while the latter have sensi-
tivity to 𝜃13 and 𝛿𝐶𝑃 [20]. At a distance 𝐿 from the neutrino production point, some
muon neutrinos will oscillate to electron neutrinos, resulting in fluxes of approximately

Φ𝐹𝑒 𝐷 (𝐸𝑡𝑟𝑢𝑒 ) ∝ 𝑃𝜈𝜇 →𝜈𝑒 (𝐸𝑡𝑟𝑢𝑒 , 𝐿) Φ𝑁 𝐷

𝜇 (𝐸𝑡𝑟𝑢𝑒 ), (1.17)

Φ𝐹𝜇 𝐷 (𝐸𝑡𝑟𝑢𝑒 ) ∝ 1 − 𝑃𝜈𝜇 →𝜈𝑒 (𝐸𝑡𝑟𝑢𝑒 , 𝐿) Φ𝑁 𝐷

[︀ ]︀
𝜇 (𝐸𝑡𝑟𝑢𝑒 ),

where the proportionality constant depends on the experiment geometry, that can

21
be affected by the different experimental acceptances at the ND and FD locations,
and 𝑃𝜈𝜇 →𝜈𝑒 is the electron neutrino appearance probability. 𝜈𝜇 → 𝜈𝑒 oscillations
are thus observed by measuring the neutrino fluxes Φ𝐹𝑒 𝐷 (𝐸𝑡𝑟𝑢𝑒 ) and Φ𝐹𝜇 𝐷 (𝐸𝑡𝑟𝑢𝑒 ) as
a function of energy and distance. The three-flavor oscillation equations are similar
but include additional terms. Charge-parity (CP) symmetry violation in the leptonic
sector would add a phase 𝛿𝐶𝑃 to the three-flavor oscillation with an opposite sign for
neutrinos and anti-neutrinos [21, 22].
Therefore, the precision to which oscillation parameters can be determined exper-
imentally depends on our ability to extract Φ𝐹𝛼 𝐷 (𝐸) from 𝑁𝛼𝐹 𝐷 (𝐸𝑟𝑒𝑐𝑜 ), as can be seen
in equation 1.16 and is graphically illustrated in figure 1-4.

Figure 1-4: Neutrino energy spectra reconstruction depends on our ability to model
the interaction of neutrinos with atomic nuclei and the propagation of particles
through the atomic nucleus. This flow chart shows the process, starting with an
oscillated far-detector incident-energy spectrum (green), differentiating the physical
neutrino interactions (green arrows) from the experimental analysis (blue arrows),
and ending up with an inferred incident-energy spectrum that hopefully matches the
actual one.

While experimental effects are generally understood and can be minimized us-

22
ing improved detectors, nuclear effects are irreducible and must be accounted for
using theoretical models, typically implemented in neutrino event generators. Thus,
the experimental sensitivity is largely determined by the accuracy of the theoreti-
cal models used to calculate the interaction cross sections and the accuracy of the
energy reconstruction. The available models have many free parameters that are
poorly constrained and are “tuned” by each neutrino experiment. Current oscillation
experiments report significant systematic uncertainties due to these interaction mod-
els [22–25]. Simulations further show that energy reconstruction errors can lead to
significant biases in extracting 𝛿𝐶𝑃 at DUNE [26]. There is a robust theoretical effort
to improve these models [27–29].

Since there are no mono-energetic high-energy neutrino beams, these models can-
not be tested for individual neutrino energies. Instead, experiments tune models to
their near-detector data, where the unoscillated flux Φ𝑁
𝛼 (𝐸𝑡𝑟𝑢𝑒 ) is calculated from
𝐷

hadronic reaction rates [30–32]. While highly informative, such integrated constraints
are insufficient to ensure that the models are correct for each value of 𝐸𝑡𝑟𝑢𝑒 . Thus, for
precision measurements using a broad-energy neutrino beam, the degree to which the
near-detector data alone can constrain models is unclear for a number of reasons. In
some cases, the near and far detectors use different target nuclei, which demands the
cross-section extrapolation between the two targets. At the same time, the fluxes at
the two detectors are not identical due to their distance from the source, and the fact
that neutrinos oscillate. The near detector measures neutrino interactions originat-
ing from pion decays in a long pipe with a length of 𝒪(50 m), while the far detector
measures neutrinos from a much smaller solid angle. Thus, the particle acceptance is
different between the two detectors. Furthermore, the selection efficiencies 𝜖𝛼 (𝐸𝑡𝑟𝑢𝑒 )
in equations 1.15 and 1.16, both for the signal and the background events at the two
detectors, are model dependent and different.

Yet another challenge for neutrino oscillation experiments is the accuracy of the
true neutrino energy reconstruction that enters the oscillation probability formula
shown in equation 1.12. The neutrino energy is reconstructed using the measured en-
ergy deposition of the final state particles. However, energy losses due to the energy

23
of neutral particles, particles below thresholds, and inactive detector regions have to
be taken into account. Inevitably, assumptions about these effects have to be made
based on the underlying modeling choices and on the detector capabilities. This pro-
cedure introduces systematic uncertainties that might limit the precise reconstruction
of the true neutrino energy. Thus, the success of forthcoming neutrino experiments
like DUNE relies on the accurate identification and reconstruction of all particles
produced. Hence, tracking detectors with low detection thresholds are key elements.
Furthermore, neutrinos traveling through matter experience a potential due to
the coherent elastic scattering with electrons and nucleons. Coherent scattering takes
place when a neutrino wavefunction interacts with matter as a whole. The implication
of this behavior is that neutrinos and antineutrinos are affected in different ways, due
to the lack of positrons in regular matter. This effect can mimic a CP violating picture
with 𝑃 (𝜈𝛼 → 𝜈𝛽 ) ̸= 𝑃 (¯
𝜈𝛼 → 𝜈¯𝛽 ), though it contains no fundamental information
related to the matter-antimatter asymmetry. Therefore, accounting for matter effects
in oscillation experiments that aim to extract the oscillation parameters is crucial.
All these open questions need to be addressed in order to ensure the success of
forthcoming high-precision neutrino oscillation experiments. Hence, this thesis pro-
gresses in that direction by improving the understanding of lepton-nucleus interactions
described in sections 1.4 and 1.5.

1.4 Neutrino-Nucleus Interactions

Neutrino-nucleus interactions are extremely complicated processes that become even

more complex on heavier nuclei, such as on argon, the target nucleus used in liquid
argon time projection chambers. These interactions are approximated by deploying
a two-step approach.
The first step involves the primary neutrino interaction with a single nucleon or
a pair of nucleons with the nucleons being treated as quasi-free objects. There are
four main types of neutrino scattering processes that play a major role in the energy
spectrum of neutrino experiments like MicroBooNE and DUNE, namely quasielastic

24
(QE), meson exchange currents (MEC), resonant production (RES), and deep inelas-
tic scattering (DIS). These four interaction types and the relevant outgoing particles
are shown in figure 1-5.

Figure 1-5: The four main interaction processes for neutrino-nucleus scattering events.

Each one of those processes dominates in different energy ranges, as illustrated in

figure 1-6. At neutrino energies below ≈ 1 GeV, QE interactions are the ones that
dominate. With this process, leptons scatter off, and liberate, a single nucleon from
the target nucleus. In the region around 𝒪(1 GeV), a set of nuclear forces include
the exchange of virtual mesons between two (or more) nucleons, which is referred to
as 2 particle-2 hole effects (2p2h). Such an effect has major contributions from MEC
events that lead to the emission of two nucleons from the primary neutrino interaction
point. For interaction energies greater than the ∆ baryon mass of 1232 MeV, RES
interactions become energetically allowed. Such processes are the dominant ones in
the energy range between 1-4 GeV. With such an interaction type, the struck nucleon
is brought into an excited state that is called a baryon resonance. The resonance
deexcites and that leads to the emission of a single nucleon and a single pion in the
final state. At even higher energies starting at ≈ 4 GeV, the incoming neutrino scatters
off a quark in the nucleus. Such an energy transfer results in the production of multiple
hadrons in the final state. Interference terms across the different interaction processes
need to be added coherently at the quantum-mechanical amplitude level. However,
commonly used neutrino event generators, such as GENIE [33], do not include such
interference terms between the different reaction modes. In other words, the total
cross section is obtained by adding the individual cross sections 𝜎𝑖 (𝐸) incoherently.
For neutrino scattering events, interaction processes can take place in the form of
either charged current (CC) or neutral current (NC) interactions. The former includes

25
Figure 1-6: Neutrino cross-section evolution as a function of the neutrino energy
illustrating the energy range where each one of the four main processes dominates.
Figure adapted from [34].

the exchange of charged W bosons, and the latter of neutral Z bosons respectively.
The corresponding Feyman diagrams are shown in figure 1-7.

Figure 1-7: Feyman diagrams illustrating charged current (CC) and neutral current
(NC) processes.

Beyond the possible interactions and processes mentioned above, neutrino-nucleon

interactions become even more sophisticated due to the fact that the interaction takes
place in a dense nuclear medium, such as an argon nucleus. These complex nuclear
effects can be further classified into initial and final state interactions.
Initial state interactions are associated with the nucleon-nucleon correlations in
the target nucleus. The nucleons inside the target nucleus are not at rest. Their

26
Figure 1-8: Nucleon momentum distribution options available in commonly used
neutrino event generators. Figure adapted from [35].

momentum distribution can be approximated in a number of ways, typical examples

of which would be a Bodek-Richie Fermi Gas [36], a Local Fermi Gas [37], or an
Effective Spectral Function [38], as illustrated in figure 1-8. The precise behavior of
the Fermi motion is unknown, and it results in the smearing of the reconstructed
value of the true neutrino energy.

Figure 1-9: Schematic illustration the wealth of possible final state interactions that
the final state hadrons can undergo. Figure adapted from [39].

27
 In the second part of the factorization process, after the primary neutrino-nucleus
interaction, the outgoing nucleons can undergo a wealth of final state interactions
while transversing the nuclear medium, before they exit the nucleus. Such re-interactions
might result in processes such as the emission of further hadrons, the absorption of the
initially emitted hadrons, charge exchange processes, and/or acceleration/deceleration
of the primary hadrons, as illustrated in figure 1-9.
All the effects described above have a direct impact on the ability to accurately
reconstruct the energy of the incoming neutrino using the properties of the final state
particles.

1.5 Connections To Electron Scattering

Neutrinos and electrons interact with atomic nuclei by exchanging intermediate vector
bosons, a massive 𝑊 ± or 𝑍 for the neutrino and a massless photon for the electron.
Electrons interact via a vector current 𝑗𝐸𝑀
𝜇
= 𝑢¯𝛾 𝜇 𝑢 and neutrinos interact via vector
and axial-vector 𝑗𝐶𝐶
𝜇
√𝑊 currents.
= 𝑢¯𝛾 𝜇 (1 − 𝛾 5 )𝑢 −𝑖𝑔
2 2

Fundamental considerations give an inclusive (𝑒, 𝑒′ ) electron-nucleon elastic scat-

tering cross section that depends on only two structure functions [40],

𝑑2 𝜎 𝑒 4𝜋𝛼2 1 − 𝑦 𝑒
[︂ ]︂
= 2 2 𝑒 2
𝐹2 (𝑥, 𝑄 ) + 𝑦 𝐹1 (𝑥, 𝑄 ) . (1.18)
𝑑𝑥𝑑𝑄2 𝑄4 𝑥

Here 𝐹1𝑒 and 𝐹2𝑒 are the standard electromagnetic vector structure functions, 𝑄2 =
q2 − 𝜈 2 is the squared momentum transfer and q and 𝜈 are the three-momentum
and energy transfers, 𝑥 = 𝑄2 /(2𝑚𝜈) is the Bjorken scaling variable, 𝑚 is the nucleon
mass, 𝑦 = 𝜈/𝐸𝑒 is the electron fractional energy loss, and 𝛼 is the fine structure
constant. This formula shows the simplest case where 𝑄2 ≫ 𝑚2 .
The corresponding inclusive charged-current (CC) (𝜈, 𝑙± ) neutrino-nucleon elastic
cross section has a similar form, where 𝑙± is the outgoing charged lepton [41]. The
vector part of the current is subject to the same fundamental considerations as above,

28
but the axial-vector part of the current does not conserve parity. This leads to a third,
axial, structure function,

𝑑2 𝜎 𝜈 𝐺2𝐹
[︂
1−𝑦 𝜈
= 𝐹2 (𝑥, 𝑄2 ) + 𝑦 2 𝐹1𝜈 (𝑥, 𝑄2 )
𝑑𝑥𝑑𝑄2 2𝜋 𝑥 (1.19)
−𝑦(1 − 𝑦/2)𝐹3𝜈 (𝑥, 𝑄2 ) .
]︀

Here 𝐹1𝜈 and 𝐹2𝜈 are the parity-conserving neutrino-nucleus vector structure functions,
𝐹3𝜈 is the axial structure function, and 𝐺𝐹 is the Fermi constant. The vector form
factors, 𝐹1𝜈 and 𝐹2𝜈 , have both vector-vector and axial-axial contributions.

These simple equations are very similar for electron-nucleus scattering. In the limit
of electron-nucleon elastic scattering (𝑥 = 1), the two structure functions reduce to the
Dirac and Pauli form factors, which are linear combinations of the electric 𝐺𝐸 (𝑄2 ) and
magnetic 𝐺𝑀 (𝑄2 ) form factors. Neutrino-nucleon elastic scattering has an additional
axial form factor. In the simplest case where a lepton scatters quasielastically from
a nucleon in the nucleus and the nucleon does not reinteract as it leaves the nucleus
shown in figure 1-10, the lepton-nucleus cross section is the integral over all initial
state nucleons,

𝑑𝜎 𝑓 𝑟𝑒𝑒
∫︁ ∫︁
𝑑𝜎
= 𝑑3 pi 𝑑𝐸𝑏 𝐾𝑆(pi , 𝐸𝑏 )
𝑑𝐸𝑑Ω pi 𝐸𝑏 𝑑Ω (1.20)
𝛿 3 (q − pf − pr )𝛿(𝜔 − 𝐸𝑏 − 𝑇𝑓 − 𝑇𝑟 ),

where pi and pf = q + pi are the initial and final momenta of the struck nucleon
in the absense of any reinteraction, pr = −pi is the momentum of the recoil 𝐴 − 1
nucleus, 𝐸𝑏 is the nucleon binding energy, 𝑆(pi , 𝐸𝑏 ) is the probability of finding a
nucleon in the nucleus with momentum pi and binding energy 𝐸𝑏 , 𝑇𝑓 and 𝑇𝑟 are
the kinetic energies of the final state nucleon and 𝐴 − 1 system, 𝑑𝜎 𝑓 𝑟𝑒𝑒 /𝑑Ω is the
lepton-bound nucleon elastic cross section, and 𝐾 is a known kinematic factor.

This simple picture is complicated by nucleon reinteractions which change the

29
Figure 1-10: Quasielastic lepton-nucleus scattering where the outgoing nucleon does
not reinteract as it leaves the nucleus.

overlap integral between the initial and final states, and thus the cross section. The
latter further changes the momentum and angle of the outgoing nucleon. Thus,
to calculate even the simplest type of lepton-nucleus interaction, the momentum and
binding energy distribution of all nucleons in the nucleus need to be known, as well as
how the outgoing nucleon wave function is distorted by the nucleon-nucleus potential,
and how the outgoing nucleon kinematics is changed by final state interactions.

Electron-nucleus scattering is much easier to understand than neutrino-nucleus

scattering for three reasons: (a) electron beams have a single, well-known, energy; (b)
electron experiments typically have low statistical uncertainty because electron beams
have high flux, and electron-nucleus cross sections are far higher than their neutrino
counterparts; and (c) electron cross sections are purely vector. The strength of the
interaction is very different: 4𝜋𝛼2 /𝑄4 for electrons versus 𝐺2𝐹 /(2𝜋) for neutrinos,
where the factor of 1/𝑄4 in the electron cross section is due to the exchanged boson
mass (i.e. massless photon) in the propagator. When compensating for the factor
of 1/𝑄4 , the shapes of the electron- and neutrino-nucleus cross sections are very
similar [42]. Nuclear medium effects such as nucleon motion, binding energy, two-
body currents, and final state interactions are similar or identical. Therefore electron-
nucleus scattering can be used to constrain models of neutrino-nucleus scattering. Any
model which fails to accurately describe electron-nucleus (vector-vector) scattering
data cannot be used with confidence to simulate neutrino-nucleus (vector-vector +
axial-axial + vector-axial) interactions. Thus, models of the neutrino-nucleus cross

30
section must be able to describe the more limited electron-nucleus cross section.

1.6 Thesis Scope

As detailed in section 1.3, the success of future precision neutrino oscillation experi-
ments depends on an unprecedented understanding of neutrino-nucleus interactions.
Insufficient knowledge of either the energy reconstruction or the cross section will limit
the experimental precision. In this thesis, the expertise in both neutrino and electron
cross-section modeling and analysis is leveraged in order to alleviate this insufficient
knowledge using three complementary approaches, illustrated in figure 1-11.

Figure 1-11: Graphic illustration of the three complementary approaches used in this
thesis to improve our understanding of lepton-nucleus interactions.

Namely, neutrino scattering data sets from the MicroBooNE detector at Fermi
National Laboratory were analysed (chapter 2). The first measurements of exclusive
cross sections with a single proton and no pions detected in the final state were
reported. These results were used to identify regions where modeling improvements
are required and specific nuclear effects can be studied in detail (chapter 3). The
exact same event topology was investigated using electron scattering data sets. For
the connection across neutrinos and electrons to be established, significant modeling
improvements took place to ensure a consistent modeling across the two particle
species in the commonly used GENIE event generator (chapter 4). Building on those
improvements, the “Electrons-For-Neutrinos” analysis reported on the first use of wide

31
phase-space electron scattering data sets from the CLAS detector at Thomas Jefferson
National Laboratory (chapter 5). The analysis identified significant shortcomings in
our lepton-nucleus interaction understanding by reporting cross sections as a function
of energy reconstruction methods and testing the validity of models commonly used
in neutrino oscillation analyses (chapter 6).

32
Chapter 2

The MicroBooNE Experiment At

Fermi National Laboratory

2.1 Fermilab Neutrino Beamlines

Fermi National Laboratory (Fermilab) takes advantage of artificially produced neu-
trino beams in order to study neutrino oscillations. The advantage of that choice is
the better control over the energy spectrum range and the neutrino flavor content.
Thus, the spectrum can be tuned so that, given the distance of the detector from the
source, the experiment is located at an oscillation maximum or minimum, depending
on the objective of the intended measurement [43]. On top of that, a narrower energy
spectrum can be obtained via the off-axis technique [44].
The neutrino beam production includes a well-defined procedure that is outlined
below for the Fermilab beamlines. The Fermilab linac is used to accelerate a proton
beam up to a kinetic energy of 400 MeV. Then the Booster synchrotron is used to
further accelerate the protons up to a kinetic energy of 8 GeV.
For the Booster Neutrino Beam (BNB), a kicker extracts the protons in a single
turn and those protons are redirected to the relevant target hall. For all the other
experiments, the protons are redirected to the Main Injector. The latter is the high-
est energy US-based accelerator facility and the source of neutrinos from the Main
Injector (NuMI), the source of muons for the muon campus experiments, and the

33
forthcoming high intensity DUNE neutrino beam. Figure 2-1 illustrates the series of
the relevant accelerator components and the different beams.

Figure 2-1: The Fermilab accelerator complex showing the accelerator components
and the different beams. Figure adapted from [45].

The BNB takes advantage of the 8 GeV protons from the Booster synchrotron.
These protons collide with a beryllium target which is located within a pulsed elec-
tromagnet called horn. Such collisions lead to the production of mesons, mainly 𝜋 ± ,
𝐾 ± , and 𝐾 0 . The channel that dominates is p + Be → 𝜋 + + X, with X corresponding
to all the other hadrons produced out of the interaction. When operating in neutrino
mode, the horn focuses the positively charged particles. The focused particle beam
then enters the decay pipe via a concrete-made collimator with a length of ≈ 2 m. In
the decay pipe, some of the particles decay, a process that results in the production
of neutrinos primarily via the channel 𝜋 + → 𝜇+ + 𝜈𝜇 . The massive particles are
stopped using a beam absorber made of steel and concrete. On the other hand, neu-
trinos transverse the absorber and, at the end of this process, a neutrino-dominated
beam is obtained. A graphic illustration of the process is shown in figure 2-2.
Apart from the desired 𝜈𝜇 beam, neutrinos can also be produced via the decay of
anti-muons coming out of the proton-target collision and the channel 𝜇+ → 𝑒+ + 𝜈𝑒 +
𝜈¯𝜇 [46]. Such interactions are the main sources of intrinsic 𝜈𝑒 contamination to the
main 𝜈𝜇 beam. Similar hadron decays result in further contamination of the 𝜈𝜇 beam.
The main source of the 𝜈¯𝜇 contamination comes from 𝜋 − ’s which are not separated

34
Figure 2-2: The ingredients for a neutrino beam include the accelerated protons,
the target, the magnetic horn, the decay pipe, and the absorbers. Figure adapted
from [45].

from the main beam by the horn. Figure 2-3 illustrates the BNB energy spectrum
while operated in neutrino mode.

Figure 2-3: The BNB neutrino flux prediction through the MicroBooNE detector for
𝜈𝜇 , 𝜈¯𝜇 , 𝜈𝑒 , and 𝜈¯𝑒 . A TPC volume with dimensions 2.56 m × 2.33 m × 10.37 m is
assumed. Figure adapted from [46].

The Fermilab-based NuMI neutrino beamline is used by the MINOS, Miner𝜈a,

ArgoNeuT, NO𝜈A, and MINOS+ experiments [47–53]. On top of that, the Mini-

35
BooNE and MicroBooNE experiments use the beamline for off-axis studies. In this
case, neutrinos are produced via the collision of a 120 GeV beam on a graphite tar-
get. Due to the large angle with respect to the NuMI beam dump, MicroBooNE
collects a significant number of low-energy neutrinos originating from kaons decay-
ing at rest. Using the off-axis NuMI beamline, MicroBooNE records events with a
narrower energy spectrum and with a higher electron-neutrino contribution.

2.2 The MicroBooNE Detector

The Micro Booster Neutrino Experiment (MicroBooNE) at Fermilab was proposed

to succeed MiniBooNE in order to resolve the Low Energy Excess (LEE) [54]. It is
located ≈ 20 m away from the MiniBooNE detector, thus establishing almost the same
𝐿/𝐸 ratio. MicroBooNE uses the same beamline as MiniBooNE, therefore the two
experiments share similar flux uncertainties. At the core of the MicroBooNE detector,
there exists a Liquid Argon Time Projection Chamber (LArTPC). This detector
technology offers high spacial and momentum resolution, allowing an unprecedented
energy reconstruction accuracy and a precise particle identification.
Some neutrinos from the two beamlines enter the MicroBooNE LArTPC and
interact with the argon nuclei. Out of these interactions, charged and neutral particles
are produced. When the charged particles transverse the liquid argon, the argon
atoms get excited and they further ionize the medium, a process that results in
the emission of ionization electrons. In the presence of a strong electric field of
273 V/cm, the ionization electrons drift towards the anode plane, where three wire
planes are located. Apart from the ionization electrons, scintillation light is also
produced and is collected by 32 photomultiplier tubes (PMTs), which are located
behind the anode plane. Both the TPC and the light collection system are embedded
withing a cylindrical 170 ton liquid argon cryostat. Figure 2-4 shows the size of the
detector.
The readout electronics are embedded withing the liquid argon in order to sig-
nificantly reduce the electronic noise. The analogue-to-digital conversion (ADC) and

36
Figure 2-4: Schematic illustration of the MicroBooNE detector and its dimensions.
Figure adapted from [55].

PMT electronics are located outside the cryostat.

Given the baseline of ≈ 470 m, neutrinos in the two beamlines need to be charac-
terized by two angles in the MicroBooNE coordinate system. As shown in figure 2-5,
𝜃 is the angle with respect to the z-direction along the beam direction. Further-
more, 𝜑 defines the orientation with respect to the XY-plane, orthogonal to the beam
direction.

Figure 2-5: The MicroBooNE coordinate system.

A Cosmic Ray Tagger (CRT) system shown in figure 2-6 was installed in 2017 [56]
to improve on the rejection of cosmics, which constitute the greatest source of back-
grounds on MicroBooNE. This detector sub-system consists of 73 scintillating modules
made of interleaved layers of scintillating plastic strips situated on the top, bottom,

37
and two sides parallel to the neutrino beam [56]. Based on simulation predictions
from CORSIKA [57] and GEANT [58], an estimated coverage of 85% is obtained.
The CRT installation aimed to improve on the identification and rejection of the
dominant cosmic background. Cosmic muons transversing the CRT result in the pro-
duction of scintillation light that can be reconstructed as hits on the CRT channels.
Such hits allow the identification of the cosmic-induced muon tracks with a time pre-
cision of ≈ 100 ns. The latter offers a complementary way to resolve the x-direction
ambiguity in the TPC reconstruction.

Figure 2-6: The design of CRT planes as part of the MicroBooNE detector. Simula-
tion of cosmic rays crossing the CRT, the brown lines represent possible cosmic ray
trajectories. There are four CRT planes: top plane, bottom plane, pipe side plane
and feedthrough side plane. The beam direction is along the z axis. Figure adapted
from [56].

2.3 Liquid Argon Time Projection Chambers

The TPC technology was introduced in the 1970s by David Nygren [59]. Carlo Rubbia
designed a LArTPC in 1977 using the same TPC principles as Nygren, but with liquid
argon instead of gas [60]. Figure 2-7 shows the working principle of a LArTPC. The
cuboid volume of the LArTPC is filled with ultra pure liquid argon. The presence

38
of a high-voltage cathode on one side of the detector and a grounded anode on the
other side create a homogeneous electric field.

Figure 2-7: Working principle of a LArTPC detector. Figure adapted from [61].

When a neutrino interaction takes place in the TPC, charged particles are pro-
duced in the final state. Along their propagation path in LAr, such particles excite
and ionize the argon nuclei, a process that results in the emission of ionization elec-
trons. In the presence of a strong electric field, these ionization electrons drift towards
the anode plane. To ensure an electron drift time of 𝒪(ms) before recombination ef-
fects take place, a very small 𝑂2 contamination of 10 parts per trillion has to be
guaranteed.
On the anode plane, there exist three wire planes with a 3 mm spacing where the
clouds of the arriving ionization electrons create signals. The first two planes corre-
spond to induction planes and are oriented ±60𝑜 with respect to the vertical axis. In
order to obtain 3D views of the particle trajectories, at least two planes with different
orientations are required. The third collection plane removes the ambiguities due to
dead wires. On top of that, the calorimetric and tracking abilities are improved. Bias
voltages are applied on the wire planes so that the two induction planes satisfy the
transparency condition outlined in [62, 63]. The condition requires that the drifting

39
electrons pass the two induction planes and are fully captured on the third collection
plane. The drifting charge induces a bipolar signal on the two induction planes and is
collected on the third one, where a unipolar signal is produced, as shown in figure 2-8
(left). The signal area is proportional to the ionization.

Figure 2-8: (Left) bipolar (U and V induction planes) and unipolar (Y collection
plane) signal induction on the three MicroBooNE planes. Figure adapted from [63].
(Right) schematic view of the wire planes. The vertical collection Y wires are shown
in pink, the induction U wires, angled at +60𝑜 are shown in blue and the induction
V wires, angled at -60𝑜 , are shown in green. Figure adapted from [64].

The MicroBooNE field cage has a height of 2.3 m, a width of 2.6 m and a length
along the beam direction of 10.4 m. The liquid argon is kept at a pressure of 1.2 atm,
a boiling temperature of 89 K, and a resulting density of 1.38 g/cm3 . The TPC is
merged into the liquid argon. The active TPC volume is 86 tons. The cathode is kept
at -70 kV and the anode is grounded, which results into a homogeneous electric field of
273 V/cm. That translates into a drift velocity of 1.14 m/ms. Thus, the MicroBooNE
readout window is 2.3 ms. The induction planes U and V are biased at -110 V and
the collection V plane is biased at 230 V. The induction planes include 2400 wires and
the collection plane consists of 3456 wires. The distance between both the different
planes and the plane wires is 3 mm. Figure 2-8 (right) shows the schematic view of
the wire planes.
Within the liquid argon, the wire signals are fed into a front end ASIC. Inter-
mediate amplifiers further amplify the signal and pass it to the feed-through. Then,
outside the cryogenic environment, the signal is digitized by readout modules with a
frequency of 16 Hz. The next step is to downsample the signal to lower frequencies

40
of 2 Hz [65].
There exist three 1.6 ms signal readout windows for each event. These frames
are further truncated to the range between -0.4 ms to 2.7 ms. The hardware-defined
trigger 0 time is obtained from the accelerator division. Given the 2.3 ms MicroBooNE
drift time, the extra buffer of 0.4 ms makes sure that there is enough time to isolate
the neutrino interaction from the cosmic rays that arrive close to the neutrino beam
trigger time.

Figure 2-9: Entry/exit points of cosmic muon tracks with a signal from a muon
counter located outside of the cryostat. In the absence of space charge effects, the
points should be located along the TPC boundaries indicated by the dashed lines.
Figure adapted from [66].

The ion drift velocity in liquid argon is ≈ 5 mm/s, orders of magnitude smaller
than the electron one. Thus, the argon ions result in the build-up of positive charge in
the LArTPC for minutes. On top of that, the continuous interaction of cosmic rays in
the TPC at ≈ 5 kHz also results in a continuous build-up of positively charged argon
ions. The existence of this positive charge leads to a distortion of the homogeneous
electric field within the TPC, as shown in figure 2-9. This distortion is referred to as

41
“space charge effect” (SCE) [66], which leads to a displacement of the reconstructed
signal ionization source by up to 𝒪(10 cm).

2.4 Scintillation Light

The scintillation light production and propagation is almost instantaneous and of the
order of 𝒪(ns). Thus, collecting the emitted scintillation light is crucial to identify
the time that the event took place and to determine the x-position along the drift
direction in the TPC. When charged particles transverse the liquid argon, scintillation
light is produced that results in the deposition of ≈ 104 𝛾 / MeV. This light can be
produced via two mechanisms, shown in figure 2-10.

Figure 2-10: The production of scintillation light in liquid argon. A charged particle
can either excite or ionise the argon. Figure adapted from [6].

The first mechanism is the self-trapped exciton luminescence. In that case, charged
particles transverse the liquid argon and they leave some of the argon atoms in an
excited state called excitons. These states are molecules with another argon atom
with a short lifetime and are called dimers or excimers. Roughly 65% of them are in
a singlet state 1 Σ𝑢 and the remaining ones are in a triplet state 3 Σ𝑢 .
With the second mechanism, the charged particles ionize the argon atoms and

42
that results in the creation of free electrons. These electrons recombine with the
positive argon ions, a process that also creates excited dimers. With this mechanism,
the probability of creating either singlets or triplets is equal.

The singlet states result in the emission of fast scintillation light with a decay
time of ≈ 6 ns. The triplet states result in a slow component and in a decay time of
≈ 1.5 𝜇s. Both the fast and the slow components have a peak wavelength at 128 nm
in the Vacuum Ultra-Violet (VUV) region. Both states have an energy minimum
which is equivalent to a distance between the atoms of ≈ 2.8 Å [67]. The liquid argon
inter-atom separation is ≈ 4 Å, which is greater than the one that corresponds to
the excimer energy minimum. Thus, liquid argon is transparent to its own light and,
therefore, the light can be detected over long distances.

The second mechanism is related to the recombination luminescence and relies on

the free electron and argon ion local density. Thus, recombination effects are stronger
for particles with higher energy deposition per unit length. This implies a particle-
type dependence of the energy deposition in the detector to the light-equivalent pro-
duction. That also means that the scintillation light yield depends on the choice for
the electric field. A strong electric field will reduce the recombination effects and,
therefore, the recombination-induced luminescence. The amount of scintillation light
can be further reduced due to impurities that result into quenching and absorption
effects. With that in mind, the liquid argon purity in MicroBooNE is monitored to
ensure an absorption length greater than the 2.5 m-long TPC width.

The MicroBooNE light detection system consists of 32 8-inch Hamamatsu cryo-

genic PMTs located behind the anode plane, shown in figure 2-11. A Tetra-Phenyl
Butadiene (TPB) coated acrylic plate is located in front of the PMT. The TPB con-
verts the 128 nm light argon scintillation light to 425 nm visible light. The tube is
surrounded by mu-metal to shield the electronics against any magnetic fields. The
PMTs operate at a voltage of 1300 V and have a gain of 107 [68]. The PMT signal is
sent to a splitter board, where it gets pre-amplified and digitized. The signals have a
60 ns rise time, much smaller than the spill windows for the BNB (1.6 𝜇s) and NuMI
(10 𝜇s) beams.

43
Figure 2-11: The MicroBooNE light collection system with the 32 PMTs. Figure
adapted from [69].

2.5 Hardware And Software Triggers

Even when accelerator-induced neutrino beams transverse the MicroBooNE detec-

tor, it is highly unlikely that any neutrino will interact due to the extremely small
neutrino-nucleus cross sections. Each BNB bunch includes ≈ 1.2 × 1012 protons and
the TPC surface area is ≈ 60 × 103 cm2 . That yields a neutrino interaction every ≈
500 bunches [19]. In order to significantly reduce the amount of data that is stored on
tape, a minimal amount of optical activity is required. Regardless, background events
with light activity within the beam spill arrival time are still the ones that dominate.
Figure 2-12 shows the differences between two such events [70]. The first one (left)
corresponds to pure cosmic activity that arrives in coincidence with the beam spill.
The second one (right) includes a neutrino interaction of interest, although there is
still a significant cosmic contamination.
Cosmic-induced background events are accounted for by recording events when
the neutrino beam is off. That leads to the existence of two trigger types. The first
one, referred to as “Beam On”, corresponds to data samples recorded when the beam
is on and the light system yielded a signal. The second one, referred to as “Beam Off”
or “External BNB” (ExtBNB), uses the same configuration as in the Beam On case,
but those data sets are recorded when the beam is off.
The MicroBooNE readout system is triggered with the arrival of a signal [64]. The
signal might originate from the BNB/NuMI accelerator clocks. Alternatively, there

44
Figure 2-12: (Left) cosmic-induced event that was stored because of the coincidence
of a 1.6 𝜇s accelerator BNB signal and light detected by the PMTs. (Right) neutrino-
induced event where the light was coming from a neutrino interaction. Figure adapted
from [70].

is a function generator in the trigger rack producing pulses at a fixed frequency. The
latter is used to record Beam Off events. The trigger board sends a signal to all the
readout crates to start recording data. On top of that, the trigger keeps track of the
trigger type and the time that the signal was received. The accelerator-based signals
are produced in couples. The former (early signal) vetos Beam Off triggers just before
the beam triggers. That process aims to avoid any trigger overlap that might result
in the reduction of the exposure to the beam. The latter signal is the one used to
trigger the readout. The MicroBooNE TPC readout is completely unbiased, thus all
the time ticks are stored and the readout is not zero-suppressed. The PMT readout
is biased though, with data being stored during specific time intervals determined by
the “discriminators”.

There exist two PMT data-taking configurations that differ by their duration and
by the suppression level. The duration is frequently expressed in tick units, with each
tick corresponding to 15.625 ns. For the BNB triggering, the beam discriminator
starts the data recording simultaneously for all PMTs. That happens by replicating
the trigger signal and by redirecting it to all the PMT boards. The duration of this
beam window is 23.4 𝜇s or 1500 optical ticks. On average, the neutrino arrival time
is ≈ 4 𝜇s after the opening of the beam window. The cosmic discriminators span a
range outside the 23.4 𝜇s window. The ultimate goal of the cosmic discriminators is to

45
suppress the amount of data that is recorded over a long time interval, once the signal
for a triggered event has arrived. The duration of the cosmic window is 4.8 ms and
spans the range of [-1.6,+3.2] ms. Only waveforms with more than 130 ADC counts
are stored. That number corresponds to ≈ 7 photo-electrons (PE). The accelerator
signals are meticulously timed so that neutrinos originating from the 1.6 𝜇s BNB spill
arrive in MicroBooNE during this time window of 23.4 𝜇s. When the beamline is fully
operational, that takes place ≈ 5 times/s. The majority of the triggered events based
on the hardware-driven signal do not include neutrino interactions. Thus, a software
trigger is further applied to determine whether an event is a neutrino candidate or
not. This trigger uses PMT optical waveforms and searches for light activity within
the 1.6 𝜇s BNB beam-spill. That action takes place after the TPC data has been sent
to the DAQ crates.

2.6 Optical Event Reconstruction

In order to perform a MicroBooNE data analysis, the low-level PMT signals have
to be converted to high-level data products. That is achieved by using the raw
PMT waveforms as an input quantity and, based on these, flashes are reconstructed.
Such flashes are indicative of optical activity in the TPC recorded by multiple PMTs
simultaneously. This light activity is collected by the 32 8-inch PMTs oriented as
shown in figure 2-13. The neutrino-induced light activity results in a higher PE yield
compared to the cosmic-induced one, as shown in figure 2-14.

Figure 2-13: Optically reconstructed flash object recorded by the MicroBooNE PMT
light collection system. The dark orange regions represent a higher PE yield. Figure
adapted from [64].

The PMTs have a low (≈ 20 ADC/PE) and a high (≈ 2 ADC/PE) gain read-

46
out [64]. The optical reconstruction merges the two streams into a corrected waveform
which corrects saturated high-gain pulses based on the information obtained from the
low-gain pulses. Depending on the discriminator type, a different baseline estimation
is used. The cosmic discriminator uses a constant baseline, called pedestal. The
beam discriminator uses a time dependent baseline estimate. This is a more accurate
estimation and addresses potential overestimations or underestimations in the signal
baseline. Once the pedestal has been identified, the waveform ADC counts are used.
Those pulses above threshold are identified and propagated to the next stage for the
“hit” reconstruction. These pulses result in the creation of data products based on this
optical hit reconstruction. Each optical hit refers to optical PMT activity, namely
the number of produced PEs and the event time. The optical hits are clustered and
the PMT PE production is summed in order to reconstruct flashes [71].

Figure 2-14: Neutrino-induced tracks (black) are matched to the corresponding light
signals collected by PMTs (red circles) and are clearly separated from the cosmic-
induced ones (dimmed color tracks). Figure adapted from [72].

2.7 TPC Signal Processing

While drifting towards the anode plane, the ionization electrons repel each other.
That results in a diffused signal arriving on the anode plane and the level of the
diffusion depends on the position that the interaction took place. Those drifting
electrons result in the induction of current on the neighboring wires. Therefore,
waveforms produced on a specific wire might have an effect on those produced on a

47
neighboring wire. The current induced on the wires is amplified and shaped by the
ASICs located within the liquid argon.
The objective of the noise filtering and signal processing on MicroBooNE is to
convert those raw digitized waveforms into the number of ionization electrons passing
through a specific wire plane at a given time [73,74]. For that to be achieved, the first
step is the application of noise filters to remove the external noise and the electronics-
induced one. The major sources of external noise originate from the TPC drift high-
voltage power supply and the low-voltage regulators for the front-end ASICs [63].
Then, the application of a deconvolution of the digitized TPC wire signals follows.
That takes place in two dimensions, with the first one being over time and the second
one being the effect across multiple wires. A region of interest (ROI) is identified
based on the deconvoluted charge distribution. The ionization charge is obtained
with a linear baseline subtraction within the start/end bins of the ROI window. At
the final step, the processed signals accounting for the number of electrons on a given
wire at a certain time are used as input to the MicroBooNE event reconstruction.
The signals are calibrated before the conversion to deposited energy is performed.
Figure 2-15 illustrates the necessity of signal processing before any reconstruction is
enabled.

2.8 Pandora Reconstruction Framework

The processed signals from the previous stage are used as the input for higher level re-
construction of objects such as vertices, tracks, and showers. In order to reconstruct
objects in the TPC, the Pandora reconstruction framework is deployed [75]. Pan-
dora uses pattern-recognition algorithms, along with the use of multiple algorithms
completing specific tasks for a given topology. The waveforms obtained with 4.8 ms
window are used as an input to the reconstruction framework.
The first step includes fitting the processed waveforms with Gaussian distributions
to each peak. This fitting process results into the creation of a 2D hit. Then, Pandora-
Cosmic is a track-focused selection that aims to tag the cosmic muons. The selection

48
Figure 2-15: Candidate neutrino event display from MicroBooNE data on one of
the induction planes. (a) The raw waveform image. (b) The image after noise-
filtering. (c) The image after 2D deconvolution. The image quality near the neutrino
interaction vertex significantly improves after the 2D deconvolution and the latter
leads to improvements in the pattern recognition. Figure adapted from [74].

results in the creation of a cosmic-free hit collection. PandoraNu identifies neutrino

interaction vertices and uses them to reconstruct tracks and showers originating from
the vertex. A parent neutrino particle is defined and the reconstructed objects are
added as daughter particles. Creating a “slice” is the first step that Pandora performs.
A slice is defined as a collection of reconstructed particles originating from the same
interaction. For the creation of the slices, the PandoraCosmic algorithm is first ran
over all the hits in order to identify the cosmic-induced muon tracks and the associated
𝛿-rays and Michel electrons under a cosmic hypothesis. The obvious cosmic activity
of through-going muon tracks is identified based on the geometric information. The
remaining hits are used by the PandoraNu algorithm and objects are reconstructed
under the neutrino hypothesis, creating a slice. Each slice is reconstructed using both

49
hypotheses. For the interactions to be reconstructed in a three-dimensional space,
Pandora requires information from at least two wire planes [75]. The 2D hits are
clustered on each wire plane and for each slice. A collection of 3D candidate vertices
is produced by identifying locations that project to the same points of the 2D clus-
ters. All the candidate vertices are propagated into a Support Vector Machine (SVM)
selection and the candidate with the highest score is isolated. The cluster matching
algorithms are ran around this candidate vertex on each plane and are compared to
improve the matching of the reconstructed objects [75]. With this process, a collec-
tion of reconstructed Particle Flow Particles (PFParticles) is constructed. Such a
PFParticle is created by combining 2D cluster objects on the three wire planes. Each
one of those PFParticles is associated with a vertex location and has a collection of 3D
points, which contain the charge details from the relevant 2D hits. These 3D points
are referred to as SpacePoints. The reconstructed PFParticles in the neutrino slice
are organized with a hierarchical structure based on parent-daughter assocciations,
as shown in figure 2-16.

Figure 2-16: Illustration of the hierarchical structure of particles reconstructed for a

simulated charged current 𝜈𝜇 event in MicroBooNE. The interaction includes a muon,
proton and charged pion in the visible final state. Figure adapted from [75].

A candidate neutrino PFParticle is assigned at the very top of the hierarchy

by the PandoraNu algorithm. That candidate neutrino will be having at least one

50
daughter PFParticle. These daughter particles are assigned a score that classifies
them as either-track-like or shower-like objects. A Support Vector Machine uses the
collection of hits to determine the nature of reconstructed object. Track-like objects
have a score close to 1 and shower-like objects score closer to 0. Based on that score,
a shower- or a track-like data product is constructed for each particle.
In the case of a track-like classification, Pandora uses a linear fit, described in detail
in [75], and returns the direction and the position of each point across the particle
trajectory in 3D. For each one of the points, the charge deposition dQ/dx and the
residual range - the distance from the end of the track - are stored. This approach
allows accurate measurements of dx that might include deflections and displacements
due to space charge effects. For track-like objects, dQ/dx is converted to dE/dx using
the inverse Modified Box model [76], as shown in figure 2-17. The advantage of this
model is that the non-linear dependence of the local density of ions is taken into
consideration.

Figure 2-17: Illustration of the measured dQ/dx vs dE/dx distribution with the mod-
ified recombination model in the MC simulation with the ArgoNeuT parametrization.
Figure adapted from [76].

51
 For shower-like objects, Pandora creates a 3D cone along the hit collection with
a fixed 3D orientation, solid angle and length. The shower energy is obtained and
calibrated using the same techniques as the ones used on the 𝜋 0 reconstruction [77].
Furthermore, showers are fitted with a Kalman filter [78]. With this fit, hits that are
longitudinally or transversely displaced from the main shower cone are removed. This
fitting process returns a track-like object. Therefore, the calorimetric tools mentioned
in the previous paragraph become available for shower-like objects too.

2.9 Cosmic Overlay Simulation

For the purposes of the oscillation and cross section analyses on MicroBooNE, neutrino-
induced interactions in our detector need to be induced. However, MicroBooNE is
a surface detector dominated by the cosmic activity. In order to simulate this cos-
mic contamination as accurately as possible, real cosmic events collected with the
unbiased trigger are used. This trigger stores events outside the beam-related trig-
ger windows and does not demand the existence of any optical activity in a specific
part of the detector. Such events are overlaid on top of GENIE simulated neutrino
interactions [79]. These resulting samples are referred to as “cosmic overlays” and an
example event display is shown in figure 2-18.

Figure 2-18: MicroBooNE event display of an event in a cosmic overlay sample. A

simulated neutrino event (orange box) is overlaid on top of cosmic events (blue box).

52
 For the simulated part of the cosmic overlays, the reconstructed information is
“backtracked” to the underlying truth-level information. That is achieved by associ-
ating the hits on each plane to the GEANT particles that resulted in the production
of these hits. Thus, it is feasible to relate reconstructed PFParrticles to the under-
lying simulated interaction products. It is further possible to identify the amount of
charge originating from the cosmic part of the overlay samples. Reconstructed track-
and shower-like objects are matched to a simulated object when they have more hits
in common than any of the other simulated particles or cosmic tracks in a given event.
Motivated by the analyzer’s involvement in the development and validation of
these samples, the analyses presented in sections 3.1 and 3.2 were the first ones
to adopt the aforementioned overlay technique. Due to the success in accurately
describing the cosmic background, this technique is currently used as the default
simulation option across the MicroBooNE collaboration.

53
54
Chapter 3
MicroBooNE Quasielastic-like
Cross-Section Results
[Phys. Rev. Lett. 125, 201803 (2020)]

3.1 First Measurement of Differential Charged Cur-

rent Quasielastic-like Scattering Cross Sections

3.1.1 Quasielastic-like Neutrino Data Analysis

As outlined in section 1.4, understanding the interaction of neutrinos with argon nuclei
is of particular importance as a growing number of neutrino oscillation experiments
employ liquid argon time projector chamber (LArTPC) neutrino detectors. Exper-
imentally, the energy of interacting neutrinos is determined from the measured mo-
menta of particles that are emitted following the neutrino interaction in the detector.
Many accelerator-based oscillation studies focus on measurements of charged-current
(CC) neutrino-nucleon quasielastic (QE) scattering interactions [80–89], where the
neutrino removes a single intact nucleon from the nucleus without producing any
additional particles. This choice is guided by the fact that CCQE reactions can
be reasonably well approximated as two-body interactions, and their experimental
signature of a correlated muon-proton pair is relatively straightforward to measure.
Therefore, precise measurements of CCQE processes are expected to allow precise
reconstruction of neutrino energies with discovery-level accuracy [90].

55
 A working definition for identifying CCQE interactions in experimental measure-
ments requires the identification of a neutrino interaction vertex with an outgoing
lepton, exactly one outgoing proton, and no additional particles. These events are
referred to herein as “CCQE-like”. This definition can include contributions from
non-CCQE interactions that lead to the production of additional particles that are
absent from the final state due to nuclear effects, such as pion absorption, or have
momenta that are below the experimental detection threshold. Pre-existing data on
neutrino CCQE-like interactions came from experiments using various energies and
target nuclei [91]. These primarily included measurements of CCQE-like muon neu-
trino (𝜈𝜇 ) cross sections for interactions where a muon and no pions were detected,
with [86–89] and without [80–85] requiring the additional detection of a proton in
the final state. While most relevant for LArTPC based oscillation experiments, no
measurements of CCQE-like cross sections on argon with the detection of a proton in
the final state existed until 2021.

This analysis presents the first measurement of exclusive CCQE-like neutrino-

argon interaction cross sections, measured using the MicroBooNE LArTPC [92]. Our
data serve as the first study of exclusive CCQE-like differential cross sections on ar-
gon as well as a benchmark for theoretical models of 𝜈𝜇 -Ar interactions, which are
key for performing a precise extraction of oscillation parameters by future LArTPC
oscillation experiments. We focused on a specific subset of CCQE-like interactions, de-
noted here as CC1p0𝜋, where the contribution of CCQE interactions is enhanced [55].
These include charged-current 𝜈𝜇 -Ar scattering events with a detected muon and ex-
actly one proton, with momenta greater than 100 MeV/𝑐 and 300 MeV/𝑐, respectively.
The measured muon-proton pairs were required to be co-planar with small missing
transverse momentum and minimal residual activity near the interaction vertex that
was not associated with the measured muon or proton. For these CC1p0𝜋 events,
the flux-integrated 𝜈𝜇 -Ar total and differential cross sections in muon and proton mo-
mentum and angle were extracted. The relevant cross sections were further reported
as a function of the calorimetric measured energy and the reconstructed momentum
transfer.

56
 The measurement used data from the MicroBooNE LArTPC detector [93], which
is the first of a series of LArTPCs to be used for precision oscillation measure-
ments [94–99]. As described in chapter 2, the MicroBooNE detector has an active
mass of 85 tons and is located along the Booster Neutrino Beam (BNB) at Fermilab,
463 m downstream from the target. The BNB energy spectrum extends to 2 GeV and
peaks around 0.7 GeV [19]. A neutrino is detected by its interaction with an argon
nucleus in the LArTPC. The secondary charged particles produced in the interac-
tion travel through the liquid argon, leaving a trail of ionization electrons that drift
horizontally and transverse to the neutrino beam direction in an electric field of 273
V/cm, to a system of three anode wire planes located 2.5 m from the cathode plane
detailed in section 2.3. The Pandora tracking package [75] described in section 2.8 is
used to form individual particle tracks from the measured ionization signals. Particle
momenta are determined from the measured track length for protons and multiple
Coulomb scattering pattern for muons [100].

The analysis presented here was performed on data collected from the BNB beam,
with an exposure of 4.59 × 1019 protons on target (POT). At nominal running con-
ditions, one neutrino interaction is expected in ≈ 500 BNB beam spills. The trigger,
based on the scintillation light detected by the 32 photomultiplier tubes (PMTs),
increases the fraction of recorded spills with a neutrino interaction to ≈ 10%. Appli-
cation of additional software selection further rejects background events, mostly from
cosmic muons, to provide a sample that contains a neutrino interaction in ≈ 15% of
the selected spills [74,101]. A CCQE-like event selection, further cosmic rejection and
neutrino-induced background rejection cuts, described in detail in [55], are applied.
Muon-proton pair candidates are identified by requiring two tracks with a common
vertex and an energy deposition profile consistent with a proton and a muon [102].
Further cuts on the track pair opening angle (|∆𝜃𝜇,𝑝 − 90∘ | < 55∘ ) and the muon and
proton track lengths (𝑙𝜇 > 𝑙𝑝 ) reduce the cosmic background rate to less than 1% [55].

The selected CC1p0𝜋 event definition includes events with any number of protons
with momenta below 300 MeV/𝑐, neutrons at any momenta, and charged pions with
momentum lower than 70 MeV/𝑐. The minimal proton momentum requirement of

57
300 MeV/𝑐 is guided by its stopping range in LAr and corresponds to five wire pitches
in the TPC, to ensure an efficient particle identification.

To avoid contributions from cosmic tracks, our CC1p0𝜋 selection considered only
pairs of tracks with a fully-contained proton candidate, and a fully or partially con-
tained muon candidate in the fiducial volume of the MicroBooNE detector. The
fiducial volume is defined by 3 < 𝑥 < 253 cm, -110 < 𝑦 < 110 cm, and 5 < 𝑧 <
1031 cm. The 𝑥 axis points along the negative drift direction with 0 cm placed at
the anode plane, 𝑦 points vertically upward with 0 cm at the center of the detector,
and 𝑧 points along the direction of the beam, with 0 cm at the upstream edge of the
detector. Tracks are fully contained if both the start point and end point are within
this volume, and partially contained if only the start point is within this volume.

We limited our analysis to a phase space region where the detector response to
our signal is well understood and its effective detection efficiency is higher than 2.5%.
This corresponds to 0.1 < 𝑝𝜇 < 1.5 GeV/𝑐, 0.3 < 𝑝𝑝 < 1.0 GeV/𝑐, −0.75 < cos 𝜃𝜇 <
0.95, and cos 𝜃𝑝 > 0.15. Additional kinematical selections were used to enhance the
contribution of CCQE interactions in our CC1p0𝜋 sample. These include requiring
that the measured muon-proton pairs be coplanar (|∆𝜑𝜇,𝑝 −180∘ | < 35∘ ) relative to the
beam axis, have small missing transverse momentum relative to the beam direction
(𝑝𝑇 = |⃗𝑝𝑇𝜇 + 𝑝⃗𝑇𝑝 | < 350 MeV/𝑐), and have a small energy deposition around the
interaction vertex that is not associated with the muon or proton tracks. This event
selection results in a CCQE dominated sample, where table 3.1 shows the fractional
contribution for each interaction channel. Figure 3-1 shows the relevant interaction
breakdown for the entire sample of selected events as a function of cos𝜃𝜇 . The same
nominal MC sample was also used to compute the purity.

After the application of the event selection requirement on the data sample, we
retained 410 CC1p0𝜋 candidate events. It is estimated that our CC1p0𝜋 CCQE-
like event selection purity equals ≈ 84% [55], with ≈ 81% of the measured events
originating from an underlying CCQE interaction as defined by the GENIE event
generator. The efficiency for detecting CC1p0𝜋 events, out of all generated CC1p0𝜋
with an interaction vertex within our fiducial volume, was estimated using our Monte

58
Carlo (MC) simulation and equals ≈ 20% [55]. We note that this efficiency includes
acceptance effects, as the typical LArTPC efficiency for reconstructing a contained
high-momentum proton or muon track is grater than ≈ 90% [75].

Table 3.1: Interaction breakdown after the application of our selection cuts.

Interaction Fractional
Mode Contribution (%)
QE 81.1
MEC 10.9
RES 6.6
DIS 1.4

3000
QE
2500 MEC
RES
2000 DIS
Event Count

1500
1000
500
0
0.6 0.4 0.2 0.0 0.2 0.4 0.6 0.8 1.0
cos( )
Figure 3-1: Interaction breakdown of the cos𝜃𝜇 plot illustrating the dominance of
CCQE interactions after the application of our selection cuts.

Single differential cross sections are reported in measured proton and muon kine-
matics. The differential cross section is given by:

d𝜎 𝑁𝑛on − 𝑁𝑛off − 𝐵𝑛
= , (3.1)
d𝑋𝑛 𝜖𝑛 · Φ𝜈 · 𝑁target · ∆𝑝𝑛

59
 where 𝑋 stands for the kinematical variable that the cross section is differential
in and 𝑛 marks the cross-section bin. In each bin 𝑛, 𝑁𝑛on is the number of measured
events when the beam is on, 𝑁𝑛off is the number of measured events when the beam is
off and cosmic-induced background events are collected, 𝐵𝑛 is the non-CC1p0𝜋 beam-
related background estimated from MC, 𝑁target is the number of scattering nuclei, Φ𝜈
is the integrated incoming neutrino flux, ∆𝜇𝑛 and ∆𝑝𝑛 are the differential bin widths,
and 𝜖𝑛 is the effective particle detection efficiency. We note that the high cos(𝜃𝜇 )
bin has large beam-related background corresponding 𝐵𝑛 in equation 3.1, which is
estimated using the GENIE v2.12.2 based MC simulation and is presented in figure 3-
2.

350 QE
MEC
NonCC1p Bkg Event Count

300 RES
250 DIS
200
150
100
50
0
0.6 0.4 0.2 0.0 0.2 0.4 0.6 0.8 1.0
cos( )
Figure 3-2: Interaction breakdown of the cos𝜃𝜇 plot illustrating the dominance of RES
interactions after the application of our selection cuts for the non-CC1p0𝜋 background
part of the MC sample.

As the detection efficiency is a multidimensional function of the interaction vertex

and the particle momentum and direction, the data were binned in three-dimensional
momentum, in-plane, and out-of-place angle bins with the effective detection efficiency
calculated for each such bin separately and integrated over the interaction vertex in
the detector. The efficiency was extracted based on simulation and is defined as the

60
ratio of the number of reconstructed CC1p0𝜋 events to the number of true generated
CC1p0𝜋 events with a vertex inside our fiducial volume in bin n. This procedure
accounts for bin migration effects such that cross-sections are obtained as a function
of true kinematical variables, as opposed to experimentally reconstructed ones.

The proton and muon efficiencies were extracted independently of each other, such
that, when the cross-section is differential in muon kinematics, the proton kinematics
is integrated over and vise versa. This is done due to the limited data and simulation
statistics and is justified since the proton and muon efficiencies are largely indepen-
dent in the region of interest. The effect of residual correlations is accounted for in the
systematic uncertainties. We further note that the missing transverse momentum re-
quirement increases the sensitivity of our efficiency corrections to the meson exchange
current (MEC) and final state interaction (FSI) models used in our simulations. We
accounted for the model sensitivity in our systematic studies detailed below.

The neutrino flux was predicted using the flux simulation of the MiniBooNE col-
laboration that used the same beamline [82]. We accounted for the small distance
between MiniBooNE and MicroBooNE. Neutrino cross section modeling uncertainties
were estimated using the GENIE framework of event reweighting [103, 104] with its
standard reweighting parameters. For both cross section and flux systematics, we
use a multisim technique [105], which consists of generating many MC replicas, each
one called a “universe”, where model parameters are varied within their uncertainties.
Each universe represents a different reweighting. The simultaneous reweighting of all
model parameters allows the correct treatment of their correlations.

A different model is followed for detector model systematic uncertainties, that

are dominated by individual detector parameters. Unisim samples [105] were gener-
ated, where one detector parameter was varied each time by 1𝜎. We then examined
the impact of each parameter variation on the extracted cross sections, by obtain-
ing the differences with respect to the central value on a bin-by-bin basis. We note
that the detection efficiency used for the cross section extraction is re-evaluated for
each variation separately, including bin migration corrections. This procedure there-
fore accounted for the systematic uncertainty in these corrections due to both the

61
cross-section and detector response modeling. We then defined the total detector
1𝜎 systematic uncertainty by summing in quadrature the effect of each individual
variation.
A dedicated MC simulation was used to estimate possible background from events
in which a neutrino interacts outside the MicroBooNE cryostat but produce parti-
cles that enter the TPC and pass the event selection cuts [85]. No such events were
found in that study, which is also supported by our observation that the 𝑧-vertex
distributions for the measured events follows a uniform distribution, as can be seen
in figure 3-3. The measured 𝑧-vertex distribution, after the beam related MC back-
ground has been subtracted, does not show an excess at low-𝑧, which indicates that
background events from interactions upstream of the detector are not accidentally
entering our selection, which would show up as a small-𝑧 enhancement in our vertex
distribution. The deficit at 𝑧 = 700 cm is due to dead wires in our detector and its
effect has been incorporated in our simulation.

175
40
Efficiency Corrected Beam On Events

150

30 125
Beam On Events

100
20
75

50
10
25

0 0
0 200 400 600 800 1000 0 200 400 600 800 1000
vertex z [cm] vertex z [cm]

Figure 3-3: Vertex 𝑧 distribution for the measured events, after the beam related MC
background has been subtracted, before (left) and after (right) detection efficiency
corrections. No small-𝑧 enhancement is observed and, with efficiency corrections, the
measured distribution is consistent with that of a uniform neutrino interaction vertex.

The MC simulation used to estimate the backgrounds and effective efficiency con-
tains real cosmic data overlayed onto a neutrino interaction simulation that uses
GENIE [103, 104] to simulate both the signal events and the beam backgrounds [55].
For the simulated portion, the particle propagation is based on GEANT4 [58], while

62
the simulation of the MicroBooNE detector is performed in the LArSoft frame-
work [106,107]. The beam-related background subtracted from the candidate CC1p0𝜋
events in the data sample is simulated.

3.1.2 Quasielastic-like Cross-Section Results

Figure 3-4 shows the flux integrated single differential CC1p0𝜋 cross section as a func-
tion of the cosine of the measured muon scattering angle. The data were compared to
several theoretical calculations and to our GENIE-based MC prediction. This predic-
tion is the result of analyzing a sample of MC events produced using our “nominal”
GENIE model and propagated through the full detector simulation in the same way as
data. This model (GENIE v2.12.2) [103,104] treats the nucleus as a the Bodek-Ritchie
Fermi Gas (RFG), used the Llewellyn-Smith CCQE scattering prescription [108], the
empirical MEC model [109], the Rein-Sehgal resonance (RES) model, the coherent
(COH) scattering model [110], and a data driven FSI model denoted as “hA” [111].
In addition, theoretical predictions by several other event generators are shown
at the cross-section level without any detector effects [112]. These include GENIE
v2.12.2 and v3.0.6 [103, 104], NuWro 19.02.1 [113], and NEUT v5.4.0 [114]. The
agreement between the “nominal” GENIE calculation (v2.12.2) and the MC predic-
tion constitutes a closure test for our analysis. The other generators all improve on
GENIE v2.12.2 by using updated nuclear interaction models, among which is the use
of a Local Fermi Gas model (LFG) [37] and Random Phase Approximation (RPA)
correction [115]. GENIE v3.0.6 and NEUT also include Coulomb corrections for the
outgoing muon [116]. The theoretical models implemented in these event generators
include free parameters that are typically fit to data, with different generators using
different data sets. We also consider the GiBUU 2019 [117] event generator which
fundamentally differs from the others due to its use of a transport equation approach.
A brief discussion of the underlying model configuration used in the different event
generator predictions included in this analysis is shown below.

• GENIE Nominal: Uses the aforementioned GENIE v2.12.2 modeling.

63
 18
MicroBooNE Data 4.59×1019 POT
15 MC
Simulation
12 GENIE Nominal
]
38 cm2
Ar
GENIE v3.0.6
10 9 NuWro 19.02.1
NEUT v5.4.0
[

GiBUU 2019
dcos( )

6
d

3
0
0.7 0.4 0.0 0.4 0.8 1.0
cos( )
Figure 3-4: The flux integrated single differential CC1p0𝜋 cross sections as a function
of the cosine of the measured muon scattering angle. Inner and outer error bars show
the statistical and total (statistical and systematic) uncertainty at the 1𝜎, or 68%,
confidence level. Colored lines show the results of theoretical absolute cross section
calculations using different event generators (without passing through a detector sim-
ulation). The blue band shows the extracted cross section obtained from analyzing
MC events propagated through our full detector simulation. The width of the band
denotes the simulation statistical uncertainty.

• NuWro 19.02.1 [113]: Using the LFG ground state model [37], the Llewellyn-
Smith CCQE scattering prescription [108], the Transverse Enhancement model
for two–body currents [118], the Adler-Rarita-Schwinger formalism to calculate
the ∆ resonance explicitly [119], the BS COH [120] scattering model and an
intranuclear cascade model for FSI.

• NEUT v5.4.0 [114]: Using the LFG ground state model [37], the Nieves CCQE
scattering prescription [121], the Nieves MEC model [122], the BS RES [123–
126] and Rein-Sehgal COH [110] scattering models, and FSI with Oset medium
correction for pions [103, 104].

• GiBUU 2019: Using somewhat similar models, but unlike other generators,
those are implemented in a coherent way, by solving the Boltzmann-Uehling-

64
 Uhlenbeck transport equation [117]. The models include: Local Fermi Gas
model [37], standard CCQE expression [127], empirical MEC model and a dedi-
cated spin dependent resonance amplitude calculation following the MAID anal-
ysis [128]. The DIS model is as in PYTHIA [129] and the FSI treatment is
different as the hadrons propagate through the residual nucleus in a nuclear
potential which is consistent with the initial state.

As can be seen in figure 3-4, all models are in overall good agreement with our
data, except for the highest cos(𝜃𝜇 ) bin with cos(𝜃𝜇 )> 0.8, where the measured cross
section is significantly lower than the theoretical predictions. This discrepancy cannot
be explained by the systematic uncertainties and is therefore indicative of an issue
with the theoretical models. Specifically, high cos(𝜃𝜇 ) correspond to low momentum
transfer events which were previously observed to not be well reproduced by theory
in inclusive reactions [84, 85] and is now also seen in exclusive reactions.
As the differential cross sections in proton kinematics and muon momentum in-
clude contributions from all muon scattering angles, their agreement with the the-
oretical calculation is affected by this disagreement in the forward muon scattering
angle. Therefore, for the results presented below, we repeated the cross-section ex-
traction exercise twice, where the first time we included all the events that satisfy
our selection criteria, and the second time we excluded those events with cos(𝜃𝜇 )>
0.8. The corresponding integrated measured CC1p0𝜋 cross sections are summarized
in table 3.2. The same table also lists the 𝜒2 for the agreement of the different models
with the data for differential cross sections for the full available phase-space and for
cos(𝜃𝜇 ) < 0.8. The values reported in the table are the simple sum of those 𝜒2 values
obtained for each distribution separately. Systematic uncertainties and correlations
were accounted for using covariance matrices.
As can be seen in table 3.2, GENIE v3.0.6 is the only model that reaches a 𝜒2 /d.o.f.
close to unity for the full phase-space. It is also the closest model to the data at the
highest cos(𝜃𝜇 ) bin. For all other models, the 𝜒2 /d.o.f. in the cos(𝜃𝜇 ) < 0.8 sample
is reduced by a factor of ∼ 2 as compared to the full phase-space sample. GENIE
v3.0.6 shows a smaller reduction in this case, and GiBUU 2019 obtains a consistently

65
Table 3.2: Integrated cross section values and 𝜒2 values for the agreement between
the measured cross sections and various event generators. Results are listed for the
full measured phase space and for a limited one of cos(𝜃𝜇 ) < 0.8.
Integrated Cross Section [10−38 cm2 ]
(Differential Cross Section 𝜒2 /d.o.f)
−0.75 < cos(𝜃𝜇 ) < 0.95 −0.75 < cos(𝜃𝜇 ) < 0.8
Data CC1𝑝0𝜋 4.93 ± 1.55 4.05 ± 1.40
Generators GENIE Nominal 6.18 (63.2/28) 4.04 (30.1/27)
GENIE v3.0.6 5.45 (34.6/28) 3.66 (21.4/27)
NuWro 19.02.1 6.67 (76.7/28) 4.39 (29.9/27)
NEUT v5.4.0 6.64 (78.5/28) 4.39 (32.2/27)
GiBUU 2019 7.00 (82.2./28) 4.78 (40.0/27)

higher 𝜒2 /d.o.f. for both the full and limited phase-space samples.

-0.65 < cos( ) < 0.95 Simulation MicroBooNE Data 15 -0.65 < cos( ) < 0.95
10 15
GENIE Nominal
GENIE v3.0.6
MC
NuWro 19.02.1
NEUT v5.4.0 10
10 GiBUU 2019
5
5
5
]

]
]
GeV/c Ar

GeV/c Ar
38 cm2
Ar
38 cm2

38 cm2
-0.65 < cos( ) < 0.95
0 -0.65 < cos( ) < 0.8 -0.65 < cos( ) < 0.8
0 -0.65 < cos( ) < 0.8
10

10 10 10
[
10

10
dcos( p )
[

[
d

5
dpp

5 5
dp

d
d

0 0
0.4 0.8 1.2 0.2 0.4 0.6 0.8 0.4 0.6 0.8 1.0
p [GeV/c] cos( p) pp [GeV/c]
Figure 3-5: As figure 3-4, but for the differential cross sections as a function of mea-
sured muon momentum (left) and measured proton scattering angle (middle) and
momentum (right). Cross sections are shown for the full measured phase-space (top)
and for events with cos(𝜃𝜇 ) < 0.8 (bottom). Inner and outer error bars show the
statistical and total (statistical and systematic) uncertainty at the 1𝜎, or 68%, con-
fidence level. Colored lines show the results of theoretical absolute cross section
calculations using different event generators (without passing through a detector sim-
ulation). The blue band shows the extracted cross section obtained from analyzing
MC events passed through our full detector simulation.

Figure 3-5 shows this comparison between the relevant cross sections in the full
available phase-space (top) and in the case where events with cos(𝜃𝜇 ) > 0.8 are
excluded (bottom). Removing this part of the phase-space significantly improves the
agreement between data and theory. The improved agreement with the data observed

66
for GENIE v3.0.6, especially for the full phase-space sample, is intriguing. Specifically,
GENIE v3.0.6 and NEUT v5.4.0 are quite similar, using the same nuclear, QE, and
MEC models, which are the most significant processes in our energy range. They
do differ in the Coulomb corrections that only GENIE v3.0.6 and NEUT have, their
free parameter tuning process, and the implementation of RPA correction, that are
known to be important at low momentum transfer [115]. Our data indicates that
these seemingly small differences can have a highly significant impact, as seen in
table 3.2.

MicroBooNE Data Simulation 10 -0.65 < cos( ) < 0.95

15 MC GENIE Nominal
GENIE v3.0.6
NuWro 19.02.1
10 NEUT v5.4.0
GiBUU 2019 5
5
]

]
GeV2 /c2 Ar

GeV Ar
38 cm2
cm2

0 -0.65 < cos( ) < 0.95

-0.65 < cos( ) < 0.8 10 -0.65 < cos( ) < 0.8
10
38

10
10

[
[

dEcal

5 5
d
dQ2CCQE
d

0 0 0.4
0.0 0.4 0.8 1.2 0.8 1.2
QCCQE [GeV2 /c2 ]
2
Ecal [GeV]
Figure 3-6: The flux integrated single differential CC1p0𝜋 cross sections as a function
of 𝑄2𝐶𝐶𝑄𝐸 = (𝐸𝜈𝑐𝑎𝑙 − 𝐸𝜇 )2 − (⃗𝑝𝜈 − 𝑝⃗𝜇 )2 and 𝐸𝜈𝑐𝑎𝑙 = 𝐸𝜇 + 𝑇𝑝 + 𝐵𝐸, where 𝐵𝐸 = 40
MeV and 𝑝⃗𝜈 = (0, 0, 𝐸𝜈𝑐𝑎𝑙 ). Colored lines show the results of theoretical absolute
cross section calculations using different event generators (without passing through a
detector simulation). The blue band shows the extracted cross section obtained from
analyzing MC events passed through our full detector simulation.

Lastly, figure 3-6 shows the flux-integrated single differential cross sections as a
function of calorimetric measured energy and reconstructed momentum transfer, with
and without events with cos(𝜃𝜇 ) > 0.8. The former is defined as 𝐸𝜈𝑐𝑎𝑙 = 𝐸𝜇 + 𝑇𝑝 + 𝐵𝐸,
and the latter as 𝑄2𝐶𝐶𝑄𝐸 = (⃗𝑝𝜈 − 𝑝⃗𝜇 )2 − (𝐸𝜈𝑐𝑎𝑙 − 𝐸𝜇 )2 , where E𝜇 is the muon energy,
T𝑝 is the proton kinetic energy, BE = 40 MeV is the effective nucleon binding energy
for argon, and 𝑝⃗𝜈 = (0, 0, 𝐸𝜈𝑐𝑎𝑙 ) is the reconstructed interacting neutrino momentum.

67
𝐸𝜈𝑐𝑎𝑙 is often used as a proxy for the true neutrino energy. Overall, good agreement is
observed between data and calculations for these complex variables, even for the full
event sample without the cos(𝜃𝜇 ) < 0.8 requirement.
The systematic uncertainty of our measurement summed up to 26.2% and included
contributions from the neutrino flux prediction and POT estimation (18.7%), detector
response modeling (18.4%), imperfect proton and muon efficiency decoupling (5.7%),
and neutrino interaction cross section modeling (7.1%).

3.1.3 Quasielastic-like Cross-Section Analysis Conclusions

In summary, the first measurement of 𝜈𝜇 CCQE-like differential cross sections on ar-

gon was reported for event topologies with a single muon and a single proton detected
in the final state using data sets from the MicroBooNE LArTPC. The data are in
good agreement with simulation predictions, except at small muon scattering angles
that correspond to low-momentum-transfer reactions. This measurement confirmed
and constrained calculations essential for the extraction of oscillation parameters and
highlights kinematic regimes where improvement of theoretical models is required.
The benchmarking of exclusive CC1p0𝜋 cross sections on argon presented here sug-
gests that measurements of CC1p0𝜋 interactions are a suitable choice for use in pre-
cision neutrino oscillation analyses, especially after theoretical models are reconciled
with the small scattering angle data.

3.2 First Multidimensional Measurement Of Kine-

matic Imbalance Cross Sections On Argon

3.2.1 Kinematic Imbalance Neutrino Data Analysis

Over the course of two years (2019-2021), the MicroBooNE collaboration made sig-
nificant improvements to the pre-existing analysis framework. These improvements
provided high statistics neutrino-argon data sets, improved signal processing [74],
reduced detector systematics [130], a theory-driven interaction modeling [79], and

68
the creation of the first MicroBooNE tune [131]. Figure 3-7 illustrates the improved
data-MC agreement after the implementation of these changes as a function of cos𝜃𝜇 ,
where the disagreement in the forward direction is longer observed. The improved
picture at cos𝜃𝜇 ≈ 1 is primarily driven by the improved modeling of the MC beam
related backgrounds.

NuWro 19.02.1 (21.86/18)

GiBUU 2021 (9.51/18)
NEUT v5.4.0 (14.38/18)
20 GENIE v3 G18 Tune (21.36/18)
MicroBooNE Data
2

(Stat ⊕ Shape Unc)

Ar
10-38 cm

6.79e+20 POT
15 Norm Unc

10
dcosθµ
dσ

0
−1 −0.5 0 0.5 1
cosθµ

Figure 3-7: Muon angular distribution after the implementation of the analysis frame-
work improvements. No data-MC disagreement is observed in the forward direction.

Motivated by these improvements, in this analysis the first study of kinematic

imbalance variables on argon, which are sensitive to nuclear effects, is reported. These
variables are studied using CC1p0𝜋 events within a neutrino slice, as defined by the
Pandora reconstruction framework and detailed in section 2.8. Exactly one muon
with 0.1 < 𝑝𝜇 < 1.2 GeV/c, exactly one proton with 0.3 < 𝑝𝑝 < 1 GeV/c, no charged
pions above the 0.07 GeV/c threshold, and no other mesons of any momenta are
required in the final state originating from charged-current 𝜈𝜇 -Ar scattering events.
The existence of any number of neutrons, electrons or photons is allowed.
Such kinematic imbalance variables of interest include the transverse variables
(TVs), namely 𝛿𝑝𝑇 , 𝛿𝛼𝑇 and 𝛿𝜑𝑇 [132, 133]. These are built specifically to charac-
terize and minimize the degeneracy between the nuclear effects most pertinent to
long-baseline oscillation experiments. In particular, the TVs facilitate the possible

69
identification of the Fermi motion of the initial state nucleon, the final state re-
interactions of the nucleons in the nucleus and the multi-nucleon interactions (2p2h).
As shown in figure 3-8, the TVs are defined by projecting the lepton and proton
momentum on the plane perpendicular to the neutrino direction.

Figure 3-8: Schematic illustration of the single transverse variables 𝛿𝑝𝑇 , 𝛿𝛼𝑇 and 𝛿𝜑𝑇 .
Figure adapted from [134].

In the absence of any nuclear effects, the proton and muon momenta are equal and
opposite in this plane and therefore the measured difference between their projections
is a direct probe of nuclear effects in quasi-elastic (QE) events. 𝛿⃗𝑝𝑇 can be fully
characterized in terms of the vector magnitude (𝛿𝑝𝑇 ) and the two angles (𝛿𝛼𝑇 and
𝛿𝜑𝑇 ):

𝛿𝑝𝑇 = |⃗𝑝𝑇 ℓ + 𝑝⃗𝑇 𝑝 | (3.2)

𝛿𝛼𝑇 = 𝑎𝑟𝑐𝑐𝑜𝑠( −⃗
𝑝𝑇 ·𝛿⃗ℓ𝑝𝑇
) (3.3)
𝑝𝑇 ℓ ·𝛿𝑝𝑇

𝛿𝜑𝑇 = 𝑎𝑟𝑐𝑐𝑜𝑠( −⃗
𝑝𝑇 ·⃗ ℓ
𝑝𝑇
)
𝑝
(3.4)
𝑝𝑇 ℓ ·𝑝𝑇 𝑝

where 𝑝⃗𝑇 ℓ
and 𝑝⃗𝑇 𝑝
are, respectively, the projections of the momentum of the
outgoing lepton and proton on the transverse plane. Different nuclear effects alter
the distributions of the TVs in different and predictable ways. Measurements of the

70
TVs therefore have a unique sensitivity to identify nuclear effects. This allows cross
sections extracted using these observables to act as a powerful tool to tune and dis-
tinguish nuclear models. Furthermore, in case of disagreement, the TV distributions
provide useful hints on the possible causes of the discrepancies.

Another kinematic variable of interest corresponds to the total momentum of

the struck nucleon. The formalism introduced in [132] is adopted. This formalism
provides an approximation for the longitudinal component of the struck nucleon mo-
mentum shown in equation 3.5, which is derived in appendix 8.1,

1 𝑚2𝐴−1 + 𝛿𝑝2𝑇
𝛿𝑝𝐿 = 𝑅 − . (3.5)
2 2𝑅

For simplicity, we defined

𝑅 ≡ 𝑚𝐴 + 𝑝𝜇𝐿 + 𝑝𝑝𝐿 − 𝐸 𝜇 − 𝐸 𝑝 (3.6)

Combining information from both the longitudinal and the transverse components
gave us access to an approximation for the total struck nucleon momentum

√︁
𝑝𝑛,𝑝𝑟𝑜𝑥𝑦 = 𝛿𝑝2𝐿 + 𝛿𝑝2𝑇 (3.7)

The muon-proton momentum imbalances introduced in [135] parallel and trans-

verse to 𝛿⃗𝑝𝑇 , as shown in figure 3-9, are explored,

𝑝𝜈 × 𝑝ˆ𝜇𝑇 ) · 𝛿⃗𝑝𝑇
𝛿𝑝𝑇 𝑥 = (ˆ
(3.8)
𝑝𝜇𝑇 · 𝛿⃗𝑝𝑇 ,
𝛿𝑝𝑇 𝑦 = −ˆ

71
 and, in terms of the magnitudes,

𝛿𝑝𝑇 𝑥 = 𝛿𝑝𝑇 · 𝑠𝑖𝑛𝛿𝛼𝑇

(3.9)
𝛿𝑝𝑇 𝑦 = 𝛿𝑝𝑇 · 𝑐𝑜𝑠𝛿𝛼𝑇 .

Figure 3-9: Schematic illustration of 𝛿𝑝𝑇 𝑥 and 𝛿𝑝𝑇 𝑦 . Figure adapted from [135].

The measured 𝛿𝑝𝑇 𝑥 event distribution shown in figure 3-10 (left) using the “com-
bined” MicroBooNE runs 1-3 exhibit a QE peak near 0. If the interaction had occurred
on a free nucleon, then a delta function would be expected at 0 because the muon
and proton final states must balance. The width of the QE peak mostly results from
the Fermi motion. If no significant deviation is assumed in the non-QE distributions
originating from MEC and RES/DIS events, then data-MC discrepancies could im-
ply an overestimation of the argon Fermi momentum, and/or a difference in the total
fraction of the FSI contribution.
Unlike the 𝛿𝑝𝑇 𝑥 distribution, a non-QE tail is observed towards the negative 𝛿𝑝𝑇 𝑦
values shown in figure 3-10 (right). Inelastic events such as 2p2h, resonance, and DIS
are inefficient at transferring the lepton momentum to the final state nucleons, since
multiple initial states particles are often involved. Therefore, the protons tagged in

72
the non-QE events will in general have less momenta then the muons and the 𝛿𝑝𝑇 𝑦
distribution is shifted to the left.
BeamOn (9051) CCQE (4744) CCMEC (1547) BeamOn (9051) CCQE (4744) CCMEC (1547)
ExtBNB (642) CCRES (1035) CCDIS (181) ExtBNB (642) CCRES (1035) CCDIS (181)

Combined CCQE = 57.7 % 2500 Combined CCQE = 57.7 %

2500 Cosmics = 7.8 % Cosmics = 7.8 %
# Events / 6.79e+20

# Events / 6.79e+20
2000
2000
1500
1500

1000
1000

500 500

0 0
δp [GeV/c] 1.3 δp [GeV/c]
1.3 T,x T,y
MC+ExtBNB

MC+ExtBNB
1.2 1.2
BeamOn

BeamOn
1.1 1.1
1 1
0.9 0.9
−0.4 −0.2 0 0.2 0.4 −0.6 −0.4 −0.2 0 0.2 0.4
δp [GeV/c] δpT,y [GeV/c]
T,x

Figure 3-10: Interaction breakdown of the CC1p0𝜋 events as a function of 𝛿𝑝𝑇 𝑥 (left)
and 𝛿𝑝𝑇 𝑦 (right). The data correspond to the “combined” MicroBooNE runs 1-3.

To avoid multiple cosmic contributions and tracks from trajectories that exit the
detector but their end-points are incorrectly reconstructed around its edges, a fiducial
volume of

10 < 𝑥 < 246, −105 < 𝑦 < 105, 10 < 𝑧 < 1026 cm (3.10)

is defined. Candidate muon and proton tracks that were fully contained in this
region were considered and their momenta were obtained based on their range [136,
137].
The log-likelihood ratio particle identification (LLR PID) score method [138] is
used to obtain our muon and proton candidates. The candidate track with the greater
LLR PID score was assigned the label of the candidate muon, while the one with the
smaller 3-plane loglikelihood was our candidate proton.
To minimize the contribution of misreconstructed tracks, we took advantage of
the fact that we had two muon momentum reconstruction methods available for con-
tained tracks, namely the momentum from range [137] and the one from Multiple
Coulomb Scattering (MCS) [100]. A quality cut was applied on the contained muons
by requiring the range and MCS momenta to be in agreement within 25%.
In order to avoid flipped tracks, it was further required that the distance between

73
the track start points and the vertex is smaller than the corresponding distance be-
tween the track end points and the vertex. It was also required that the distance
between the start points of the two candidate tracks is smaller than the one between
the two end points.
The maximal possible signal contribution was ensured, while the majority of the
cosmic contamination and the beam related MC backgrounds were rejected, by re-
quiring that the proton LLP PID score is less than 0.05.
The application of our event selection resulted in 9051 candidate events in our data
sample. Using the MC, it was estimated that our event selection yielded a purity of
≈ 70% and an efficiency of ≈ 10%. There was also some contribution from the
remaining cosmic contamination (≈ 8%). After the application of the event selection,
topological and interaction breakdowns for the kinematic variables of interest, such
as the ones shown in figure 3-11 for 𝛿𝑝𝑇 , were obtained.
BeamOn (9051) Overlay CC1p (5673) Overlay NonCC1p (1837) BeamOn (9051) CCQE (4744) CCMEC (1547)
ExtBNB (642) Dirt CC1p (0) Dirt NonCC1p (62) ExtBNB (642) CCRES (1035) CCDIS (181)

Combined CC1p = 69 % Combined CCQE = 57.7 %

Cosmics = 7.8 % Cosmics = 7.8 %
# Events / 6.79e+20

# Events / 6.79e+20

1500 1500

1000 1000

500 500

0 0
δ p [GeV/c] δ p [GeV/c]
T T
MC+ExtBNB

MC+ExtBNB

1.2 1.2
BeamOn

BeamOn

1.1 1.1
1 1
0.9 0.9
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

Figure 3-11: Topological (left) and interaction (right) breakdown after the application
of the event selection for 𝛿𝑝𝑇 .

3.2.2 Cross-Section Extraction Technique

The unfolded cross-section results reported in this analysis took advantage of the
Wiener-SVD unfolding [139]. This method combines the use of the singular value
decomposition (SVD) unfolding and a Wiener filter. SVD unfolding [140], such as
the Tikhonov regularisation [141], unfolds a distribution by minimising a 𝜒2 function
comparing a prediction to data. To avoid the large variance introduced, a penalty

74
term is added to regularise the curvature (second derivative) of the results. The
strength of such a term is determined by finding an appropriate trade-off between the
bias and the variance between the data and the MC. More details on the technique
are included in appendix 8.2. In order to report the cross-section results, two key
ingredients are required, namely the response and covariance matrices.

The construction of the response matrices uses the selected MC CC1p0𝜋 events
to construct a two-dimensional (2D) object, where each entry in true bin i and re-
constructed bin j (𝑁 𝑡𝑟𝑢𝑒 𝑖,𝑟𝑒𝑐𝑜 𝑗 ) is divided by the true number of events generated in
bin i (𝑆 𝑡𝑟𝑢𝑒 𝑖 ). These response matrices serve as “2D local efficiencies”, as defined in
equation 3.11 and can be seen in figure 3-12 for 𝛿𝑝𝑇 .

𝑁 𝑡𝑟𝑢𝑒 𝑖,𝑟𝑒𝑐𝑜 𝑗
𝑀𝑖𝑗 = (3.11)
𝑆 𝑡𝑟𝑢𝑒 𝑖

Combined Response Matrix

0.8
0.00 0.00 0.00 0.00 0.00 0.00 0.00
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.01
± 0.00
0.03
± 0.00 0.06
Reco δp [GeV/c]

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.05 0.01
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.01 ± 0.00

0.6 0.00 0.00 0.00 0.00 0.00 0.00

± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
0.00
± 0.00
0.00
± 0.00
0.01
± 0.00
0.05
± 0.00
0.02
± 0.00
0.00
± 0.00

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.05 0.02 0.00 0.00
0.04
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
T

0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01 0.06 0.02 0.00 0.00 0.00
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00

0.4 0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.01
± 0.00
0.05
± 0.00
0.02
± 0.00
0.01
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00 0.00 0.00 0.00 0.00 0.02 0.05 0.02 0.01 0.00 0.00
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
0.00
± 0.00
0.00
0.00
± 0.00
0.00
0.00
± 0.00
0.00
0.00
± 0.00
0.02
0.01
± 0.00
0.05
0.06
± 0.00
0.03
0.02
± 0.00
0.00
0.00
± 0.00
0.00
0.00
± 0.00
0.00
0.00
± 0.00
0.00
0.00
± 0.00
0.00
± 0.00
0.00
0.02
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
0.2 0.00
± 0.00
0.00
± 0.00
0.02
± 0.00
0.06
± 0.00
0.02
± 0.00
0.01
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.00
± 0.00
0.01 0.02 0.06 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
0.03 0.06 0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
0.05 0.01 0.00 0.00 0.00 0.00 0.00 0.00 0.00
± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00 ± 0.00
0 0
0 0.2 0.4 0.6 0.8
True δp [GeV/c]
T

Figure 3-12: Response matrices of 𝛿𝑝𝑇 using the selected CC1p0𝜋 MC events.

The method also uses a covariance matrix constructed from the MC flux nor-
malised event rate as an input. The total covariance matrix incorporates information
related to the systematic and statistical uncertainties. The flux-normalized MC event

75
rates in reconstructed space were obtained as

𝑁 𝑟𝑒𝑐𝑜 𝑖
˜ 𝑟𝑒𝑐𝑜 𝑖 =
𝜎 , (3.12)
Φ𝐶𝑉
𝜈 × 𝑁𝑡𝑎𝑟𝑔𝑒𝑡𝑠
where 𝑁 𝑟𝑒𝑐𝑜 𝑖 = 𝑀𝑖𝑗 × 𝑆 𝑡𝑟𝑢𝑒 𝑗 + 𝐵 𝑟𝑒𝑐𝑜 𝑖 is the total number of reconstructed events in
bin i, 𝑀𝑖𝑗 is the response matrix corresponding to reco bin i and true bin j as defined in
equation 3.11, 𝑆 𝑡𝑟𝑢𝑒 𝑗 is the true signal without any detector or reconstruction effects
in bin j, and 𝐵 𝑟𝑒𝑐𝑜 𝑖 is the total number of reconstructed beam-related MC background
events in bin i. Substituting 𝑁 𝑟𝑒𝑐𝑜 𝑖 into equation 3.12 yields

𝑀𝑖𝑗𝑢𝑛𝑖𝑣 × 𝑆 𝑡𝑟𝑢𝑒 𝑗 𝐶𝑉 + 𝐵 𝑟𝑒𝑐𝑜 𝑖 𝑢𝑛𝑖𝑣

˜ 𝑟𝑒𝑐𝑜 𝑖 =
𝜎 . (3.13)
Φ𝐶𝑉
𝜈 × 𝑁𝑡𝑎𝑟𝑔𝑒𝑡𝑠
For each systematic variation, each term in equation 3.13 labelled with “univ” is
reweighted/modified and each term labelled with “CV” is fixed to the central value.
The integrated flux, Φ𝐶𝑉
𝜈 , remains fixed for each variation.

In the case of the cross section variations, the calculation of the response matrix
in each universe is slightly modified via the normalization to the true signal in a given
universe 𝑆 𝑡𝑟𝑢𝑒 𝑗 𝑢𝑛𝑖𝑣 ,

𝑁 𝑡𝑟𝑢𝑒 𝑗,𝑟𝑒𝑐𝑜 𝑖 𝑢𝑛𝑖𝑣

𝑀𝑖𝑗𝑢𝑛𝑖𝑣 = (3.14)
𝑆 𝑡𝑟𝑢𝑒 𝑗 𝑢𝑛𝑖𝑣
This treatment of the systematic uncertainties addresses both the signal and the
beam related background uncertainties.
Using these flux-normalized event rates 𝜎
˜ , a covariance matrix, 𝐸𝑖𝑗 , can be calcu-
lated using the central value and 𝑁𝑢𝑛𝑖𝑣 multisims with the covariance formalism,

𝑁∑︁
𝑢𝑛𝑖𝑣
1
𝐸𝑖𝑗 = 𝜎𝑖𝑢𝑛𝑖𝑣 − 𝜎
(˜ ˜𝑖𝐶𝑉 )(˜
𝜎𝑗𝑢𝑛𝑖𝑣 − 𝜎
˜𝑗𝐶𝑉 ) (3.15)
𝑁𝑢𝑛𝑖𝑣 𝑠=0

where 𝑁𝑢𝑛𝑖𝑣 is number of alternative universes, 𝜎

˜𝑖𝑢𝑛𝑖𝑣 corresponds to the variation
and 𝜎
˜𝑖𝐶𝑉 to the central-value prediction.
The unfolding model uncertainty is accessed by comparing the data spectra un-
folded with G18 to the data spectra unfolded with two alternative G18 configurations.

76
The former one did not include the effect of the MicroBooNE tune and the latter in-
cluded an additional weight of 2 on the MEC events. The spread between the three
configurations on a bin-by-bin basis normalized to (2) is assigned as an additional
√︀

uncertainty [142].
The statistical uncertainty of our measurement is 1.5%. The total uncertainty
sums to 13% and includes contributions from the neutrino flux prediction (7.3%),
unfolding model uncertainty (7.3%), neutrino interaction cross section modeling (5%),
detector response modeling (4.9%), POT estimation (2.3%), number-of-scattering-
targets (1.15%), reinteractions (1%), and out-of-cryostat interaction modeling (0.2%).
Figure 3-13 shows the total covariance matrix due to the aforementioned sources
of uncertainty for 𝛿𝑝𝑇 .

Combined Total Covariance Matrix×10−3

0.8
j bin δp [GeV/c]

0.4
0.6
T

0.4
0.2

0.2

0 0
0 0.2 0.4 0.6 0.8
i bin δp [GeV/c]
T

Figure 3-13: Total covariance matrix for 𝛿𝑝𝑇 .

The Wiener SVD unfolding machinery returns an unfolded data cross section
along with an unfolded covariance matrix and an additional smearing matrix, 𝐴𝑐 .
The corresponding 𝐴𝑐 matrix for 𝛿𝑝𝑇 is shown in figure 3-14. The smearing matrix
𝐴𝑐 contains information about the regularisation of the measurement and is applied
to the true model cross section predictions when compared to the data. Therefore, the
result of our measurement lives in a “regularized” phase-space, which is not identical
to the true phase-space.

77
 Combined
0.6
0.8 0.5
0.4

δp [GeV/c]
0.6
0.3
T
0.4 0.2
0.1
0.2 0
−0.1
0
0 0.2 0.4 0.6 0.8
δp [GeV/c]
T

Figure 3-14: Additional smearing matrix 𝐴𝑐 for 𝛿𝑝𝑇 .

3.2.3 Event Generator Modeling And Configurations

The extracted cross sections were compared to GENIE v3.0.6 G18_10a_02_11a

(G18) and the theory-driven GiBUU 2021 (GiBUU) event generator. Additional com-
parisons to the corresponding events generators when FSI are turned off were also
included (G18 No FSI and GiBUU No FSI). G18 uses the Local Fermi Gas model [37],
the Nieves CCQE scattering prescription [121] which includes Coulomb corrections
for the outgoing muon [116] and Random Phase Approximation correction [115], the
Nieves MEC model [122], the KLN-BS RES [123–126] and Berger-Sehgal COH [120]
scattering models, the hA2018 FSI model [143], and the T2K tune weights [131].

GiBUU uses somewhat similar models, but, unlike GENIE, those are imple-
mented in a coherent way, by solving the Boltzmann-Uehling-Uhlenbeck transport
equation [117]. The modeling includes the Local Fermi Gas model [37], a standard
CCQE expression [127], an empirical MEC model and a dedicated spin dependent
resonances amplitude calculation following the MAID analysis [128]. The DIS model
is as in PYTHIA [129] and the FSI treatment is different as the hadrons propagate
through the residual nucleus in a nuclear potential which is consistent with the initial
state.

78
 Apart from the nominal G18 prediction, we further included a comparison to the
recently added theory driven GENIE v3.0.6 G21_11b_00_000 configuration (G21
hN). The latter uses the SuSAv2 model for QE and MEC interactions [28], and the
hN2018 FSI model [144]. The modeling options for RES, DIS, and COH interactions
are the same as for G18. We investigated the effect of the FSI modeling choice by
comparing the G21hN results to the ones obtained with G21 hA, where the hA2018
FSI model was used instead, and to G21 G4 with the recently coupled Geant4 FSI
framework [145].
Lastly, our results present the comparison between the nominal G18 LFG model
and predictions using the same G18 modeling configuration but different nuclear
model options available in the GENIE event generator, namely the Bodek-Ritchie
Fermi Gas (G18 RFG) [36] and an Effective Spectral Function (G18 EffSF) [38].
Furthermore, the prediction without Random Phase Approximation (RPA) effects
was used for comparison (G18 No RPA) [115].

3.2.4 Kinematic Imbalance Differential Cross-Section Results

The single- and double- in 𝛿𝛼𝑇 bins differential unfolded cross sections as a func-
tion of 𝛿𝑝𝑇 are presented in figure 3-15. The single-differential results as a function
of 𝛿𝑝𝑇 using all the events that satisfy our selection are shown in the top panel.
The peak height of both generator predictions is ≈ 30% higher when FSI effects are
turned off. Yet, all distributions illustrate a transverse missing momentum tail that
extends beyond the Fermi momentum whether FSI effects are activated or not. The
ratio between the generator predictions with and without FSI is shown in the insert
and illustrates significant shape variations across the range of interest. The double-
differential result using events with 𝛿𝛼𝑇 < 45𝑜 shown in the bottom left panel of
figure 3-15 is dominated by events that primarily occupy the region up to the Fermi
momentum and do not exhibit a high momentum tail. The corresponding ratio in-
sert illustrates a fairly uniform behavior indicative of transparency effects ranging
between 50-70%. The double-differential results using events with 135𝑜 < 𝛿𝛼𝑇 < 180𝑜
is shown in the bottom right panel of figure 3-15 and illustrate the high transverse

79
missing momentum up to 1 GeV/c. The case without FSI effects is strongly disfa-
vored and the ratio insert illustrates strong shape variations. Therefore, the high 𝛿𝛼𝑇
region is an appealing candidate for neutrino experiments to benchmark and tune the
FSI modeling in event generators.
(a) All events
GiB No FSI (102.7/13) GiB FSI (21.6/13)
G18 No FSI (53.6/13) G18 FSI (5.8/13)
50 MicroBooNE Data (Stat ⊕ Shape)
6.79e+20 POT

GeV/c Ar
dσ 10-38 cm2 Norm
40
6 FSI/No FSI
4
30 2
0 0.2 0.4 0.6
20
dδpT

10

0
0 0.2 0.4 0.6 0.8
δpT [GeV/c]
(b) δαT < 45o (c) 135o < δαT < 180o
GiB No FSI (58.8/11) GiB FSI (3.2/11) GiB No FSI (83.7/13) GiB FSI (9.3/13)
deg GeV/c Ar

deg GeV/c Ar

0.3 G18 No FSI (27.1/11) G18 FSI (9.9/11) 0.3 G18 No FSI (91.9/13) G18 FSI (12.5/13)
cm2

cm2

0.25 0.25

0.2 1 FSI/No FSI 0.2 6 FSI/No FSI

4
0.5 2
10-38

10-38

0.15 0 0.2 0.4 0.6 0.8 0.15 0 0.2 0.4 0.6

dδαTdδpT

dδαTdδpT

0.1 0.1
d 2σ

d 2σ

0.05 0.05

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

Figure 3-15: The flux-integrated single- (top) and double- in 𝛿𝛼𝑇 bins (bottom)
differential CC1p0𝜋 cross sections as a function of the transverse missing momentum
𝛿𝑝𝑇 . Inner and outer error bars show the statistical and total (statistical and shape
systematic) uncertainty at the 1𝜎, or 68%, confidence level. The gray band shows the
normalization systematic uncertainty. Colored lines show the results of theoretical
absolute cross section calculations with and without FSI based on the GENIE and
GiBUU event generators.

The single-differential results as a function of 𝛿𝛼𝑇 using all the events that satisfy
our selection are shown in top panel of figure 3-16. The result without FSI illustrates
a uniform behavior across the whole distribution and is disfavored. The addition of
FSI effects leads to a ≈ 30% asymmetry around 𝛿𝛼𝑇 = 90𝑜 due to the fact that the
proton in our selection undergoes FSI. The three FSI models used here for comparison
result in a comparable performance, also shown in terms of the ratio plot of the

80
different FSI options to the prediction without FSI. The double-differential result
using events with 𝛿𝑝𝑇 < 0.2 GeV/c shown in the bottom left panel of figure 3-16
illustrates a uniform distribution indicative of the suppressed FSI impact in that part
of the phase-space. The double-differential result using events with 𝛿𝑝𝑇 > 0.4 GeV/c
is shown in the bottom right panel of figure 3-16 and illustrates the presence of strong
FSI effects. The case without FSI effects is disfavored and the asymmetry around 90𝑜
is significantly enhanced. Therefore, the high 𝛿𝛼𝑇 region is an appealing candidate for
neutrino experiments to benchmark and tune the FSI modeling in event generators.

(a) All events, MicroBooNE Preliminary

0.12
G21 No FSI (84.4/7) G21 hA (5.9/7)
G21 hN (7.9/7) G21 G4 (5.8/7)
0.1 MicroBooNE Data (Stat ⊕ Shape)
6.79e+20 POT
deg Ar

Norm
dσ 10-38 cm2

0.08

0.06

0.04
dδαT

1.2 FSI/No FSI

1
0.8
0.02 0.6
0 50 100 150

0
0 20 40 60 80 100 120 140 160 180
δαT [deg]
(b) δpT < 0.2 GeV/c, MicroBooNE Preliminary (c) δpT > 0.4 GeV/c, MicroBooNE Preliminary
0.35 G21 No FSI (40.6/7) G21 hA (1.4/7) G21 No FSI (7.2/7) G21 hA (1.7/7)
deg GeV/c Ar

deg GeV/c Ar

G21 hN (2.1/7) G21 G4 (5.1/7) G21 hN (2.0/7) G21 G4 (1.2/7)

0.3 0.05
cm2

cm2

2 FSI/No FSI
0.25 0.04 1.5
1
0.2 0.03 0 50 100 150
10-38

10-38

0.15
0.02
dδαTdδpT

dδαTdδpT

FSI/No FSI
0.1 0.8
d2σ

d2σ

0.6
0.05 0.01
0 50 100 150

0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
δαT [deg] δαT [deg]

Figure 3-16: The flux-integrated single- (top) and double- in 𝛿𝑝𝑇 bins (bottom) dif-
ferential CC1p0𝜋 cross sections as a function of the angle 𝛿𝛼𝑇 . Inner and outer error
bars show the statistical and total (statistical and shape systematic) uncertainty at
the 1𝜎, or 68%, confidence level. The gray band shows the normalization systematic
uncertainty. Colored lines show the results of theoretical absolute cross section calcu-
lations with a number of FSI modeling options based on the GENIE event generator.

Lastly, figure 3-17 shows the single- (top) and double- in 𝛿𝑝𝑇,𝑦 bins (bottom)
differential unfolded cross sections as a function of 𝛿𝑝𝑇,𝑥 . The event distributions of

81
𝛿𝑝𝑇,𝑥 and 𝛿𝑝𝑇,𝑦 have already been presented in figure 3-11. The single differential
result (top panel) illustrates a fairly broad symmetric distribution centered around 0.
The double-differential result for events where 𝛿𝑝𝑇,𝑦 < -0.15 GeV/c (bottom left panel)
illustrates an even broader distribution where all predictions yield comparable results.
Unlike the asymmetric part of the 𝛿𝑝𝑇,𝑦 tail, the double-differential result for events
with -0.15 < 𝛿𝑝𝑇,𝑦 < 0.15 GeV/c (bottom right panel) shows a much narrower peak
which strongly depends on the choice of the underlying model and the addition or
not of nuclear effects, such as RPA ones. The G18 LFG and G18 No RPA predictions
are favored in that part of the phase-space.

(a) All events

G18 LFG (7.8/11) G18 No RPA (3.5/11)
50
G18 RFG (37.5/11) G18 EffSF (57.3/11)
MicroBooNE Data (Stat ⊕ Shape)
GeV/c Ar

6.79e+20 POT
40 Norm MicroBooNE Preliminary
dσ 10-38 cm2

σData = 0.17
30

20
dδpT,x

10

0
−0.4 −0.2 0 0.2 0.4
δpT,x [GeV/c]
(b) δpT,y < -0.15 GeV/c (c) -0.15 < δpT,y < 0.15 GeV/c
14 G18 LFG (3.9/11) G18 No RPA (3.2/11) G18 LFG (4.9/11) G18 No RPA (4.1/11)
GeV2/c2 Ar

GeV2/c2 Ar

G18 RFG (3.8/11) G18 EffSF (16.3/11) G18 RFG (39.3/11) G18 EffSF (52.3/11)
12
cm2

cm2

80
MicroBooNE Preliminary MicroBooNE Preliminary
10
σData = 0.25 60 σData = 0.14
10-38

10-38

8
6 40
T,y

T,y
dδp

dδp

4
d2σ

d2σ

20
T,x

T,x

2
dδp

dδp

0 0
−0.4 −0.2 0 0.2 0.4 −0.4 −0.2 0 0.2 0.4
δpT,x [GeV/c] δpT,x [GeV/c]

Figure 3-17: The flux-integrated single- (top) and double- in 𝛿𝑝𝑇,𝑦 bins (bottom) dif-
ferential CC1p0𝜋 cross sections as a function of the angle 𝛿𝑝𝑇,𝑥 . Inner and outer error
bars show the statistical and total (statistical and shape systematic) uncertainty at
the 1𝜎, or 68%, confidence level. The gray band shows the normalization system-
atic uncertainty. Colored lines show the results of theoretical absolute cross section
calculations with a number of event generators.

The 𝜒2 per degree of freedom (d.o.f.) data comparison for each prediction shown

82
on the results in figures 3-15, 3-16, and 3-17 takes into account the total covariance
matrix including the off-diagonal elements. Figures 3-18 - 3-26 show in the interaction
breakdown of the aforementioned results.

G18 All events G18 No FSI All events

QE MEC
50 RES DIS 50
MicroBooNE Data
GeV/c Ar

GeV/c Ar
(Stat ⊕ Shape Unc)
cm2

cm2
40 Norm Unc 40

30 30
10-38

10-38
20 20
T

T
dσ

dσ
dδp

dδp
10 10

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

GiBUU All events GiBUU No FSI All events

50 50
GeV/c Ar

GeV/c Ar
cm2

cm2
40 40

30 30
10-38

10-38
20 20
T

T
dσ

dσ
dδp

dδp
10 10

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

Figure 3-18: Cross-section interaction breakdown for all the selected events. The
breakdown is shown for (top left) the G18 configuration with FSI effects, (top right)
the G18 configuration without FSI effects, (bottom left) GiB with FSI effects, and
(bottom right) GiB without FSI effects.

G18 δαT < 45o G18 No FSI δαT < 45o

QE MEC
deg GeV/c Ar

deg GeV/c Ar

0.3 RES DIS 0.3

MicroBooNE Data
cm2

cm2

0.25 (Stat ⊕ Shape Unc) 0.25

Norm Unc
0.2 0.2
10-38

10-38

0.15 0.15
0.1 0.1
dδαTdδpT

dδαTdδpT
d2σ

d2σ

0.05 0.05
0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

GiBUU δαT < 45o GiBUU No FSI δαT < 45o

deg GeV/c Ar

deg GeV/c Ar

0.3 0.3
cm2

cm2

0.25 0.25
0.2 0.2
10-38

10-38

0.15 0.15
0.1 0.1
dδαTdδpT

dδαTdδpT
d2σ

d2σ

0.05 0.05
0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

Figure 3-19: Cross-section interaction breakdown for events with 𝛿𝛼𝑇 < 45𝑜 . The
breakdown is shown for (top left) the G18 configuration with FSI effects, (top right)
the G18 configuration without FSI effects, (bottom left) GiB with FSI effects, and
(bottom right) GiB without FSI effects.

83
 G18 135o < δαT < 180o G18 No FSI 135o < δαT < 180o
QE MEC

deg GeV/c Ar

deg GeV/c Ar
0.3 RES DIS 0.3
MicroBooNE Data

cm2

cm2
0.25 (Stat ⊕ Shape Unc) 0.25
Norm Unc
0.2 0.2

10-38

10-38
0.15 0.15
0.1 0.1

dδαTdδpT

dδαTdδpT
d2σ

d2σ
0.05 0.05
0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

GiBUU 135o < δαT < 180o GiBUU No FSI 135o < δαT < 180o
deg GeV/c Ar

deg GeV/c Ar
0.3 0.3
cm2

cm2
0.25 0.25
0.2 0.2
10-38

10-38
0.15 0.15
0.1 0.1
dδαTdδpT

dδαTdδpT
d2σ

d2σ
0.05 0.05
0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
δpT [GeV/c] δpT [GeV/c]

Figure 3-20: Cross-section interaction breakdown for events with 135𝑜 < 𝛿𝛼𝑇 < 180𝑜 .
The breakdown is shown for (top left) the G18 configuration with FSI effects, (top
right) the G18 configuration without FSI effects, (bottom left) GiB with FSI effects,
and (bottom right) GiB without FSI effects.

G21 hA All events G21 hN All events

0.12 0.12
QE MEC
RES DIS
0.1 MicroBooNE Data 0.1
(Stat ⊕ Shape Unc)
deg Ar

deg Ar
dσ 10-38 cm2

dσ 10-38 cm2

0.08 Norm Unc 0.08

0.06 0.06

0.04 0.04
dδαT

dδαT

0.02 0.02

0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
δαT [deg] δαT [deg]

G21 G4 All events G21 No FSI All events

0.12 0.12

0.1 0.1
deg Ar

deg Ar
dσ 10-38 cm2

dσ 10-38 cm2

0.08 0.08

0.06 0.06

0.04 0.04
dδαT

dδαT

0.02 0.02

0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
δαT [deg] δαT [deg]

Figure 3-21: Cross-section interaction breakdown for all the selected events. The
breakdown is shown for (top left) the G21 hA configuration with the hA2018 FSI
model, (top right) the G21 hN configuration with the hN FSI model, (bottom left)
the G21 G4 configuration with the G4 FSI model, and (bottom right) the G21 No
FSI configuration without FSI effects.

84
 G21 hA δpT < 0.2 GeV/c G21 hN δpT < 0.2 GeV/c
0.35 QE MEC 0.35

deg GeV/c Ar

deg GeV/c Ar
RES DIS
0.3 MicroBooNE Data 0.3

cm2

cm2
(Stat ⊕ Shape Unc)
0.25 Norm Unc 0.25
0.2 0.2

10-38

10-38
0.15 0.15

dδαTdδpT

dδαTdδpT
0.1 0.1

d2σ

d2σ
0.05 0.05
0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
δαT [deg] δαT [deg]

G21 G4 δpT < 0.2 GeV/c G21 No FSI δpT < 0.2 GeV/c
0.35 0.35
deg GeV/c Ar

deg GeV/c Ar
0.3 0.3
cm2

cm2
0.25 0.25
0.2 0.2
10-38

10-38
0.15 0.15
dδαTdδpT

dδαTdδpT
0.1 0.1
d2σ

d2σ
0.05 0.05
0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
δαT [deg] δαT [deg]

Figure 3-22: Cross-section interaction breakdown for events with 𝛿𝑝𝑇 < 0.2 GeV/c.
The breakdown is shown for (top left) the G21 hA configuration with the hA2018
FSI model, (top right) the G21 hN configuration with the hN FSI model, (bottom
left) the G21 G4 configuration with the G4 FSI model, and (bottom right) the G21
No FSI configuration without FSI effects.

G21 hA δpT > 0.4 GeV/c G21 hN δpT > 0.4 GeV/c
QE MEC
deg GeV/c Ar

deg GeV/c Ar

0.05 RES DIS 0.05

MicroBooNE Data
cm2

cm2

(Stat ⊕ Shape Unc)

0.04 Norm Unc 0.04

0.03 0.03
10-38

10-38

0.02 0.02
dδαTdδpT

dδαTdδpT
d2σ

d2σ

0.01 0.01

0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
δαT [deg] δαT [deg]

G21 G4 δpT > 0.4 GeV/c G21 No FSI δpT > 0.4 GeV/c
deg GeV/c Ar

deg GeV/c Ar

0.05 0.05
cm2

cm2

0.04 0.04

0.03 0.03
10-38

10-38

0.02 0.02
dδαTdδpT

dδαTdδpT
d2σ

d2σ

0.01 0.01

0 0
0 20 40 60 80 100 120 140 160 180 0 20 40 60 80 100 120 140 160 180
δαT [deg] δαT [deg]

Figure 3-23: Cross-section interaction breakdown for events with 𝛿𝑝𝑇 > 0.4 GeV/c.
The breakdown is shown for (top left) the G21 hA configuration with the hA2018
FSI model, (top right) the G21 hN configuration with the hN FSI model, (bottom
left) the G21 G4 configuration with the G4 FSI model, and (bottom right) the G21
No FSI configuration without FSI effects.

85
 G18 LFG All events G18 RFG All events
QE MEC
50 RES DIS 50
MicroBooNE Data

GeV/c Ar

GeV/c Ar
40 (Stat ⊕ Shape Unc) 40

cm2

cm2
Norm Unc
30 30

10-38

10-38
20 20

T,x

T,x
dσ

dσ
dδ p

dδ p
10 10

0 0
−0.4 −0.2 0 0.2 0.4 −0.4 −0.2 0 0.2 0.4
δp [GeV/c] δp [GeV/c]
T,x T,x

G18 EffSF All events G18 No RPA All events

50 50
GeV/c Ar

GeV/c Ar
40 40
cm2

cm2
30 30
10-38

10-38
20 20
T,x

T,x
dσ

dσ
dδ p

dδ p
10 10

0 0
−0.4 −0.2 0 0.2 0.4 −0.4 −0.2 0 0.2 0.4
δp [GeV/c] δp [GeV/c]
T,x T,x

Figure 3-24: Cross-section interaction breakdown for all the selected events. The
breakdown is shown for (top left) the G18 LFG configuration, (top right) the G18
RFG configuration, (bottom left) the G18 EffSF configuration, and (bottom right)
the G18 No RPA configuration.

G18 LFG δpT,y < -0.15 GeV/c G18 RFG δpT,y < -0.15 GeV/c
14 QE MEC 14
GeV2/c2 Ar

GeV2/c2 Ar

RES DIS
12 MicroBooNE Data 12
cm2

cm2

(Stat ⊕ Shape Unc)

10 Norm Unc 10
10-38

10-38

8 8
6 6
dδpT,x dδpT,y

dδpT,x dδpT,y

4 4
d2σ

d2σ

2 2
0 0
−0.4 −0.2 0 0.2 0.4 −0.4 −0.2 0 0.2 0.4
δp [GeV/c] δp [GeV/c]
T,x T,x

G18 EffSF δpT,y < -0.15 GeV/c G18 No RPA δpT,y < -0.15 GeV/c
14 14
GeV2/c2 Ar

GeV2/c2 Ar

12 12
cm2

cm2

10 10
10-38

10-38

8 8
6 6
dδpT,x dδpT,y

dδpT,x dδpT,y

4 4
d2σ

d2σ

2 2
0 0
−0.4 −0.2 0 0.2 0.4 −0.4 −0.2 0 0.2 0.4
δp [GeV/c] δp [GeV/c]
T,x T,x

Figure 3-25: Cross-section interaction breakdown for events with 𝛿𝑝𝑇,𝑦 < -0.15 GeV/c.
The breakdown is shown for (top left) the G18 LFG configuration, (top right) the G18
RFG configuration, (bottom left) the G18 EffSF configuration, and (bottom right)
the G18 No RPA configuration.

86
 G18 LFG -0.15 < δpT,y < 0.15 GeV/c G18 RFG -0.15 < δpT,y < 0.15 GeV/c
QE MEC

GeV2/c2 Ar

GeV2/c2 Ar
RES DIS
MicroBooNE Data

cm2

cm2
80 80
(Stat ⊕ Shape Unc)
Norm Unc
60 60

10-38

10-38
40 40

dδpT,x dδpT,y

T,y
dδp
d2σ

d 2σ
20 20

T,x
dδp
0 0
−0.4 −0.2 0 0.2 0.4 −0.4 −0.2 0 0.2 0.4
δp [GeV/c] δpT,x [GeV/c]
T,x

G18 EffSF -0.15 < δpT,y < 0.15 GeV/c G18 No RPA -0.15 < δpT,y < 0.15 GeV/c
GeV2/c2 Ar

GeV2/c2 Ar
cm2

cm2
80 80

60 60
10-38

10-38
40 40
T,y

T,y
dδp

dδp
d 2σ

d 2σ
20 20
T,x

T,x
dδp

dδp
0 0
−0.4 −0.2 0 0.2 0.4 −0.4 −0.2 0 0.2 0.4
δpT,x [GeV/c] δpT,x [GeV/c]

Figure 3-26: Cross-section interaction breakdown for events with -0.15 < 𝛿𝑝𝑇,𝑦 <
0.15 GeV/c. The breakdown is shown for (top left) the G18 LFG configuration, (top
right) the G18 RFG configuration, (bottom left) the G18 EffSF configuration, and
(bottom right) the G18 No RPA configuration.

3.2.5 Kinematic Imbalance Cross-Section Analysis Conclusions

The first measurement of 𝜈𝜇 CC1p0𝜋 single and double differential cross sections on
argon as a function of kinematic imbalance variables for event topologies with a single
muon and a single proton detected in the final state was reported. The unfolded data
results were compared to a number of event generators, available model configurations
and FSI modeling options. This measurement identified regions of the phase-space
which are ideal to provide constraints for nuclear and final state interaction effects in
generator predictions essential for the extraction of oscillation parameters.

3.3 Prospects With Future Neutrino Experiments

The MicroBooNE experiment is a crucial step in the understanding of the underlying
neutrino-argon interactions that will be used to drastically reduce the uncertainties of
forthcoming high precision neutrino oscillation measurements. Furthermore, Micro-
BooNE is the first LArTPC in a neutrino beam with automated event reconstruction
and selection.

87
 This work paved the path towards precision 𝜈𝜇 CC cross section measurements
with a single proton and no pions in the final state. However, only three out of
the five available run periods are used in the results presented in this thesis, as
shown in figure 3-27. Within the next year, runs 4 and 5 will also become available.
Therefore, the statistical uncertainties will be further reduced, and the path to further
multidimensional analyses will be further explored.

Figure 3-27: MicroBooNE total Protons on Target (POT) collected with the Booster
Neutrino Beam (BNB) during the five run periods. In this thesis, the first three run
periods were used.

Beyond MicroBooNE, experiments of the Short-Baseline Neutrino (SBN) Pro-

gram at Fermilab, namely ICARUS and SBND, have already turned (or will very
soon turn) on. The progress of SBN, which shares the same technology and beam
as MicroBooNE, will largely benefit from the LArTPC expertise developed by the
MicroBooNE collaboration. With these experiments, the largest neutrino-argon scat-
tering data sets will be collected and will be used to test the performance of our
theory predictions in multiple variables simultaneously.
Despite the abundance of neutrino-argon cross sections that will be reported at the
SBN program, the relevant energy range is significantly lower than the DUNE energy
spectrum. However, there is a wealth of cross-section opportunities that will become
available once the ArgonCube demonstrator, a 2×2 grid engineering prototype for
the LArTPC DUNE Near Detector (ND) module design, is commissioned in Fall
2022. ArgonCube will provide sufficiently high statistics to measure cross sections

88
and to evaluate the corresponding systematic uncertainties using both multi-particle
channels and multidimensional analyses and will pave the path towards the final
design of the DUNE ND.
All these neutrino experimental efforts will be complemented by the continuous
benchmarking of the neutrino event generators predictions against external data sets,
such as against electron scattering data sets. A major step in this direction is made
in chapter 4 with the unification of the event generation process and of the modeling
across the two particle species. Furthermore, chapters 5 and 6 detail the analysis of
electron scattering data sets from the CLAS detector at Thomas Jefferson Laboratory
following neutrino data analysis methods and the testing against the performance of
commonly used GENIE event generator.

89
90
Chapter 4
Inclusive Electron Scattering And
The GENIE Event Generator
[Phys. Rev. D 103, 113003 (2021)]

4.1 Electron-Nucleus Modeling Development

As already discussed in section 1.2, the extraction of neutrino mixing parameters
from neutrino oscillation experiments [23, 146, 147] relies on comparing the energy-
dependent neutrino event distribution for a particular neutrino flavor near the neu-
trino production point with the corresponding event distribution at a significant dis-
tance away. In practice, the yield at each neutrino energy is extracted from the
measured neutrino-nucleus interactions in a detector, as reconstructed from the mea-
sured particles ejected in the neutrino-nucleus interaction. This requires detailed
knowledge of the neutrino-nucleus interactions.
Unfortunately, measuring the neutrino-nucleus interaction is difficult due to the
wide energy spread of accelerator-produced neutrino beams [148], as can be seen in
figure 4-1, and the tiny neutrino-nucleus cross section. A relatively small body of data
has been published [24], which suffers from poor statistics and is flux-averaged over
a wide range of neutrino energies. This data is then supplemented with theoretical
models and implemented into event generator codes such as GENIE [33], which is
extensively used across US-based neutrino experiments, to simulate the neutrino-
nucleus interactions across a wide range of energies and targets. GENIE simulations
are then used to aid in extraction of the incident neutrino flux as a function of

91
 energy from the neutrino-nucleus scattering events measured in neutrino detectors.
However, the theoretical models need to describe many different interaction processes
for medium to heavy nuclei (typically C, O, or Ar) where nuclear effects complicate the
interactions. As a result, the uncertainties in the extraction of oscillation parameters
are often dominated by the lack of knowledge of the neutrino-nucleus interactions [23,
147].

DIS (32%)
60 DIS (28%)
RES (37%) RES (36%)
30 MEC (7%) MEC (8%)
dσ [Arb Units]

QE (22%) QE (26%)
40
20
dEν

10 20

0 0
1 2 3 4 5 1 2 3 4 5
Eν [GeV]
Figure 4-1: Charged-current cross sections as a function of neutrino energy obtained
using GENIE for muon neutrino scattering using the DUNE near detector (left) and
far detector (right) oscillated fluxes. The shaded bands show the fractional contri-
bution for each interaction mechanism, quasielastic scattering (QE), meson-exchange
currents (MEC), resonance excitation (RES), and deep inelastic scattering (DIS). See
text for details of the interaction mechanisms. The numbers in parentheses indicate
the percentage of the cross section due to each interaction mechanism.

Figure 4-1 shows such a wide energy spectrum for the DUNE near detector flux-
averaged cross sections (left) and the far detector oscillated flux-averaged cross sec-
tions (right) using one model configuration in GENIE. All four neutrino-nucleus re-
action mechanisms contribute significantly and all four need to be well understood.
This is especially true because different reaction mechanisms contribute differently in
the different oscillation peaks. Understanding one reaction mechanism better than
the others could have significant implications for oscillation analyses.
To improve our understanding of neutrino-nucleus interactions, we can take ad-
vantage of the fact that neutrinos and electrons are both leptons. Thus, they interact

92
with atomic nuclei in similar ways via the same reaction mechanisms, as illustrated
in figure 4-2 and detailed in sections 1.4 and 1.5.

Figure 4-2: (Left) electron-nucleus inclusive scattering via one-photon exchange and
(right) charged current neutrino-nucleus inclusive scattering via 𝑊 exchange with a
final state charged lepton.

The most common lepton-nucleus interaction mechanisms are shown in figure 4-

3 include: (a) quasielastic (QE) scattering from individual moving nucleons in the
nucleus; (b) two-nucleon knockout, due to interactions with a meson being exchanged
between two nucleons referred to two-particle two-hole excitations, 2p2h or its major
component, meson exchange currents (MEC); (c) interactions which leave the struck
nucleon in an excited resonance state (RES); and (d) nonresonant interactions with
a quark within the nucleon (DIS).

Figure 4-3: Reaction mechanisms for lepton-nucleus scattering (a) quasielastic scat-
tering (QE) where one nucleon is knocked out of the nucleus, (b) 2p2h where two
nucleons are knocked out of the nucleus, (c) RES resonance production where a nu-
cleon is excited to a resonance which decays to a nucleon plus meson(s), and (d) DIS
where the lepton interacts with a quark in the nucleon.

All these interaction processes have to be described in detailed in event generators

like GENIE. GENIE started as an event generator that could exclusively handle

93
neutrino interactions. In recognition of the importance of electron scattering, the
latter was added as a new option in close conjunction with the neutrino scattering
section. As much as possible, the neutrino cross section references vector and axial
contributions separately and uses the same modeling for vector interactions as the
electron section. Some models were developed separately for electrons and others
were developed for both applications in tandem.

An earlier electron version of GENIE (v2.12.10) had already been tested by com-
paring with inclusive (𝑒, 𝑒′ ) data [149]. Although the QE peak was well-described for
a variety of energies and nuclei, the RES region was poorly described. However, the
establishment of full compatibility between the electron and neutrino versions was
then still in its early stages.

With this work, we significantly improved both neutrino and electron versions of
GENIE to address these and other issues. We fixed significant errors in the previous
version, including an error in the Mott cross section in the electron QE Rosenbluth
interaction, a missing Lorentz boost in the MEC interaction affecting both electron
and neutrino interactions, and incorrect electron couplings used in the RES interac-
tions. We worked to better integrate the electron and neutrino codes for QE and
MEC models. We also added more up-to-date models such as SuSAv2 [150]. These
changes have been incorporated in the latest GENIE version (v3.2.0). We refer to
the electron-scattering component of the widely-used GENIE [33] event generator as
eGENIE.

The GENIE improvements can be seen in figure 4-4 [151]. The QE peak (at
𝜔 ≈ 0.15 GeV/c) predicted by the older GENIE v2 is too large and is slightly shifted
to higher energy transfer than the data, while the first simulated resonance peak
is at a larger energy transfer than the one observed in the data. The QE peak
predicted by the updated GENIE v3 has about the correct integral and is at the
correct energy transfer (but is slightly too narrow) and the first resonance peak is
located at 𝑚Δ − 𝑚 ≈ 300 MeV beyond the QE peak, as expected. Details of the
calculations and of the discrepancies between GENIE v3 and the data are discussed
in detail below.

94
 0.56 GeV, θ = 60
o

d2σ [µb/sr/GeV]
6 v2
4
v3
dΩ dE
2

0
0.1 0.2 0.3 0.4
Energy Transfer [GeV]
Figure 4-4: Comparison between GENIE v2 and v3 descriptions of inclusive C(𝑒, 𝑒′ )
scattering cross sections at 𝐸0 = 0.56 GeV, 𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.24 GeV2 . Black
points show the data, solid black line shows the GENIE v3 results and dashed black
line shows the GENIE v2 results.

In this analysis, we specifically focused on testing our knowledge of the electron-

nucleus cross section by benchmarking eGENIE against existing inclusive electron
scattering data for different target nuclei, beam energies and scattering angles. The
goals are very similar to the ones in reference [149], but we test a much more modern
version of eGENIE and we also compare different models within eGENIE. In addition,
if eGENIE describes electron-nucleus scattering well, then it would be an improvement
on the former empirical fit [152] and would be valuable for helping simulate a variety
of electron experiments.

For fixed incident beam energy and scattered electron angle, the dominant process
changes from QE at low energy transfer (𝜔 ≈ 𝑄2 /2𝑚) through MEC to RES and to
DIS at high energy transfer. Therefore, examining the agreement of eGENIE with
data as a function of energy transfer can provide valuable insight into the specific
shortcomings of the eGENIE models and their implementations. This separation
according to the underlying physics interactions gives valuable insights which are not
presently possible with neutrino cross sections, because only broad-energy beams are
available.

The GENIE simulation framework offers several models of the nuclear ground
state, multiple models for each of the electron- or neutrino-nucleus scattering mecha-

95
nisms accompanied by various tunable model parameters, and a number of models for
hadronic final state interactions (FSI), i.e., intranuclear rescattering of the outgoing
hadrons [33, 104, 153]. We describe the different models relevant for this work and
the electron-specific effects that we accounted for during the eGENIE development
below [42].

Since our goal is to use electron scattering data to validate neutrino interaction
modeling in GENIE, the GENIE code for electron and neutrino interactions are unified
in many places. The neutrino interacts with a nucleus via the weak interaction and
massive 𝑊 or 𝑍 exchange, whereas the electron interacts mostly electromagnetically
via massless photon exchange, as shown in figure 4-2. This causes the cross sections
to differ by an overall factor of

8𝜋 2 𝛼2 1
(4.1)
𝐺2𝐹 𝑄4

when equations 1.18 and 1.19 are compared. In the code, both interactions use
the same nuclear ground state and many of the nuclear reaction effects, such as FSI,
are very similar or identical. Except for mass effects and form factors, the electron
nucleus cross section can be obtained by setting the axial part of the interaction to
zero. We also accounted for isoscalar and isovector terms appropriately.

Many of the models reported in this work, except for SuSAv2, use the GENIE
implementation of the Local Fermi gas (LFG) model to describe the nuclear ground
state. In the simplest Fermi gas model, nucleons occupy all momentum states up to
the global Fermi momentum 𝑘𝐹 with equal probability. In the LFG model, the Fermi
momentum at a given radial position depends on the local nuclear density obtained
from measurements of nuclear charge densities. To account for this radial depen-
dence, GENIE selects an initial momentum for the struck nucleon by first sampling
an interaction location 𝑟 inside the nucleus according to the nuclear density. The
nucleon momentum is then drawn from a Fermi distribution using the local Fermi
momentum 𝑘𝐹 (𝑟).

Another commonly used nuclear model is the Relativistic Fermi Gas (RFG). Here

96
a global momentum distribution is used for the entire nucleus, independent of the
interaction location in the nucleus. However, a high-momentum tail of nucleons with
momenta above the Fermi-momentum is included. This tail is meant to approximately
account for the effects of two-nucleon short-range correlations [154, 155] and follows
a 1/𝑘 4 distribution, where 𝑘 is the nucleon momentum.
We consider two distinct sets of eGENIE configurations:

• G2018, which uses the LFG nuclear model, the Rosenbluth cross section for
QE scattering, and the empirical MEC model [109]. This model set is formally
marked as the G18_10a_02_11a configuration of GENIE v3.

• GSuSAv2 [28], which follows the universal SuSAv2 super-scaling approach

to lepton scattering. This new model set is included in the latest GENIE
v3.2.0 release as the GEM21_11b_00_000 configuration for electron scattering
and G21_11b_00_000 for neutrino interactions.

In both model sets, RES is modeled using the Berger-Sehgal model [125] and DIS
reactions are modeled using Bodek and Yang [156]. The models are described in more
detail below.
In QE interactions, a lepton scatters on a single nucleon, removing it from the
spectator 𝐴 − 1 nucleus unless final-state interactions lead to reabsorption. The
electron QE interaction in the G2018 configuration of GENIE uses the Rosenbluth
cross section with the vector structure function parametrization of reference [157].
We corrected the implementation of this model for eGENIE and modified the cross
section to account for the identified issues. This electron QE cross section differs in
important ways from the Valencia CCQE model [158] used in the G2018 configuration
for neutrinos. Most notably, the Rosenbluth treatment lacks medium polarization
corrections.
A new QE model in GENIE, based on the SuSAv2 approach [28, 159, 160], uses
superscaling to write the inclusive cross section in terms of a universal function inde-
pendent of momentum transfer and nucleus. For EM scattering, the scaling function
may be expressed in the form

97
 𝑑2 𝜎
′
𝑓 (𝜓 ) = 𝑘𝐹 𝑑Ω𝑒 𝑑𝜈
′ ′ , (4.2)
𝜎𝑀 𝑜𝑡𝑡 (𝑣𝐿 𝐺𝑒𝑒
𝐿 + 𝑉𝑇 𝐺𝑒𝑒
𝑇 )

where 𝜓 ′ is a dimensionless scaling variable, 𝑘𝐹 is the nuclear Fermi momentum,

the denominator is the single-nucleon elastic cross section, 𝑣𝐿 and 𝑣𝑇 are known
′ ′
functions of kinematic variables, and 𝐺𝑒𝑒
𝐿 (𝑞, 𝜔) and 𝐺𝑇 (𝑞, 𝜔) are the longitudinal
𝑒𝑒

and transverse nucleon structure functions linearly related to 𝐹1𝑒 and 𝐹2𝑒 [161]. For
eGENIE, we extended the original neutrino implementation [28] to the electron case.

The original SuSAv2 QE cross section calculations used a Relativistic Mean Field
(RMF) model of the nuclear ground state [162, 163]. This approach includes the
effects of the real part of the nucleon-nucleus potential on the outgoing nucleons
which creates a “distorted” nucleon momentum distribution.

Although GENIE lacks the option to use an RMF nuclear model directly, we
achieve approximate consistency with the RMF-based results by using a two-step
strategy for QE event generation. First, an energy and scattering angle for the out-
going lepton are sampled according to the inclusive double-differential cross section.
This cross section is computed by interpolating precomputed values of the nuclear
′ ′
responses 𝐺𝑒𝑒
𝐿 (𝑞, 𝜔) and 𝐺𝐿 (𝑞, 𝜔) which are tabulated on a two-dimensional grid in
𝑒𝑒

(𝑞, 𝜔) space. The responses were obtained using the original RMF-based SuSAv2
calculation. Second, the outgoing nucleon kinematics are determined by choosing its
initial momentum from an LFG distribution. The default nucleon binding energy used
in GENIE for the LFG model is replaced for SuSAv2 with an effective value tuned to
most closely duplicate the RMF distribution. The outgoing nucleon kinematics are
not needed for the comparisons to inclusive (𝑒, 𝑒′ ) data shown in this work.

MEC describes an interaction that results in the ejection of two nucleons from
the nucleus, thus is often referred to as 2p2h. It typically proceeds via lepton inter-
action with a pion being exchanged between two nucleons or by interaction with a
nucleon in an Short Range Correlated (SRC) pair. MEC is far less understood than
other reaction mechanisms because, unlike the others, it involves scattering from two

98
nucleons simultaneously. GENIE has several models for MEC.
The G2018 configuration of eGENIE uses the empirical MEC model [109] that
is useable for both electron- and neutrino-nucleus scattering. It assumes that the
MEC peak for inclusive scattering has a Gaussian distribution in 𝑊 and is located
between the QE and first RES peaks. Although both versions of the model use the
same effective form factors, the amplitude of the MEC peak was tuned separately to
electron and neutrino scattering data. This model was developed in the context of
empirically fitting GENIE to MiniBooNE inclusive neutrino scattering data and is still
used for neutral-current interactions [109]. For charged-current neutrino interactions,
𝜈GENIE G2018 uses the very different Valencia 2p2h model [158, 164] instead of the
empirical model.
For the description of the 2p2h MEC contributions, the SuSAv2 model uses the
fully relativistic calculations from [165]. This treatment allows for a proper separation
of neutron-proton and proton-proton pairs in the final state via the analysis of the
direct-exchange interference terms [166]. This approach is capable of reproducing the
nuclear dynamics and superscaling properties observed in inclusive electron-nucleus
scattering reactions [167–169]. The latter serves as a robust test for nuclear models.
It further provides an accurate description of existing neutrino data [169–173]. As in
the case for the SuSAv2 QE model, we extended the original GENIE implementation
of SuSAv2 MEC for neutrinos to the electron case for eGENIE [28, 159, 174, 175].
The SuSAv2 MEC approach is the only fully relativistic model that can be ex-
tended without approximations to the full-energy range of interest for neutrino scat-
tering events. Therefore, it is a very promising modeling choice for present and future
neutrino experiments for one of the least understood interaction channels.
RES production in GENIE is simulated using the Berger-Sehgal model [125], in
which the lepton interacts with a single moving nucleon and excites it to one of
16 resonances. The cross sections are calculated based on the Feynman-Kislinger-
Ravndal (FKR) model [176], without any interferences between them. Form factors
are derived separately for vector and axial probes [110] but have not been updated
to include recent electron scattering results.

99
 The GENIE treatment of DIS used in this work is based on that of Bodek and
Yang [156]. Hadronization is modeled using an approach which transitions gradually
as a function of the hadronic invariant mass 𝑊 between the AGKY model [177] and
the PYTHIA 6 model [178]. At low 𝑊 values, the Bodek-Yang differential cross
section is scaled by tunable parameters that depend on the multiplicity of hadrons in
the final-state [104].

Integration of the RES and DIS contributions is complicated by the need for a
model of nonresonant meson production. There is no definite separation of RES and
DIS contributions. GENIE makes a sharp cutoff at 𝑊 = 1.93 GeV in the latest tune
and uses a suppression factor to enable usage of the Bodek-Yang cross section at low
𝑊 in place of a true nonresonant model. These features were recently retuned by the
GENIE collaboration using measurements of charged-current 𝜈𝜇 and 𝜈¯𝜇 scattering on
deuterium [179]. The W cutoff and suppression factors apply to both electron- and
neutrino-nucleus models.

Final state interactions of outgoing hadrons with the residual nuclei are calcu-
lated in eGENIE using the INTRANUKE [153, 180] package and one of two options.
The first, hA, an empirical data-driven method, uses the cross-section of pions and
nucleons with nuclei as a function of energy up to 1.2 GeV and the CEM03 [181]
calculation for higher energies. The second, hN, is a full intra-nuclear cascade cal-
culation of the interactions of pions, kaons, photons, and nucleons with nuclei. In
the hN model, each outgoing particle can interact successively with any or all the
nucleons it encounters on its path leaving the nucleus, and any particles created in
those interactions can also subsequently reinteract. The ability of the two models to
describe hadron-nucleus data is very similar. The eGENIE G2018 configuration uses
the hA FSI model, while GSuSAv2 uses hN. However, the choice of FSI model has no
effect on the inclusive cross sections considered in the present work.

100
4.2 Inclusive Electron Scattering Data Comparisons

To test eGENIE, we compared inclusive electron scattering data to theoretical pre-

dictions made using two different program configurations which differ in their choice
of QE and MEC models. Namely, G2018 adopts the Rosenbluth model for QE in-
teractions and the empirical Dytman model for MEC events, while GSuSAv2 uses
SuSAv2 for both QE and MEC interactions.

Figures 4-5, 4-6 and 4-7 show the inclusive C(𝑒, 𝑒′ ) cross sections for a wide range
of beam energies and scattering angles compared to the G2018 and GSuSAv2 mod-
els [151, 182–189].

The QE peak is the one at lowest energy transfer (𝜈 ≈ 𝑄2 /2𝑚) in each plot.
The next peak at about 300 MeV larger energy transfer corresponds to the ∆(1232)
excitation and the “dip-region” is between the two peaks. The ∆ peak in the data
is separated from the QE peak by less than the 300 MeV ∆-nucleon mass difference,
indicating that it is shifted in the nuclear medium. This shift is more visible at lower
momentum transfer where the ∆ peak is more prominent.

GSuSAv2 clearly describes the QE and dip-regions much better than G2018, es-
pecially at the three lowest momentum transfers, as shown in figure 4-5. G2018 has
particular difficulty describing the data for 𝐸0 = 0.24 GeV and 𝜃𝑒 = 60∘ , where
𝑄2 = 0.05 GeV2 at the QE peak. G2018 also predicts too small a width for the QE
peak and too small a MEC contribution for 𝐸0 = 0.56 GeV and 𝜃𝑒 = 60∘ . GSuSAv2
describes both features far better.

At intermediate momentum transfers shown in figure 4-6, GSuSAv2 describes

the data somewhat better than G2018, although it overpredicts the dip-region cross
section at 𝐸0 = 1.299 GeV and 𝜃𝑒 = 37.5∘ . The MEC contribution for G2018 appears
to be much too small for 𝐸0 = 2.222 GeV and 𝜃𝑒 = 15.54∘ (𝑄2𝑄𝐸 = 0.33 GeV2 ).
Both model sets significantly disagree with the data in the resonance region, where
they use the same RES and DIS models. The 0.961 GeV, 37.5∘ and the 2.222 GeV,
15.54∘ data are taken at almost identical 𝑄2𝑄𝐸 . The lower beam-energy data is more
transverse, since it is at larger scattering angle. The GSuSAv2 MEC contribution is

101
 ×103 GSuSav2 ×103 G2018
C, 0.24 GeV, θ = 60
12 o

50 50

0 3 0 3
×100.02 0.04 0.06 0.08 0.1 ×100.02 0.04 0.06 0.08 0.1
C, 0.56 GeV, θ = 36o
12
d2σ [µb/sr/GeV]

50 50
dΩ dE

0 0
0 0.1 0.2 0.3 0.40 0.1 0.2 0.3 0.4
C, 0.56 GeV, θ = 60o
12
6 6

4 4

2 2

0 0
0 0.1 0.2 0.3 0.4 0 0.1 0.2 0.3 0.4
Energy Transfer [GeV] Energy Transfer [GeV]
Figure 4-5: Comparison of inclusive C(𝑒, 𝑒′ ) scattering cross sections for data and for
GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top) 𝐸0 = 0.24 GeV,
𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.05 GeV2 (middle) 𝐸0 = 0.56 GeV, 𝜃𝑒 = 36∘ and 𝑄2𝑄𝐸 ≈ 0.11
GeV2 , and (bottom) 𝐸0 = 0.56 GeV, 𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.24 GeV2 . Black points
show the data, solid black lines show the total GENIE prediction, colored lines show
the contribution of the different reaction mechanisms: (blue) QE, (red) MEC, (green)
RES and (orange) DIS.

similar for both data sets, but the G2018 MEC contribution is far smaller for the
higher beam-energy data. The GSuSAv2 MEC contribution describes the dip-region
better in the higher beam-energy data set. The RES model appears to agree with the

102
 ×103 GSuSav2 ×103 G2018
C, 0.961 GeV, θ = 37.5
12 o

5 5

0 0
×103 0.2 0.4 0.6 ×103 0.2 0.4 0.6
C, 1.299 GeV, θ = 37.5o
12
d2σ [µb/sr/GeV]

2 2

1 1
dΩ dE

0 0
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8
C, 2.222 GeV, θ = 15.54o
12

40 40

20 20

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Energy Transfer [GeV] Energy Transfer [GeV]
Figure 4-6: Comparison of inclusive C(𝑒, 𝑒′ ) scattering cross sections for data and
for GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top) 𝐸0 = 0.96
GeV, 𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.32 GeV2 , (middle) 𝐸0 = 1.30 GeV, 𝜃𝑒 = 37.5∘ and
𝑄2𝑄𝐸 ≈ 0.54 GeV2 , and (bottom) 𝐸0 = 2.22 GeV, 𝜃𝑒 = 15.5∘ and 𝑄2𝑄𝐸 ≈ 0.33 GeV2 .
Black points show the data, solid black lines show the total GENIE prediction, colored
lines show the contribution of the different reaction mechanisms: (blue) QE, (red)
MEC, (green) RES and (orange) DIS.

data slightly better for the lower beam-energy, more transverse, data set.

At the highest momentum transfers (𝑄2 ≈ 1 GeV2 ) shown in figure 4-7, the dis-
agreement at the larger energy transfers is far greater. The empirical G2018 MEC

103
 ×103 GSuSav2 ×103 G2018
C, 1.501 GeV, θ = 37.5
12 o

1.5 1.5
1 1
0.5 0.5
0 0
0.4 0.6 0.8 0.2 0.4 0.6 0.8
C, 3.595 GeV, θ = 16o
12
d2σ [µb/sr/GeV]

4 4

2 2
dΩ dE

0 0
0.5 1 1.5 0.5 1 1.5
C, 3.595 GeV, θ = 20o
12

0.6 0.6
0.4 0.4
0.2 0.2
0 0
0.4 0.5 0.6 0.7 0.8 0.4 0.5 0.6 0.7 0.8
Energy Transfer [GeV] Energy Transfer [GeV]
Figure 4-7: Comparison of inclusive C(𝑒, 𝑒′ ) scattering cross sections for data and for
GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top) 𝐸0 = 1.501 GeV,
𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.92 GeV2 , (middle) 𝐸0 = 3.595 GeV, 𝜃𝑒 = 16∘ and 𝑄2𝑄𝐸 ≈ 1.04
GeV2 , and (bottom) 𝐸0 = 3.595 GeV, 𝜃𝑒 = 20∘ and 𝑄2𝑄𝐸 ≈ 1.3 GeV2 . Black points
show the data, solid black lines show the total GENIE prediction, colored lines show
the contribution of the different reaction mechanisms: (blue) QE, (red) MEC, (green)
RES and (orange) DIS.

model contributions are negligible, in marked contrast to the GSuSAv2 MEC con-
tributions. The RES and DIS contributions are very significant at high 𝑄2 and in
general the GENIE model is larger than the data in the region dominated by RES

104
 ×103 GSuSav2 ×103 G2018
Fe, 0.56 GeV, θ = 60
56 o

20 20

10 10

0 0
×1030.1 0.2 0.3 0.4
×1030.1 0.2 0.3 0.4
40 56Fe, 0.961 GeV, θ = 37.5o 40
d2σ [µb/sr/GeV]

20 20
dΩ dE

0 0
0.2 0.4 0.6 0.2 0.4 0.6
Fe, 1.299 GeV, θ = 37.5o
56

10 10

5 5

0 0
0.2 0.4 0.6 0.8 0.2 0.4 0.6 0.8
Energy Transfer [GeV] Energy Transfer [GeV]
Figure 4-8: Comparison of inclusive Fe(𝑒, 𝑒′ ) scattering cross sections for data and for
GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top) Fe(𝑒, 𝑒′ ), 𝐸0 = 0.56
GeV, 𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.24 GeV2 , (middle) Fe(𝑒, 𝑒′ ), 𝐸0 = 0.96 GeV, 𝜃𝑒 = 37.5∘
and 𝑄2𝑄𝐸 ≈ 0.32 GeV2 , (bottom) Fe(𝑒, 𝑒′ ), 𝐸0 = 1.30 GeV, 𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.54
GeV2 . Black points show the data, solid black lines show the total GENIE prediction,
colored lines show the contribution of the different reaction mechanisms: (blue) QE,
(red) MEC, (green) RES and (orange) DIS.

interactions [149]. In addition, GENIE does not include the nuclear medium depen-
dent ∆-peak shift, so that the predicted location of the ∆-peak is at larger energy
transfer than that of the data.

105
 GSuSav2 G2018
d2σ [µb/sr/GeV] Ar, 2.222 GeV, θ = 15.54o
40

100 100

50 50
dΩ dE

0 0
0 0.2 0.4 0.6 0.8 0 0.2 0.4 0.6 0.8
Energy Transfer [GeV] Energy Transfer [GeV]
Figure 4-9: Comparison of inclusive Ar(𝑒, 𝑒′ ) scattering cross sections for data and for
GENIE at 𝐸0 = 2.22 GeV, 𝜃𝑒 = 15.5∘ and 𝑄2𝑄𝐸 ≈ 0.33 GeV2 . (left) data vs GSuSAv2
and (right) data vs G2018. Black points show the data, solid black lines show the
total GENIE prediction, colored lines show the contribution of the different reaction
mechanisms: (blue) QE, (red) MEC, (green) RES and (orange) DIS.

Figure 4-8 shows the inclusive Fe(𝑒, 𝑒′ ) cross sections for several beam energies and
scattering angles compared to the G2018 and GSuSAv2 models. The GSuSAv2 model
describes the QE region better for all three data sets. As described previously, the
GSuSAv2 MEC model is independently calculated. The empirical G2018 MEC model
was fit using GENIE v2 QE and RES models. The fit will have to be redone once the
QE and RES models stabilize. The GSuSAv2 MEC contributions are significantly
larger than the empirical G2018 MEC contributions and match the dip-region data
far better at 𝑄2𝑄𝐸 = 0.24 and 0.32 GeV2 . However, it overpredicts the dip-region
cross section at 𝑄2𝑄𝐸 = 0.54 GeV2 . The RES and DIS models describe the Fe data
better than the C data at large energy transfers.
Figure 4-9 shows the inclusive Ar(𝑒, 𝑒′ ) cross sections for 𝐸0 = 2.222 GeV and
𝜃𝑒 = 15.54∘ [185] compared to the G2018 and GSuSAv2 models. The GSuSAv2 model
reproduces the data very well in the QE-peak region and the G2018 reproduces the
data moderately well. The GSuSAv2 MEC model describes the dip-region much
better than the G2018 model. Again, there is significant disagreement with the RES
and DIS models at larger energy transfers.
The quality of the agreement between data and GENIE depends more on the
beam energy and angle than on the target mass from C to Fe. There is a possible

106
 Proton Deuterium
1000 2.445 GeV, θ = 20 o
1000

500 500

0 0
0.5 1 1.5 0.5 1 1.5
3.245 GeV, θ = 26.98o
d2σ [nb/sr/GeV]

50 50
dΩ dE

0 0
1 1.5 2 1 1.5 2
5.5 GeV, θ = 41o
1 1

0.5 0.5

0 0
3.4 3.6 3.8 4 3.4 3.6 3.8 4
Energy Transfer [GeV] Energy Transfer [GeV]
Figure 4-10: Comparison of inclusive proton (left) and deuterium (right) (𝑒, 𝑒′ ) scat-
tering cross sections for data and for GENIE using G2018. (top) 𝐸0 = 2.445 GeV
and 𝜃𝑒 = 20∘ , (middle) 𝐸0 = 3.245 GeV and 𝜃𝑒 = 26.98∘ , and (bottom) 𝐸0 = 5.5 GeV
and 𝜃𝑒 = 41∘ . Black points show the data, solid black lines show the total GENIE
prediction, colored lines show the contribution of the different reaction mechanisms:
(green) RES and (orange) DIS. The first peak at lowest energy transfer is the ∆(1232)
resonance.

momentum-transfer dependent shift in the location of the SuSAv2 QE peak in Fe due

to the extrapolation via scaling from C to Fe. The GSuSAv2 QE model generally
describes the data as well as or better than the G2018 model.

107
 ×103 Proton ×103 Deuterium
4.499 GeV, θ = 4 o

200 200

100 100

0 0
×103 0.5 1 1.5 ×103 0.5 1 1.5
6.999 GeV, θ = 4o
d2σ [µb/sr/GeV]

100 100

50 50
dΩ dE

0 0
0.5 1 1.5 0.5 1 1.5
9.993 GeV, θ = 4o
40 40

20 20

0 0
0.5 1 1.5 0.5 1 1.5
Energy Transfer [GeV] Energy Transfer [GeV]
Figure 4-11: Comparison of inclusive proton (left) and deuterium (right) (𝑒, 𝑒′ ) scat-
tering cross sections for data and for GENIE using G2018. (top) 𝐸0 = 4.499 GeV
and 𝜃𝑒 = 4∘ , (middle) 𝐸0 = 6.699 GeV and 𝜃𝑒 = 4∘ , and (bottom) 𝐸0 = 9.993 GeV
and 𝜃𝑒 = 4∘ . Black points show the data, solid black lines show the total GENIE
prediction, colored lines show the contribution of the different reaction mechanisms:
(green) RES and (orange) DIS. The first peak at lowest energy transfer is the ∆(1232)
resonance.

The GSuSAv2 MEC model appears to be significantly superior to the empirical

MEC model, especially at 𝑄2 < 0.5 GeV2 or at smaller scattering angles. The empir-
ical MEC contribution is often much smaller than needed to explain the dip-region

108
cross section. However, as an empirical model, it can be tuned to better describe the
data.
eGENIE dramatically overpredicts the large-energy transfer data at higher mo-
mentum transfers (𝑄2 > 0.5 GeV2 ), indicating issues with the RES (Berger-Sehgal)
and DIS (Bodek and Yang) models used.
This discrepancy at larger momentum and energy transfers is due to the elemen-
tary electron-nucleon cross section in the RES and DIS regions, rather than to the
nuclear models, since eGENIE also significantly overpredicts the proton and deuteron
cross sections. That is the case especially above the ∆ peak, as shown in figures 4-10
and 4-11. The discrepancy becomes even more pronounced due to the double counting
of processes common across the two interaction channels. This shows that tuning the
RES and DIS models to neutrino data [179] is not sufficient to constrain the vector
part of the cross section.

4.3 Implications For Neutrinos

Electron-scattering data can be a very effective tool for testing neutrino event gener-
ators due to the similarity between the interactions. Figure 4-12 shows the remark-
ably similar cross-section shapes for electron-nucleus and neutrino-nucleus scattering
for semi-exclusive 1.16 GeV lepton-carbon scattering with exactly one proton with
𝑄2 ≥ 0.1 GeV2 and 𝑃𝑝 ≥ 300 MeV/c, no charged pions with 𝑃𝜋 ≥ 70 MeV/c and
no neutral pions or photons of any momenta. This corresponds approximately to
the JLab CLAS detector thresholds. When comparing electron and neutrino distri-
butions, the electron events are each weighted by 𝑄4 to reflect the difference in the
electron and neutrino elementary interactions.
Exploiting these similarities within the same code is invaluable for minimizing the
systematic uncertainties of future high-precision neutrino-oscillation experiments. Os-
cillation analysis uncertainties exceeding 1% for signal and 5% for backgrounds may
substantially degrade the experimental sensitivity to CP violation and mass hierar-
chy [146]. Such uncertainties already include the relevant neutrino-nucleus interaction

109
 0.1
0.2
Area Normalized

Area Normalized
0.08
0.15
0.06
e
0.1 ν
0.04
0.05 ν
0.02 e
0 0
0 0.2 0.4 0.6 0.8 1 0.2 0.4 0.6 0.8 1 1.2 1.4
Energy Transfer [GeV] Q2 [GeV2/c2]

Figure 4-12: Comparison of semi-exclusive 1.16 GeV lepton-carbon scattering for

𝑄2 ≥ 0.1 GeV2 . The number of generated events is plotted versus energy transfer
(left) and 4-momuntm transfer squared (right) for events with exactly one proton
with 𝑃𝑝 ≥ 300 MeV/c, no charged pions with 𝑃𝜋 ≥ 70 MeV/c and no neutral pions or
photons of any momentum for eGENIE electrons (orange) and GENIE CC 𝜈𝜇 (blue).
The electron events have been weighted by 𝑄4 . Both curves are area normalized.

uncertainties. These uncertainties are driven by the choices of the nuclear models and
cross-section configurations available in event generators like GENIE.

Figure 4-13 shows that there is a larger difference among QE scattering models
than there is between QE electron and neutrino scattering using the same nuclear
model. All six panels show a “ridge”, a maximum in the cross section as a function of
energy transfer and momentum transfer. The length of the ridge, namely the decrease
in intensity as the energy and momentum transfers increase, reflects the momentum
transfer dependence of the nucleon form factors used in the cross section model. The
width of the distribution perpendicular to the ridge reflects the width of the nuclear
momentum distribution.

The momentum distribution of the LFG model cuts off at about 260 MeV/c for
C, whereas the RMF and the RFG models have “tails” that extend to much larger
momenta, as shown in figure 4-14. The Nieves cross section decreases more slowly
with momentum transfer than the others. For GSuSAv2, the electron cross section
appears to decrease slightly faster with momentum transfer than the neutrino cross
section, possibly reflecting differences in the axial and vector nucleon form factors.

Our ability to use the GENIE code to transfer knowledge gained from electron

110
 ν SuSav2 ν Nieves ν Llewellyn-Smith
1 Relativistic MF 1 Local FG 1 Relativistic FG

0.8 0.8 0.8

ω [GeV]

ω [GeV]

ω [GeV]
0.6 0.6 0.6

0.4 0.4 0.4

0.2 0.2 0.2

0.4 0.6 0.8 1 1.2

e SuSav2 1.4 1.6 0.4 0.6 0.8 1 1.2
e Rosenbluth 1.4 1.6 0.4 0.6 0.8 1 1.2
e Rosenbluth 1.4 1.6
1 Relativistic qMF
3
[GeV/c] 1 Local FG q3 [GeV/c] 1 Relativistic qFG
3
[GeV/c]

0.8 0.8 0.8

ω [GeV]

ω [GeV]

ω [GeV]
0.6 0.6 0.6

0.4 0.4 0.4

0.2 0.2 0.2

0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
q [GeV/c] q [GeV/c] q [GeV/c]
3 3 3

Figure 4-13: Number of simulated events for QE scattering on 12 C at 1.161 GeV

with 𝑄2 ≥ 0.1 shown as a function of the energy transfer 𝜔 and the momentum
transfer 𝑞3 = |⃗𝑞 | for all the available nuclear models in GENIE for neutrinos (top)
and for electrons (bottom). (left) the GSuSAv2 model which uses a Relativistic Mean
Field momentum distribution, (middle) the Nieves or Rosenbluth cross section with
the Local Fermi Gas momentum distribution, and (right) the Llewellyn-Smith or
Rosenbluth cross section with the Relativistic Fermi Gas momentum distribution.
The electron events have been weighted by 𝑄4 .

scattering depends critically on the implementation of its components. Because of

its modular design, all reaction models in GENIE use the same nuclear model, for
instance RFG or LFG. Although the electron scattering capability was added after
the initial code release, many of the reaction models used electron scattering data
to construct the vector components of neutrino interactions. This was true for the
RES [110,125] and the DIS [156] interactions. The difference between vector neutrino-
and electron-scattering is an overall factor illustrated in equation 4.1 and an appro-
priate change in the form factors.

Similarly, figure 4-15 shows that the distribution of MEC events is very similar for
electrons and for neutrinos within the same model. Thus, measurements of electron
scattering will be able to significantly constrain models of neutrino scattering.

Both QE and MEC models use the same vector form factors for neutrino and for

111
 Probability Density
0.08
RFG
0.06

0.04 LFG

0.02

0
0 0.1 0.2 0.3 0.4 0.5
pp [GeV/c]

Figure 4-14: Initial momentum distribution of protons in simulated QE C(𝑒, 𝑒′ 𝑝)

events at 𝐸 = 1.161 GeV for the Local Fermi Gas (solid histogram) and Relativistic
Fermi Gas (dotted histogram) models. The two curves are normalized to have the
same area.

MEC ν e
1 SuSav2 1
0.8 0.8
ω [GeV]

ω [GeV]

0.6 0.6
0.4 0.4
0.2 0.2
0 0
0.4 0.6 0.8 1 1.2 1.4 1.6 0.4 0.6 0.8 1 1.2 1.4 1.6
q [GeV/c] q [GeV/c]
3 3

Figure 4-15: Number of simulated events as a function of the energy transfer 𝜔 and
of the momentum transfer 𝑞3 = |⃗𝑞 | for neutrinos (left) and for electrons (right) using
GSuSav2 for MEC interactions. The electron events have been scaled by 𝑄4 and all
the samples have been generated with 𝑄2 ≥ 0.1.

electron scattering. QE models can use nucleon form factors from electron scatter-
ing [157], but MEC models must calculate the form factors.
The GENIE DIS cross section comes from the Bodek-Yang model [156] for the
full cross section which extends to 𝜋N threshold. The cross section is scaled in the
RES region so that it agrees with neutrino-deuterium data [179]. Since a single factor

112
is used to fit the model to the neutrino data, the high-quality electron-proton and
electron-deuterium data will be poorly described. While the total neutrino cross sec-
tion and some of the hadronic content of the final state are loosely constrained by the
neutrino-deuterium data, the vector component of the models is poorly constrained.
The QE models describe the data reasonably well in the low energy-transfer region.
Similarly, the largest energy-transfer portions of figures 4-7 and 4-10 show a reasonable
agreement between GENIE and data. However, at intermediate energy transfer, the
RES modeling disagrees with the data for both nuclear and nucleon targets, also
observed in [149]. This is due to the use of RES form factors that are not up-to-date
and the way the nonresonant contribution was modeled.
Improvements are in progress but are not simple and therefore not available at this
time. A possible short-term fix would be to include the electron-proton and electron-
deuterium inclusive electron-scattering models of Bosted and Christy [190,191]. Alter-
natively, the vector resonant form factors could be updated using electroproduction
data from JLab and elsewhere. A fit to that data is available [192] and partially
implemented in GENIE, but it does not include nonresonant scattering. A more
comprehensive solution would be to use the recent DCC model [193, 194] to simul-
taneously describe both resonant and nonresonant scattering of both electrons and
neutrinos.
The comparisons presented in this work are focused on inclusive electron cross-
section measurements. Yet, forthcoming neutrino oscillation experiments like DUNE
will use 4𝜋-coverage liquid argon time projection chamber tracking detectors in order
to investigate exclusive interaction channels. Until 2021, no prior tests of the event
generator performance against such exclusive channels existed with electron scattering
events. Chapters 5 and 6 present the work that alleviated this shortage using wide
phase-space electron scattering data sets from Hall B and the e2a experiment at JLab.

113
114
Chapter 5

The CLAS Electrons-For-Neutrinos

Experiment At Thomas Jefferson
National Laboratory

5.1 Continuous Electron Beam Accelerator Facility

To test the event generator performance against wide phase-space exclusive interac-
tion channels with electron-nucleus scattering events, data sets from Hall B and the
e2a experiment at Thomas Jefferson Laboratory (JLab) in Newport News VA were
used.

The Continuous Electron Beam Accelerator Facility (CEBAF) electron accelera-

tor [195] provided electron beams up to 6 GeV to three experimental halls, referred
to as halls A, B, and C, until 2015. CEBAF has since been upgraded to 12 GeV and
a new hall (Hall D) was constructed in the context of the 12 GeV upgrade. A laser
shone on a gallium arsenide cathode to emit photo-electrons. The electrons were then
accelerated by cavities operating at 1.497 GHz. Each of the halls received electron
bunches every 2 ns. Those bunches can have different electron densities so that each
hall could be supplied with a different beam current. A schematic of the facility is
shown in figure 5-1.

115
Figure 5-1: Schematic view of the accelerator facility and the experimental halls at
Jefferson Lab. Figure adapted from [196].

The electrons are accelerated through the beamline. The latter consists of two
parallel linacs that are connected with two arcs with curvature radii of 80 m. Each
one of the linacs could increase the electron beam energy by 550 MeV. The beam was
circulated up to 5 times, for a maximum energy of 6 GeV. The beam energy spread
was ≈ 10−4 .

5.2 The CEBAF Large Acceptance Spectrometer

The e2a data sets were collected at Hall B using the CEBAF Large Acceptance
Spectrometer (CLAS) spectrometer shown in figure 5-2. The spectrometer had ≈
50% angular coverage and ran at luminosities up to 1034 cm−2 sec−1 .
CLAS used a toroidal magnet for momentum reconstruction. The magnet di-
vided CLAS into six almost identical sectors. Each sector had three regions of drift
chambers (DC) for charged particle trajectory measurements, threshold Cherenkov
counters (CC) for electron identification, scintillator counters (SC) for timing and
for charged hadron identification, and electromagnetic calorimeters (EC) for electron
identification and for photon and neutron detection. The 𝜃-angle coverage for the DC
is 8𝑜 − 140𝑜 , for the SC 9𝑜 − 140𝑜 , and for the EC 8𝑜 − 45𝑜 .

116
 Scintillators (timing)
Cherenkov (e– ID)
Drift chambers
(tracking)
e–b
eam

Target Calorimeters
(energy)

≈8m

Figure 5-2: Drawing of the CLAS detector showing the sector structure and the
different detectors. The beam enters from the upper left side. The target is located
at the center of the detector.

5.3 Toroidal Magnet

The torus magnet consisted of six sectors and was made of iron-free super conducting
coils oriented around the beam axis [197]. The torus had a 2 T maximum magnetic
field. The coils had a length of 5 m, and a width of 2.5 m. The magnetic field was 5
times smaller at large angles. The central region of the spectrometer had no magnetic
field allowing for polarized target operation.

5.4 Drift Chambers

The Drift Chambers (DC) are used for tracking and momentum measurements of the
charged particles produced out of the interactions. The DC have the structure shown

117
Figure 5-3: The CLAS superconducting toroidal magnet. Figure adapted from [197].

in figure 5-4 (left) and more details can be found in [198].

Figure 5-4: (Left) Illustration of the region 3 drift chamber structure for one of the
CLAS sectors. (Right) Schematic representation of the thee drift chamber regions.
Figures adapted from [198].

Each one of the modules is filled with a gas mixture of 90% Argon and 10% CO2 .
The charged particles that transverse the DC ionize the gas mixture. The produced
electrons drift towards the anode wires and the corresponding ions drift towards the
cathode wires. The relevant electrical signal is used to determine the distance from
the charged particle trajectory to the wires.
As the ionization electrons approach the wires, the increased electric field results in
the multiplication of the electron-atom collisions. The latter results in the production
of more electrons by a factor of ≈ 104 and results into an “avalanche”. The CO2
molecule serves as a quenching gas and prevents the creation of secondary avalanches.

118
 The DC consist of three radial regions as shown in figure 5-4 (right). Regions 1 and
3 are located in a low magnetic field region. Each one of these regions is made of two
superlayers, one axial and one stereo (at a stereo angle of 6𝑜 ), to allow for complete
coordinate reconstruction. The layers consist of six layers of 192 hexagonal drift cells
with the 20-micron sense wire at the center and six shared field wires creating the
electric field, with the exception of the innermost layer in region 1, which is made of
four layers. The DC structure provides a position resolution ≈ 400 𝜇m, ≈ 5 mrad for
the angular variables and ≈ 1% (0.5%) for the hadron (electron) momenta.

5.5 Electromagnetic Calorimeter

The electromagnetic calorimeter (EC) [199] was used to trigger on and to identify
the electons. It further had a high momentum neutron detection efficiency of ≈ 50%
and a high photon detection efficiency for energies above 300 MeV. It was also used
to measure neutral particles with an efficiency of ≈ 60%. It consists of 6 modules
corresponding to each one of the detector sectors. Each one of the modules has a
triangular shape with a projected vertex at the CLAS target location ≈ 5 m. Each
element has a thickness corresponding to 16 radiation lengths and is made of 39 layers
of alternating 10 mm thick scintillator strips and 2.2 mm of lead sheets. All the layers
have triangular shapes. In each layer, the scintillator strips are parallel to a given
side of the triangle. Each subsequent layer is rotated by 120𝑜 with respect to the
previous layer. That results in the formation of three views, namely u,v, and w, as
can be seen in figure 5-5.
Each view consists of 13 layers. The scintillator material thickness for each EC
module is 39 cm and the lead one is 8.4 cm. This proportionality between the scin-
tillator and the lead layers leads to one third of the shower energy being deposited
in the scintillator part. The time resolution for electrons and neutrons is 200 and
600 ps, respectively. The 13 layers are combined into an inner (5 layers) and an outer
(8 layer) stack in order to provide longitudinal sampling of the showers and a hadron
identification.

119
Figure 5-5: Schematic view of a CLAS electromagnetic calorimeter module. Figures
adapted from [199].

The EC energy resolution is given by the expression

∆𝐸 0.093
= 0.003 + √︀ (5.1)
𝐸 𝐸 [𝐺𝑒𝑉 ]

5.6 Cherenkov Counters

There are six Cherenkov Counters (CC) [200], which are used to trigger on the elec-
trons and to separate between pions and electrons. A schematic of the CC is shown
in figure 5-6.
Each CC covers the polar angular range of 𝜃 = 8 - 45𝑜 and consists of eighteen
regions. Each region is made of two modules called segments. The CC is filled with
perfluorobutane (𝐶4 𝐹10 ) gas with an index of refraction of n = 1.00153. That yields
a 2.5 GeV energy threshold for pions and a 10 MeV threshold for electrons. When
the particle velocity (v) is greater than the speed of light (c) in the detector medium
(c/n), Cherenkov light is emitted. The produced light is then directed to the light
collections cone using elliptical and hyperbolic mirrors in order to get focused to the
PMT.

120
Figure 5-6: Optical arrangement of one of the optical modules of the CLAS Cherenkov
detector, showing the optical and light collection components. Figures adapted
from [200].

5.7 Time-Of-Flight Detector

The time-of-flight detectors (TOF) [201] were made of scintillator paddles and are
shown in figure 5-7. Each scintillator was 5 cm thick and 10-15 cm wide. They ranged
in length between 25-450 cm. Using the time-of-flight (t) from the TOF subsystem
and the distance (d) based on tracking information from the DC, the velocity of the
charged particles would be obtained (v = d/t). The particle mass would be calculated
by combining the information of the particle momentum from the DC, as shown in
equation 5.2.

√︀
1 − 𝛽2
𝑝
𝑚= (5.2)
𝛽

There exist 57 TOF paddles in each sector of the detectororiented perpendicular

to the beamline. Each one covers ≈ 2𝑜 . The overall coverage in 𝜃 starts at 8𝑜 and
goes up to 142𝑜 , organized in four panels. There is a PMT attached at the end of
each paddle. The time resolution for electrons in the TOF is 163 ps.

121
Figure 5-7: Schematic view of the TOF counters in one sector illustrating the grouping
into four panels. Figure adapted from [201].

5.8 E2a Targets

The Hall B e2a experiment ran between April 15-May 27 1999. During the data taking
period, data were collected using the energies and targets presented in table 5.1.

Table 5.1: Target areal densities and integrated charges for the 𝑒4𝜈 data sets.

Target Energy Areal Density [g/cm2 ] Integrated Charge [mC]

3
He 1.159 0.2770 0.08
3
He 2.257 0.2770 0.08
3
He 4.453 0.2770 0.11
4
He 2.257 0.5375 1.17
4
He 4.453 0.5375 0.98
12
C 1.159 0.1786 0.08
12
C 2.257 0.1786 1.79
12
C 4.453 0.1786 2.84
56
Fe 2.257 0.1181 0.22
56
Fe 4.453 0.1181 0.31

The 2.257 and 4.453 GeV data sets used a torus current of 2250 A, while the
1.159 GeV ones were obtained with a 750 A torus current. The beam current ranged

122
between 3-18 nA. The solid targets (12 C and 56
Fe) were 0.9 × 0.9 cm2 square plates
with a thickness of 1 mm. The liquid targets (3 He and 4 He) were stored in cylindrical-
shaped vessel with a diameter of 2.8 cm. More details on the used data sets can be
found in the already published analyses [202–207].

123
124
Chapter 6
Electrons-For-Neutrinos Results
[Nature 599, 565–570 (2021)]

6.1 Electron Data Mining Analysis

As discussed in section 1.5, the incomplete lepton-nucleus knowledge can be lever-
aged using the fact that electrons and neutrinos interact similarly with nuclei. While
previous work has compared these interaction models with inclusive electron scatter-
ing [42,149], this analysis is the first comparison of electron scattering data with these
interaction models where events with one or more detected hadrons are used [208].
The data presented here can therefore test and constrain neutrino-nucleus interaction
models to be used in analysis of neutrino oscillation measurements.
For the purposes of the “Electrons-For-Neutrinos” (e4𝜈) analysis, electron scatter-
ing data sets from the CEBAF Large Acceptance Spectrometer (CLAS) [209] at the
Thomas Jefferson National Accelerator Facility (JLab) were used. The components
of the detector are described in detail in chapter 5.
The e2a data sets detailed in section 5.8 were used, which were measured in 1999
and reported in many published analyses [202–207]. These include electron scattering
events on 4 He, 12
C, and 56
Fe nuclei at beam energies of 1.159, 2.257 and 4.453 GeV.
The beam energy equaled the injector energy plus the pass number times the linac
energy. The three-pass beam energy was measured using the Hall A arc measurement
and the four pass energy was measured using the Hall C arc measurement. These
gave a central linac energy of 1.0979 GeV and the Hall B one-, two-, and four-pass

125
beam energies of 1.159, 2.257, and 4.453 GeV, respectively. An uncertainty of 2×10−3
was assigned to these energies, based on the difference between the Hall A and Hall C
measurements. The incident energies used in this analysis span the range of typical
accelerator-based neutrino beam energies [20, 210], as can be seen in figure 6-1. The
carbon data are relevant for scintillator-based experiments such as MINER𝜈A and
NO𝜈A [211] and similar to the oxygen in water-based Cherenkov detectors such as
Super-Kamiokande (SK) [22, 23] and HK [212]. The iron is similar to the argon in
the liquid argon time projection chambers of MicroBooNE [93], the Fermilab short-
baseline oscillation program [94] and DUNE [213]. Many nuclear interaction processes
are mass dependent, so it is important to measure a range of target nuclei.

(a) DUNE (b)

T2K
NOvA
ν Flux [arb.]

ν Flux [arb.]

0 1 2 3 4 5 1 2 3 4 5
Eν [GeV] Eν [GeV]

Figure 6-1: The expected energy distribution of different 𝜈𝜇 beams, (left) before
oscillation at the near detector and (right) after oscillation at the far detector. The
vertical lines show the three electron beam energies of this measurement. The NO𝜈A
far-detector beam flux is calculated using the near detector flux and the neutrino
oscillation parameters from [1].

Electrons with energies 𝐸𝑒 ≥ 0.4, 0.55 and 1.1 GeV for 𝐸𝑏𝑒𝑎𝑚 = 1.159, 2.257,
and 4.453 GeV respectively, and angles 15∘ ≤ 𝜃𝑒 ≤ 45∘ were detected. Protons
with momenta 𝑝𝑝 ≥ 300 MeV/𝑐 and angles 𝜃𝑝 ≥ 10∘ , charged pions with momenta
𝑝𝜋 ≥ 150 MeV/𝑐 and angles 𝜃𝜋+ ≥ 10∘ and 𝜃𝜋− ≥ 22∘ , and photons with energy
𝐸𝛾 ≥ 300 MeV and 8 ≤ 𝜃𝛾 ≤ 45∘ were detected. Separate fiducial cuts were applied
for electrons, negatively-charged pions, positively-charged particles, and photons, to
select momentum-dependent regions of CLAS where the detection efficiency was con-

126
stant and close to one. These hadron detection thresholds are similar to those of
neutrino detectors [87], however neutrino detectors have full angular coverage and
lower lepton energy thresholds.
Apart from the aforementioned momentum and angular cuts, additional angular
outlines are applied to account for the detector acceptance. The minimum electron
angle as a function of electron momentum 𝑝 for each beam energy was determined as

7∘
𝜃𝑒1.1 ≥ 17∘ + (6.1)
𝑝 [GeV]
10.5∘
𝜃𝑒2.2 ∘
≥ 16 + (6.2)
𝑝 [GeV]
15∘
𝜃𝑒4.4 ≥ 13.5∘ + (6.3)
𝑝 [GeV]
and the minimum 𝜋 − angle as

4∘
𝜃𝜋1.1 ∘
− ≥ 17 + (6.4)
𝑝 [GeV]

and

7∘
𝜃𝜋2.2,4.4 ≥ 25∘ + (6.5)
𝑝 [GeV]
−

for 𝑝𝜋− < 0.35 GeV/c and

10∘
𝜃𝜋2.2,4.4 ≥ 16∘ + (6.6)
𝑝 [GeV]
−

for 𝑝𝜋− ≥ 0.35 GeV/c. The minimum 𝜋 + and proton angle was 𝜃 > 12∘ for all
data sets and momenta.
The momentum and charge of the outgoing charged particles were obtained from
their measured positions in the drift chambers and the curvature of their trajectories
in the magnetic field. Electrons were identified by requiring that the track originated
in the target, produced a time-correlated signal in the Cherenkov counter, and de-

127
posited enough energy in the electromagnetic calorimeter. The charged pions and
protons were separated by requiring that the track originated in the target and that
the measured time of flight agreed with that calculated from the particle’s momentum
and assumed mass. Photons were identified by requiring a signal in the electromag-
netic calorimeter which implied a velocity greater than ≈ 0.96 c [214].
Elastic electron scattering from hydrogen was used to correct the electron momen-
tum as a function of angle for uncertainties in the CLAS magnetic field and in the
tracking chamber locations. These corrections also significantly narrowed the elastic
peak width. Typical correction factors were less than 1%. The momentum correction
factors at lower scattered electron energies were tested using the H(𝑒, 𝑒′ 𝜋 + )𝑋 and
3
He(𝑒, 𝑒′ 𝑝𝑝)𝑋 reactions and found that they gave the correct missing mass for the
undetected neutron [208].
Low momentum protons were corrected for energy losses traversing the target
and detector material. The CLAS GEANT Monte Carlo (MC) was used to simulate
the proton energy loss in CLAS as a function of proton momentum. The maximum
correction was about 20 MeV/c for a proton momentum of 300 MeV/c. The correction
was negligible for protons with momenta greater than 600 MeV/c.
The results from the e2a data sets were compared to predictions from the GE-
NIE [33] simulation, which is used by most neutrino experiments in the USA and has
an electron-scattering version (eGENIE) that was recently overhauled to be consistent
with the neutrino counterpart (𝜈GENIE), as detailed in [42]. GENIE includes quasi-
elastic lepton scattering (QE), meson exchange currents (MEC), resonance production
(RES) and deep inelastic scattering (DIS), as well as rescattering via final state in-
teractions (FSI) of the outgoing hadrons. The two GENIE configurations already
presented in section 4.1 were compared. These include the significantly improved
G2018 setup which reproduces measured neutrino [215] and electron inclusive cross
sections, and newly implemented SuSAv2 that uses modern, theoretically-inspired,
recently-implemented QE and MEC models [28]. The resulting simulated events
were then analyzed using the same code as the data and the two were compared.
Electrons, unlike neutrinos, radiate bremsstrahlung photons in the electric field

128
of the nucleus. Events where the photons from scattered-electron radiation were
detected in CLAS were vetoed. It was assumed that the photons came from either
radiation by the outgoing electron approximately parallel to its motion or from 𝜋 0
decay. The radiated photons were identified by requiring that they be detected within
∆𝜑𝛾,𝑒′ ≤ 30∘ and ∆𝜃𝛾,𝑒′ ≤ 40∘ of the scattered electron and removed them from the
data set. The events that were removed are indicated by the red box in figure 6-2.

Data 56Fe @ 2.2 GeV

100
103
80
∆θγ ,e' [deg]

60 γ from radiation
102

40
10
20 γ from π
0
γ from π
0

0 1
−150 −100 −50 0 50 100 150
∆φ [deg]
γ ,e'

Figure 6-2: ∆𝜃𝛾,𝑒′ as a function of ∆𝜑𝛾,𝑒′ . The red box indicated the region with
radiated photons which was removed in our analysis.

The incoming and outgoing electrons can each radiate a real photon, which changes
the kinematics of the interaction or the detected particles, and there can be vertex
or propagator corrections that change the cross section. When comparing electron
scattering data to models, either the data or the model needs to be corrected for ra-
diative effects. Published electron scattering cross sections are typically corrected for
radiative effects, but this correction is complicated and somewhat model-dependent.
A framework for electron radiative corrections in GENIE was implemented for the
first time to allow comparisons to nonradiatively corrected data. The framework al-
lows electron radiation, which can change the kinematics of the event by changing
either the incident or scattered electron energy through radiation of a real photon.

129
We modeled external radiation in the same way as the JLab SIMC event genera-
tor [216, 217]. The implementation takes advantage of the peaking approximation
that greatly simplifies the calculation of the angular distribution of the emitted pho-
ton radiation by making the assumption that radiation along the direction of a given
particle can be interpreted as radiation due to that particle [218]. Future versions
of eGENIE will incorporate cross section changes due to vertex and propagator cor-
rections. The radiative correction procedure was validated by comparing a simulated
sample to electron scattering from protons at JLab. Figure 6-3 shows the data com-
pared to the GENIE simulation with and without radiative corrections [219]. The
radiatively corrected calculation is clearly much closer to the data. This correction
can be used for comparisons with nonradiatively corrected data.

×103
# events

12 Data
GENIE + radiative correction
10
GENIE default
8

0
4.3 4.305 4.31 4.315 4.32 4.325 4.33
1
H(e,e'p) E [GeV]
cal

Figure 6-3: Number of events vs 𝐸𝑐𝑎𝑙 = 𝐸𝑒′ + 𝑇𝑝 the scattered electron energy plus
proton kinetic energy for 4.32 GeV H(𝑒, 𝑒′ 𝑝). Black points are data, the blue histogram
shows the unradiated GENIE prediction and the black histogram shows the GENIE
prediction with electron radiation. The GENIE calculations have been scaled to have
the same integral as the data.

The primary focus of this analysis was events with one electron and zero pions
or photons from 𝜋 0 decay above threshold, which are referred to as (e,e’)0𝜋 . That
choice was made to maximize the contribution of QE events where the incident lepton
scattered from a single nucleon in the nucleus, as is done in many neutrino oscillation

130
analyses [1, 220]. Furthermore, events with one detected electron, one proton, and
zero pions, denoted here as (e,e’p)1𝑝0𝜋 were examined. These events were expected to
be dominated by well-understood QE events.

Because the CLAS geometrical coverage is incomplete (≈ 50%), undetected pions

and photons needed to be subtracted to obtain true (e,e’)0𝜋 and (e,e’p)1𝑝0𝜋 event sam-
ples. The undetected pion, photon, and proton (if applicable) contribution was quan-
tified from the events with detected pions, photons or protons. The pion-production
cross section was assumed to be independent of 𝜑𝑞𝜋 . The latter is the angle between
the electron-scattering plane (the plane containing the incident and scattered elec-
trons and the virtual photon) and the hadron plane (the plane containing the virtual
photon and pion). The data-driven correction for these undetected hadrons is out-
lined below for the (e,e’p)1𝑝0𝜋 case. For each detected (𝑒, 𝑒′ 𝑝𝜋) event, the proton-pion
pair was rotated around the momentum transfer direction ⃗𝑞 randomly multiple times,
as can be seen in figure 6-4. For each rotation, it was determined whether the particle
was within the fiducial region of the detector. The particle acceptance would then
be 𝐴𝜋 = 𝑁𝑑𝑒𝑡 /𝑁𝑟𝑜𝑡 , where 𝑁𝑟𝑜𝑡 is the number of rotations and 𝑁𝑑𝑒𝑡 is the number of
times the pion would have been detected. The corresponding number of undetected
(𝑒, 𝑒′ 𝑝𝜋) events for that detected (𝑒, 𝑒′ 𝑝𝜋) event is (𝑁𝑟𝑜𝑡 − 𝑁𝑑𝑒𝑡 )/𝑁𝑑𝑒𝑡 . That was used
as a weight to subtract for the undetected pion events. For example, if one specific
(𝑒, 𝑒′ 𝑝𝜋) event would have been detected 250 times out of 1000 rotations, then it was
inferred that for each detected event, there were three more that were not detected.
The appropriate variables were calculated for that event and subtracted it from the
corresponding distributions with a weight of three. That was done separately for
𝜋 + , 𝜋 − and photons in order to obtain a true (e,e’p)1𝑝0𝜋 sample. The same process is
repeated for the (e,e’)0𝜋 channel, where the corrections included undetected protons.

Higher multiplicity events were also accounted for, such as for events with two
detected 𝜋 ± or photons. When these events were rotated, each rotated event could
have been detected as a 2𝜋 event, a 1𝜋 event, or a 0𝜋 event. If it appeared as a
0𝜋 event, its contribution was subtracted from the various 0𝜋 spectra as described
above. If it appeared as a 1𝜋 event, it was included in the set of 1𝜋 events with the

131
Figure 6-4: Schematic illustration of the data driven background correction using
detected (e,e’p𝜋) events.

appropriate negative weight. It was then treated as a regular 1𝜋 event, which was
then rotated and added to the 0𝜋 data set. Some of the detected 1𝜋 events were
actually 2𝜋 events with an undetected pion. When the effect of these events was
accounted for, there were fewer true 1𝜋 events left. This reduced the contamination
of the 1𝜋 events in the 0𝜋 channel.

In practice, the process was initiated with the highest multiplicity events. Then,
their contributions to each of the detected lower multiplicity channels were subtracted.
The process was repeated recursively by rotating the higher multiplicity events. In
this way, their contributions to the lower multiplicity channels were determined and
subtracted, and then each of the lower multiplicity channels in turn were considered.
Event multiplicities up to three pions and photons (total) for the (𝑒, 𝑒′ )0𝜋 channel and
up to three protons, pions and photons (total) for the (𝑒, 𝑒′ 𝑝)1𝑝0𝜋 channel, where the
subtraction converged, were considered. The effects of the subtraction and its con-
vergence can be seen in figure 6-5 for 𝐸𝑄𝐸 . The number of events with an undetected
𝜋 ± or photon is about equal to the number of events with a detected 𝜋 ± or photon,
consistent with the ≈ 50% CLAS geometrical acceptance. The effect of including two
𝜋 ± or photon events is much less than that of the one 𝜋 ± or photon events and the

132
effect of including three 𝜋 ± or photon events is negligible.

(a) No cuts (b)

56
Fe Detected 1 π±/ γ 56
Fe No detected π±/ γ
Weighted Events / GeV

Weighted Events / GeV

0.2 Undetected 1π / γ (-)
±
0.6 Subtract undetected π±/ γ
Undetected 2π±/ γ (+)

0.4
0.1
0.2

0 0
2 3 4 5 2 3 4 5
EQE[GeV] EQE [GeV]

Figure 6-5: The effect of undetected pion subtraction. The number of weighted
events as a function of reconstructed energy 𝐸𝑄𝐸 for 4.453 GeV Fe(𝑒, 𝑒′ ) events for
(left) events with a detected 𝜋 ± or photon (blue), events with one (red) or two (light
brown) undetected 𝜋 ± or photons and (right) all (𝑒, 𝑒′ 𝑋) events with detected or
undetected 𝜋 ± or photon (blue), (𝑒, 𝑒′ ) events with no detected 𝜋 ± or photon (red),
and (𝑒, 𝑒′ ) events after subtraction for undetected 𝜋 ± or photon (light brown).

The subtraction method was tested by applying it to eGENIE events. The result-
ing subtracted spectra agreed reasonably with the true 1𝑝0𝜋 spectra as can be seen
in figure 6-6. The method diverges for total hadron multiplicities greater than four
due to the proton and pion multiplicity differences shown in figure 6-7. It is clear
that eGENIE dramatically overpredicts the number of events with large proton and
pion multiplicities.
The CLAS acceptance maps were used to determine the probability that each
particle produced by eGENIE was detected as a function of the momentum, the
angular orientation, and the particle species. Figure 6-9 shows the electron acceptance
map for 12 C at 𝐸𝑏𝑒𝑎𝑚 = 1.159 GeV as a function of (left) cos𝜃𝑒 vs 𝜑𝑒 and (right) cos𝜃𝑒
vs momentum 𝑝𝑒 illustrating an acceptance greater than 90% across the majority of
the detector fiducial volume.
The particle momenta were smeared with an effective CLAS resolution. Namely,
electrons and proton momentum resolutions of 0.5% and 1%, respectively, for the
2.257 and 4.453 GeV data and 1.5% and 3% for the 1.159 GeV data, which was taken
with a lower torus magnetic field, were used.

133
 12
C @ 2.257 GeV
0.1

Normalized Yield
0.08

0.06

0.04 Unsubtracted (e,e'p)1p0π

0.02 True (e,e'p)1p0π

Subtracted (e,e'p)1p0π
0
0.8 1 1.2 1.4 1.6 1.8 2 2.2 2.4
(e,e'p)1p0π Ecal [GeV]

Figure 6-6: Illustration of the successful closure test of the data driven correction
for undetected particles as a function of 𝐸 𝐶𝑎𝑙 using the (e,e’p)1𝑝0𝜋 channel on 12 C at
𝐸𝑏𝑒𝑎𝑚 = 2.257 GeV. The contribution of the unsubtracted (e,e’p)1𝑝0𝜋 spectrum (black)
is reduced to the subtracted (e,e’p)1𝑝0𝜋 spectrum (magenta), which is in reasonable
agreement with the true (e,e’p)1𝑝0𝜋 spectrum (green).

107

106 Protons
# Events

105 π±

104 SuSav2
G2018
0 1 2 3 4
Multiplicities

Figure 6-7: The proton (black) and charged pion (blue) multiplicities for data
(points), SuSav2 (solid histogram) and G2018 (dashed histogram) for 2.257 GeV
carbon. Error bars show the 68% (1𝜎) confidence limits for the statistical and point-
to-point systematic uncertainties added in quadrature. Error bars are not shown
when they are smaller than the size of the data point. Normalization uncertainties of
3% not shown.

The cross section as a function of variables of interest for particles above the
minimum angles shown in equations 6.1-6.6 was determined in several steps. All

134
 1.159 GeV (a) 2.257 GeV (b) 4.453 GeV (c)

Acceptance Correction

Acceptance Correction

Acceptance Correction
12 12 12
10 12
10 10
C
8 8 4 8
He
6 6 6
4 4 56 4
Fe
2 2 2
Acceptance Uncertainty [%]

Acceptance Uncertainty [%]

Acceptance Uncertainty [%]

0.6 0.8 1 (d) 1.2 1 1.5 2 (e) 2 3 4 (f)
15 15 15

10 10 10

5 5 5

0 0 0
0.6 0.8 1 1.2 1 1.5 2 2 3 4
1.6 (g) 1.6 (h) 1.6 (i)
Radiation Correction

Radiation Correction

Radiation Correction
1.4 1.4 1.4
1.2 1.2 1.2
1 1 1
0.8 0.8 0.8
0.6 0.6 0.6
0.6 0.8 1 1.2 1 1.5 2 2 3 4
(e,e'p)1p0π Ecal [GeV]

Figure 6-8: (Top row) Acceptance correction factors, (middle row) acceptance cor-
rection factor uncertainties, and (bottom row) electron radiation correction factors
plotted vs E𝑐𝑎𝑙 for the three incident beam energies. Results for carbon are shown
in black, helium in green and iron in magenta. The left column (a,d,g) shows the
1.1 GeV results, the middle column (b,e,h) shows the 2.2 GeV results and the right
column (c,f,i) shows the 4.4 GeV results.

100 2.5 100

0.9 98 2 98
Acceptance (%)

Acceptance (%)
pe [GeV/c]

0.8 96 1.5 96
cosθe

0.7 94 94
1
0.6 92 92
0.5
90 90
0 50 100 150 200 250 300 0.6 0.7 0.8 0.9
φe [deg] cosθe

Figure 6-9: Electron acceptance maps for 12 C at 𝐸𝑏𝑒𝑎𝑚 = 1.159 GeV as a function of
(left) cos𝜃𝑒 vs 𝜑𝑒 and (right) cos𝜃𝑒 vs momentum 𝑝𝑒 .

events were first weighted by a factor of 𝑄4 to account for the major difference in
electron- and neutrino-nucleus scattering. The number of weighted events was then

135
determined and corrected (if appropriate) for events with undetected pions, photons,
and extra protons. The background-subtracted event distribution was divided by the
number of target nuclei per area and the number of incident beam electrons to get
the normalized yield. The delivered integrated beam charge was measured using the
CLAS Faraday Cup [209]. Electron radiation effects were corrected for by multiplying
the resulting spectra by the ratio of eGENIE without electron radiation divided by
eGENIE with electron radiation, as shown in panels g-i of figure 6-8. This includes
a multiplicative factor to account for the effects of internal radiation. Electron and
proton acceptance and other detector effects were corrected for using eGENIE. The
acceptance correction factor is the ratio of the number of true signal events without
detector effects to the number of true signal events with detector effects. The detector
effects included momentum resolution, fiducial cuts, and acceptance map effects. The
fiducial cuts determine the useful areas of the detector as a function of particle mo-
menta and angles, and the acceptance maps describe the efficiency of the detector as
a function of particle momenta and angles. This factor corrects the effective electron
and proton solid angles to almost 4𝜋. It excludes all electrons, pions and protons
below their minimum angles defined in equation 6.1-6.6. The acceptance correction
factor was obtained using both G2018 and SuSav2 shown in panels a-c of figure 6-8
as the bin-by-bin average of the two configurations. The G2018 results were shifted
so that the energy reconstruction peaks lined up at the correct beam energy. Finally,
the bin width division was taken into account.

In order to perform a sanity check of our cross-section extraction procedure, the

inclusive 1.159 37.5∘ GeV C(𝑒, 𝑒′ ) cross section was determined as follows:

𝑑𝜎 𝑁𝑒
= (6.7)
𝑑Ω𝑑𝜔 ∆Ω𝑁𝑖 𝑁𝑡

where 𝑁𝑒 is the number of detected electrons in Sector 1 within 36∘ ≤ 𝜃𝑒 ≤ 39∘

and a 12∘ range in 𝜑𝑒 , ∆Ω = sin 𝜃𝑒 𝑑𝜃𝑒 𝑑𝜑𝑒 = 6.68 msr, 𝑁𝑖 is the number of incident
electrons, and 𝑁𝑡 = 0.179 g/cm2 = 8.95×10−9 nuclei/𝜇b. The extracted cross section

136
was further compared to measurements from SLAC [182], as can be seen in figure 6-
10. The measured JLab cross section is in reasonable agreement with the GENIE
predictions and also consistent with the SLAC measurements at lower and higher
energies [182].

Figure 6-10: Comparison between the inclusive C(𝑒, 𝑒′ ) cross sections measured at
37.5∘ for data (points) and SuSav2 (lines) for the 0.961 and 1.299 GeV SLAC data
and our 1.159 GeV CLAS data.

Several major sources of systematic uncertainties were considered, including the

angular dependence of the pion-production cross section for the undetected-pion sub-
traction, the effects of fiducial cuts on undetected particle subtraction, photon iden-
tification cuts, the sector-to-sector variation of the data to eGENIE ratio, the model-
dependence of the acceptance correction, and uncertainties in the normalization mea-
surement. Table 6.1 shows the summary of the total systematic uncertainties used in
the e4𝜈 analysis.
When events containing pions were rotated around the momentum transfer vec-
tor, the cross section was assumed to not depend on 𝜑𝑞𝜋 . The 𝜑𝑞𝜋 independence of
the pion-production cross section was tested by weighting the subtraction using the
measured 𝜑𝑞𝜋 -dependent 𝐻(𝑒, 𝑒′ 𝑝𝜋) cross sections of reference [221]. This changed
the subtracted spectra by about 1% and was included as a systematic uncertainty.

137
Table 6.1: Summary of the total systematic uncertainties used in the e4𝜈 analysis.

Source Uncertainty (%)

Detector acceptance
Identification cuts 2, 2.1, 4.7
Number of rotations (@1.1,2.2,4.4 GeV)
𝜑𝑞𝜋 cross-section dependence
Sector dependence 6
Acceptance correction 2-15
Overall normalization 3
Electron inefficiency 2

The subtraction of events with undetected pions depends on the CLAS acceptance
for such particles. The final spectrum should be independent of the CLAS pion
acceptance. The effect of varying the CLAS acceptance on the undetected particle
subtraction was estimated by comparing the results using the nominal fiducial cuts
and using fiducial cuts with the 𝜑 acceptance in each CLAS sector reduced by 6∘
or about 10-20%. This changed the resulting subtracted spectra by about 1% at
1.159 and 2.257 GeV and by 4% at 4.453 GeV. This difference was included as a
point-to-point systematic uncertainty.
The photon identification cuts were also varied. Photons were also identified
as neutral particle hits in the calorimeter with a velocity greater than 2𝜎 (3𝜎 at
1.159 GeV) below the mean of the photon velocity peak at 𝑣 = 𝑐. This limit was
varied by ±0.25𝜎. This gave an uncertainty in the resulting subtracted spectra of
0.1%, 0.5% and 2% at 1.159, 2.257 and 4.453 GeV, respectively.
CLAS had six almost identical sectors. The primary difference among the sectors
is the distribution of dead detector channels. These dead channels were accounted for
in our fiducial cuts and in our acceptance maps, where the effect of the dead detectors
on the particle detection efficiency was measured and applied that efficiency to the
particles generated in the eGENIE simulation. If our fiducial cuts and acceptance
maps completely accounted for the effect of the dead and inefficient detector channels,
then the ratio of data to eGENIE should be the same for all six sectors. Sectors with
anomalous data to eGENIE ratios were discarded. More precisely, sectors 3/5 were

138
discarded at 𝐸𝑏𝑒𝑎𝑚 = 1.159 GeV, as well as sectors 3/4/5 at 𝐸𝑏𝑒𝑎𝑚 = 2.257 GeV. All the
sectors were used at 𝐸𝑏𝑒𝑎𝑚 = 4.453 GeV. The variance of the ratios for the remaining
sectors was used as a measure of the uncertainty in the measured normalized yields.
This gave a point-to-point systematic uncertainty of 6%.
The acceptance correction factor uncertainty was obtained using both G2018 and
√
SuSav2 and their bin-by-bin difference divided by 12. The uncertainty was averaged
over the entire peak to avoid large uncertainties due to small misalignments, as shown
in panels d-f of figure 6-8.
The overall normalization was determined using inclusive 4.4 GeV H(𝑒, 𝑒′ ) mea-
surements. The measured and simulated H(𝑒, 𝑒′ ) cross sections agreed to within an
uncertainty of 3%, which is used as a normalization uncertainty [222].
The statistical uncertainty and the point-to-point systematic uncertainties were
added in quadrature and displayed on the data points. The total point-to-point
systematic uncertainties ranged between 7-25%, with the largest uncertainties for the
smallest cross sections.

6.2 Incident Energy Reconstruction Results

There are two general approaches for reconstructing the incident neutrino energy
based on the particle detection capabilities of the neutrino detector.
Water Cherenkov detectors only measure charged leptons and pions. If the neu-
trino scattered quasi-elastically from a stationary nucleon in the nucleus, its energy
can be reconstructed from the measured lepton as:

2𝑀𝑁 𝜖 + 2𝑀𝑁 𝐸𝑙 − 𝑚2𝑙

𝐸𝑄𝐸 = , (6.8)
2(𝑀𝑁 − 𝐸𝑙 + 𝑘𝑙 cos 𝜃𝑙 )
where 𝜖 ≈ 20 MeV is the average nucleon separation energy, 𝑀𝑁 is the nucleon
mass, and (𝑚𝑙 , 𝐸𝑙 , 𝑘𝑙 , 𝜃𝑙 ) are the scattered lepton mass, energy, momentum, and angle.
Figure 6-11 shows the 𝐸𝑄𝐸 distribution for 1.159 GeV C(𝑒, 𝑒′ )0𝜋 events, which are
most relevant for T2K and HK. A broad peak is observed centered at the real beam
energy with a large tail extending to lower energies. The peak is doppler-broadened

139
by the motion of the nucleons in the nucleus. The tail is caused by non-QE reactions
that pass the (𝑒, 𝑒′ )0𝜋 selection. The tail is cut off at the lowest energies by the CLAS
minimum detected electron energy of 0.4 GeV. The SuSAv2 eGENIE peak has the
correct width, but is somewhat larger than the data. It overestimates the tail by
about 25%. The G2018 eGENIE peak also exceeds the data, but is too narrow, with
a Gaussian width of 𝜎 = 76 MeV, compared to 89 MeV for the data. This is due to
inexact modeling of the nuclear ground state momentum distribution. The tail dips
below the data at around 0.9 GeV, and is larger than the data at lower reconstructed
energies. Neither model describes the data quantitatively well.

Data

1 SuSav2 (Total)
QE MEC
dEQE GeV

RES DIS
µb

G2018

0.5
dσ

Ebeam

0 ⇓
0.6 0.8 1 1.2 1.4
C(e,e')0π EQE [GeV]

Figure 6-11: The 1.159 GeV C(𝑒, 𝑒′ )0𝜋 cross section plotted as a function of the
reconstructed energy 𝐸𝑄𝐸 for data (black points), GENIE SuSAv2 (solid black curve)
and GENIE G2018 (dotted black curve). The colored lines show the contributions
of different processes to the GENIE SuSAv2 cross section: QE (blue), MEC (red),
RES (green) and DIS (orange). Error bars show the 68% (1𝜎) confidence limits
for the statistical and point-to-point systematic uncertainties added in quadrature.
Error bars are not shown when they are smaller than the size of the data point.
Normalization uncertainty of 3% not shown.

Figure 6-12 shows the cross section as a function of 𝐸𝑄𝐸 for 1.159, 2.257 and
4.453 GeV C(𝑒, 𝑒′ )0𝜋 events and 2.257 and 4.453 GeV Fe(𝑒, 𝑒′ )0𝜋 events. It is clear
that the mismodeling already observed in the 1.159 GeV C(𝑒, 𝑒′ )0𝜋 sample becomes

140
even more pronounced for higher energies and heavier nuclei.

1.159 GeV (x1/2) 2.257 GeV 4.453 GeV (x5)

dEQE GeV 0.4 0.4 0.4
µb

12
C
0.2 0.2 0.2
dσ

0 0 0
0.6 0.8 1 1.2 1.4 1 1.5 2 2.5 2 3 4 5
2 2
Data

dEQE GeV
1.5 1.5
SuSav2 (Total) 56
Fe µb
QE MEC
dσ 1 1
RES DIS 0.5 0.5
G2018 0 0
1 1.5 2 2.5 2 3 4 5
(e,e')0π EQE [GeV]

Figure 6-12: The 𝐴(𝑒, 𝑒′ 𝑝)0𝜋 cross section plotted as a function of the reconstructed
quasielastic energy 𝐸𝑄𝐸 for data (black points), SuSAv2 (black solid curve) and G2018
(black dotted curve). Different panels show results for different beam energy and
target nucleus combinations: (top row) Carbon target at (left to right) 1.159, 2.257
and 4.453 GeV, and (bottom) Iron target at (left) 2.257 and (right) 4.453 GeV incident
beam. The 1.159 GeV yields have been scaled by 1/2 and the 4.453 GeV yields have
been scaled by 5 to have the same vertical scale. Colored lines show the contributions
of different processes to the SuSAv2 GENIE simulation: QE (blue), MEC (red),
RES (green) and DIS (orange). Error bars show the 68% (1𝜎) confidence limits
for the statistical and point-to-point systematic uncertainties added in quadrature.
Error bars are not shown when they are smaller than the size of the data point.
Normalization uncertainties of 3% not shown.

Tracking detectors measure all charged particles above their detection thresholds.
The “calorimetric” incident neutrino energy is then the sum of all the detected particle
energies:

∑︁
𝐸𝑐𝑎𝑙 = 𝐸𝑖 + 𝜖, (6.9)

where 𝐸𝑖 are the detected nucleon kinetic energies and the lepton and meson total
energies and 𝜖 is the average total removal energy for the detected particles. This
quantity 𝜖, used in reconstructing the incident energies in equations 6.8 and 6.9, was

141
determined from the data. It is defined using the difference in the binding energies
for knocking a proton out of nucleus 𝐴 as 𝜖 = |𝑀𝐴 − 𝑀𝐴−1 − 𝑚𝑝 | + ∆𝜖. The removal
energy correction ∆𝜖 was adjusted so that the peaks in the 𝐸𝑐𝑎𝑙 spectrum for low
transverse missing momentum events reconstructed to the correct beam energy. It
was found that ∆𝜖 = 5 and 11 MeV for 12 C and 56 Fe, respectively, which are consistent
with average excitation energies from single-nucleon knockout from nuclei.
Figure 6-13 shows the cross section as a function of 𝐸𝑐𝑎𝑙 for 1.159, 2.257 and
4.453 GeV C(𝑒, 𝑒′ 𝑝)1𝑝0𝜋 events and 2.257 and 4.453 GeV Fe(𝑒, 𝑒′ 𝑝)1𝑝0𝜋 events. All
spectra show a sharp peak at the real beam energy, followed by a large tail at lower
energies. For carbon, only 30-40% of the events reconstruct to within 5% of the real
beam energy, as illustrated in table 6.2.

Table 6.2: (𝑒, 𝑒′ 𝑝)1𝑝0𝜋 events reconstructed to the correct beam energy. Peak Fraction
refers to the fraction of events reconstructed to the correct beam energy and Peak Sum
refers to the integrated weighted cross section (as shown in Fig. 6-13) reconstructed
to the correct beam energy. The peak integration windows are 1.1 ≤ 𝐸𝑐𝑎𝑙 ≤ 1.22
GeV, 2.19 ≤ 𝐸𝑐𝑎𝑙 ≤ 2.34 GeV, and 4.35 ≤ 𝐸𝑐𝑎𝑙 ≤ 4.60 GeV, respectively, for the three
incident beam energies. SuSAv2 is not intended to model nuclei lighter than 12 C.

1.159 GeV 2.257 GeV 4.453 GeV

Peak Peak Peak Peak Peak Peak
Fraction Sum [𝜇b] Fraction Sum [𝜇b] Fraction Sum [𝜇b]
Data 39 4.13 31 1.26 32 0.34
12
C SuSAv2 44 5.33 27 1.76 12 0.20
G2018 51 6.53 37 2.44 23 0.43
Data - - 20 3.73 23 1.01
56
Fe SuSAv2 - - 21 5.28 10 0.58
G2018 - - 30 8.22 19 1.48

For iron this fraction is only 20-25%, highlighting the crucial need to well model
the low-energy tail of these distributions. eGENIE overpredicts the fraction of events
in the peak at 1.159 GeV and significantly underpredicts it at 4.453 GeV. eGENIE
using SuSAv2 dramatically overpredicts the peak cross section at 1.159 and 2.257
GeV, and significantly underestimates the peak cross section at 4.453 GeV, as shown

142
 1.159 GeV (x1/2) (a) 2.257 GeV (b) 4.453 GeV (x5) (c)
2 2 2

cal Section

Cross Section

Cross Section
0.2 0.2 0.2

GeV
1.5 1.5 1.5
12
C dσ µb 0.1 0.1 0.1
dECross 1 1 1

0.5 0.5 0.5

0 0 0
0.6 0.8 1 1.2 1 1.5 2 2 3 4
(d) (e)
6 6
Data

Section

Cross Section
0.6 0.6

cal GeV
dσ µb
0.4 0.4
SuSav2 (Total) 56 4 4
Fe

dECross
0.2 0.2
QE MEC
RES DIS 2 2

G2018 0 0
1 1.5 2 2 3 4
(e,e'p)1p0π Ecal [GeV]

Figure 6-13: The 𝐴(𝑒, 𝑒′ 𝑝)1𝑝0𝜋 cross section plotted as a function of the reconstructed
calorimetric energy 𝐸𝑐𝑎𝑙 for data (black points), SuSAv2 (black solid curve) and G2018
(black dotted curve). Different panels show results for different beam energy and
target nucleus combinations: (top row) Carbon target at (left to right) 1.159, 2.257
and 4.453 GeV, and (bottom) Iron target at (left) 2.257 and (right) 4.453 GeV incident
beam. The 1.159 GeV yields have been scaled by 1/2 and the 4.453 GeV yields
have been scaled by 5 to have the same vertical scale. The insets show the cross
sections with the same horizontal scale and an expanded vertical scale. Colored lines
show the contributions of different processes to the SuSAv2 GENIE simulation: QE
(blue), MEC (red), RES (green) and DIS (orange). Error bars show the 68% (1𝜎)
confidence limits for the statistical and point-to-point systematic uncertainties added
in quadrature. Error bars are not shown when they are smaller than the size of the
data point. Normalization uncertainties of 3% not shown.

in table 6.2. eGENIE using the older G2018 models overestimates the peak cross
section at all three incident energies. It also reconstructs the peak position (i.e. the
incident energy) to be 10, 25 and 36 MeV too low for 4 He, C and Fe, respectively,
at all three beam energies. This is due to an error in the G2018 QE modeling.
This beam-energy dependence of the data-GENIE discrepancy could have significant
implications for the neutrino flux reconstruction.

At 1.159 GeV, eGENIE using SuSAv2 slightly overpredicts the low energy tail
and eGENIE using G2018 is reasonably close. Both models dramatically overpredict

143
the low energy tail at the higher beam energies shown in the insets of figure 6-
13. The tail seems to be dominated by RES and DIS at 4.453 GeV that did not
result in the production of other charged particles above detection threshold. This
overprediction has already been observed in inclusive electron scattering from the
proton and deuteron, and thus appears to be due to the electron-nucleon interaction,
rather than to the nuclear modeling [42].
SuSAv2 describes the peak cross section - the part of the cross section that re-
constructs to the correct beam energy - equally well for C and for Fe, while G2018
over estimates the peak cross section more for Fe than for C. Both models predict a
greater peak fraction relative to the data for Fe than for C, particularly at 2.2 GeV,
as shown in table 6.2. While the (𝑒, 𝑒′ )0𝜋 QE reconstruction of equation 6.8 gives a
much broader peak at the true beam energy than the calorimetric energy 𝐸𝑐𝑎𝑙 due to
the effects of nucleon motion, as shown in figures 6-14 and 6-15, it has the same tail
of lower energy events for the same (𝑒, 𝑒′ 𝑝)1𝑝0𝜋 data set.

Data/SuSav2 12
C (a) 1.4 Data/SuSav2 12
C (b)
/ 1.159 GeV / 1.159 GeV
/ 2.257 GeV 1.2 / 2.257 GeV
1 / 4.453 GeV(x4) / 4.453 GeV(x4)
1
d σ µb

d σ µb

0.8
dEFeed

dEFeed
QE
cal

10−1 0.6
0.4
0.2
10−2
−0.6 −0.4 −0.2 0 −0.6 −0.4 −0.2 0 0.2
(e,e'p)1p0π Ecal Feeddown (e,e')0π EQE Feeddown

Figure 6-14: Energy feed-down cross-sections (𝐸𝑟𝑒𝑐 − 𝐸𝑡𝑟𝑢𝑒 )/𝐸𝑡𝑟𝑢𝑒 for data (points)
and SuSav2 (lines) for 1.159 GeV (red triangles and dotted lines), 2.257 GeV (green
squares and dashed lines) and 4.453 GeV (blue dots and solid lines) on carbon for (a)
𝐸 𝑐𝑎𝑙 , and (b) 𝐸 𝑄𝐸 .

6.3 Kinematic Imbalance Results

Neutrino experiments use the “transverse variables” (TVs) outlined in section 3.2.1
to enhance their sensitivity to different aspects of the reaction mechanism. These

144
 10 Data/SuSav2 56 Data/SuSav2 56
Fe (c) Fe (d)
/ 2.257 GeV / 2.257 GeV
/ 4.453 GeV(x4) 6 / 4.453 GeV(x4)

1
d σ µb

d σ µb
4
dEFeed

dEFeed
QE
cal

10−1
2

10−2
−0.6 −0.4 −0.2 0 −0.6 −0.4 −0.2 0 0.2
(e,e'p)1p0π Ecal Feeddown (e,e')0π EQE Feeddown

Figure 6-15: Energy feed-down cross-sections (𝐸𝑟𝑒𝑐 − 𝐸𝑡𝑟𝑢𝑒 )/𝐸𝑡𝑟𝑢𝑒 for data (points)
and SuSav2 (lines) for 1.159 GeV (red triangles and dotted lines), 2.257 GeV (green
squares and dashed lines) and 4.453 GeV (blue dots and solid lines) on iron for (c)
𝐸 𝑐𝑎𝑙 , and (d) Fe 𝐸 𝑄𝐸 .

TVs are independent of the neutrino energy and use the momentum of the detected
particles transverse to the incident lepton [132, 223, 224] as shown in equation 6.12,
′
where 𝑃⃗𝑇𝑒 and 𝑃⃗𝑇𝑝 are the three-momenta of the detected lepton and proton perpen-
dicular to the direction of the incident lepton, respectively. The 𝑃⃗𝑇 vector is intended
to characterize the nuclear ground state, 𝛿𝛼𝑇 the FSI and ∆𝜑𝑇 is intended to probe
regions where MEC events dominate [132, 223, 224].

′
𝑃⃗𝑇 = 𝑃⃗𝑇𝑒 + 𝑃⃗𝑇𝑝 (6.10)
′
𝑃⃗ 𝑒 · 𝑃⃗𝑇
𝛿𝛼𝑇 = arccos(− 𝑇 𝑒′ ) (6.11)
𝑃𝑇 𝑃𝑇
′
𝑃⃗𝑇𝑒 · 𝑃⃗𝑇𝑝
𝛿𝜑𝑇 = arccos(− 𝑒′ 𝑝 ) (6.12)
𝑃𝑇 𝑃𝑇

Purely QE events without final state interactions, where the lepton scattered
from a bound moving proton, will have small 𝑃𝑇 , consistent with the motion of the
struck nucleon. Events with small 𝑃𝑇 should thus reconstruct to the correct incident
energy. Non-QE events, where neutral or sub-detection-threshold charged particles
were produced, will have larger 𝑃𝑇 and will not reconstruct to the correct incident
energy. 𝑃𝑇 is thus an ideal observable for tuning reaction models to ensure they

145
correctly account for non-QE processes.

The 𝑃𝑇 distribution for 2.257 GeV C(𝑒, 𝑒′ 𝑝)1𝑝0𝜋 is shown in figure 6-16 and the
other targets and energies are shown in figure 6-17. Both data and eGENIE peak
at relatively low momenta, as expected, and both have a large tail extending out to
1 GeV/𝑐 and containing about half of the measured events. The high-𝑃𝑇 tail is pre-
dominantly due to resonance production that did not result in an additional pion or
nucleon above the detection threshold. eGENIE using SuSAv2 reproduces the shape
of the data moderately well, suggesting adequate reaction modeling, including the
contribution of non-QE processes such as resonance production. As expected, both
data and eGENIE/SuSAv2 events with 𝑃𝑇 < 200 MeV/𝑐 almost all reconstruct to
the correct incident energy. However, events with 𝑃𝑇 ≥ 400 MeV/𝑐 do not reconstruct
to the correct energy and are poorly reproduced by eGENIE. This disagreement be-
comes even more pronounced at higher energies and heavier nuclei, and indicates that
including high-𝑃𝑇 data in oscillation analyses could bias the extracted parameters.
As high-𝑃𝑇 data accounts for 25 − 50% of the measured events, care must be taken
to improve the models implemented in GENIE, so that they can reproduce the high-
𝑃𝑇 data. This will be especially true at the higher incident neutrino energies expected
for DUNE.

The opening angle 𝛿𝛼𝑇 measures the angle between 𝑃𝑇 and the transverse mo-
′
mentum transfer (⃗𝑞𝑇 = −𝑃⃗𝑇𝑒 ) in the transverse plane and is isotropic in the absence
of final state interactions. 𝛿𝜑𝑇 measures the opening angle between the detected pro-
ton momentum and the transverse momentum transfer and is forward peaked. The
𝛿𝛼𝑇 distributions become progressively less isotropic at higher energies and heavier
targets, indicating the increasing importance of FSI and of non-QE reaction mech-
anisms. GENIE agrees best with data at the lowest beam energy. At the higher
beam energies GENIE describes the relatively flat smaller angles much better than
the back-angle peak. GENIE also describes the lowest energy 𝛿𝜑𝑇 distribution. At
higher energies, GENIE overestimates the height of the forward peak, as shown in
figure 6-18.

In line with the neutrino-based MicrobooNE analysis presented in section 3.2.4,

146
 2 (a)
0 < PT < 200 [MeV/c]
0.08
0.06
1 0.04
0.02
2.257 GeV
Data 0
0.5 SuSav2 (Total) 0.2 200 < PT < 400 [MeV/c] (b)

cal GeV
QE MEC

µb
0.4
dPT GeV/c

RES DIS
µb

0.3 G2018
0.1

dσ
dE
dσ

0.2
0
0.1 PT > 400 [MeV/c] (c)
0 0.05
0 0.2 0.4 0.6 0.8 1
C(e,e'p) PT [GeV/c]
1p0π

0
1 1.5 2
(e,e'p)1p0π Ecal [GeV]

Figure 6-16: (Left) the 2.257 GeV C(𝑒, 𝑒′ 𝑝)1𝑝0𝜋 cross section plotted versus missing
transverse momentum, 𝑃𝑇 , for data (black points), SuSav2 (black solid line) and
G2018 (black dashed line). The vertical lines at 200 MeV/𝑐 and at 400 MeV/𝑐 separate
the three bins in 𝑃𝑇 . Colored lines show the contributions of different processes to the
SuSAv2 GENIE simulation: QE (blue), MEC (red), RES (green) and DIS (orange).
(Right) The cross section plotted versus the calorimetric energy 𝐸𝑐𝑎𝑙 for different bins
in 𝑃𝑇 : (top) 𝑃𝑇 < 200 MeV/𝑐, (middle) 200 MeV/𝑐 ≤ 𝑃𝑇 ≤ 400 MeV/𝑐, and (bottom)
𝑃𝑇 > 400 MeV/𝑐.

these kinematic variables were further investigated in the form of multidimensional

cross sections. Figure 6-19 shows the data-simulation cross-section comparisons for
12
C at 2.261 GeV as a function of 𝑃𝑇 for (top) all the events, (bottom left) events with
𝛿𝛼𝑇 < 45𝑜 dominated by QE interactions and no reinteractions, and (bottom right)
events with 135𝑜 < 𝛿𝛼𝑇 < 180𝑜 maximally affected by FSI and multi-hadron channels.
Using all the events (top) that satisfy our selection yielded a QE-rich region up to
≈ 300 MeV/c and a RES-dominated tail that extended to ≈ 1 GeV/c. Slicing the
available 𝑃𝑇 phase-space in 𝛿𝛼𝑇 regions revealed regions with specific features. More
precisely, the region with 𝛿𝛼𝑇 < 45𝑜 can be used to isolate primarily QE events and
to test nuclear models in event generators. Even in this region though, differences
of 20-30% in the QE strength are observed by both model configurations used for

147
 1.159 GeV (a) 2.257 GeV (b) 4.453 GeV (x4) (c)
0.8 0.8 0.8

dPT GeV/c
µb
0.6 0.6 0.6
12
C dσ
0.4 0.4 0.4

0.2 0.2 0.2

0 0 0
0 0.2 0.4 0.6 0.8 01 0.2 0.4 0.6 0.8 01 0.2 0.4 0.6 0.8 1
2
(d) 2
(e)
Data

dPT GeV/c
1.5 1.5

µb
56
SuSav2 (Total) Fe
QE MEC 1 1

dσ
RES DIS 0.5 0.5
G2018 0 0
0 0.2 0.4 0.6 0.8 01 0.2 0.4 0.6 0.8 1

(e,e'p) PT [GeV/c]
1p0π

Figure 6-17: The cross section plotted vs transverse missing momentum 𝑃𝑇 for data
(black points), SuSAv2 (black solid curve) and G2018 (black dotted curve). Different
panels show results for different beam energy and target nucleus combinations: (top
row) Carbon target at (left to right) 1.159, 2.257 and 4.453 GeV, and (bottom) Iron
target at (left) 2.257 and (right) 4.453 GeV. The 4.453 GeV yields have been scaled by
four to have the same vertical scale. Colored lines show the contributions of different
processes to the SuSAv2 GENIE simulation: QE (blue), MEC (red), RES (green) and
DIS (orange).

comparison in this analysis. Meanwhile, the 135𝑜 < 𝛿𝛼𝑇 < 180𝑜 region is dominated
by RES interactions and could be used to tune FSI parameters and to improve the
RES modeling.
Figure 6-20 shows the data-simulation comparisons for 12
C at 2.261 GeV as a
function of 𝛿𝛼𝑇 for (left) all the events, (middle) events with 𝑃𝑇 < 0.2 GeV/c, and
(right) events with 𝑃𝑇 > 0.4 GeV/c. The 𝑃𝑇 < 0.2 GeV/c slice is dominated by QE
events with minimal FSI effects that result in a fairly uniform distribution with a
slight enhancement in the forward direction due to the RES contamination. The
𝑃𝑇 > 0.4 GeV/c slice is dominated by multi-hadron and enhanced-FSI events that
result in a sharp peak in the region close to 150𝑜 . Therefore, this investigation of the
𝛿𝛼𝑇 phase-space in slices of 𝑃𝑇 is complimentary to the one illustrated in figure 6-19.
Using the exact same 𝑒4𝜈 data sets and cross-section extraction technique, the

148
 3 1.159 GeV (a) 3 2.257 GeV (b) 3 4.453 GeV (x2) (c)

dαT deg
2 2 2

dσ nb
12
C
1 1 1

0 0 0
0 50 100 150 0 50 100 150 0 50 100 150
(d) (e)
Data 10 10

dαT deg
dσ nb
SuSav2 (Total) 56
Fe
QE MEC 5 5
RES DIS
G2018 0 0
0 50 100 150 0 50 100 150

(e,e'p) δαT [deg]

1p0π

8 (f) 8 (g) 8 (x2) (h)

6 6 6
dφT deg
dσ nb

12
C 4 4 4

2 2 2

0 0 0
0 50 100 150 0 50 100 150 0 50 100 150
20
(i) 20
(j)

15 15
dφT deg
dσ nb

56
Fe 10 10

5 5

0 0
0 50 100 150 0 50 100 150

(e,e'p) δφ [deg]
1p0π T

Figure 6-18: The cross section plotted vs 𝛿𝛼𝑇 (a-e) and vs 𝛿𝜑𝑇 (f-j) for data (black
points), SuSAv2 (black solid curve) and G2018 (black dotted curve). Different panels
show results for different beam energy and target nucleus combinations: (top row)
Carbon target at (left to right) 1.159, 2.257 and 4.453 GeV, and (bottom) Iron target
at (left) 2.257 and (right) 4.453 GeV. The 4.453 GeV yields have been scaled by two
to have the same vertical scale. Colored lines show the contributions of different
processes to the SuSAv2 GENIE simulation: QE (blue), MEC (red), RES (green)
and DIS (orange).

cross section as a function of the total struck nucleon momentum approximation,

derived following the approach in section 3.2.1, was reported. The only difference
compared to the already outlined formalism on 40
Ar is that B = 0.09216 GeV is the

149
 12C @ 2.257 GeV, All events

Data
0.5 SuSav2 (Total)
QE MEC
0.4

dPT GeV/c
RES DIS

µb
G2018
0.3

dσ
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1
(e,e'p) PT [GeV/c]
1p0π

12C @2.257 GeV, δαT < 45o 12C @2.257 GeV, 135o < δαT < 180o
×10−3 ×10−3
4
dδαTdPT deg GeV/c

dδαTdPT deg GeV/c

2
µb

µb
3
1.5
2
1
d2σ

d2σ
0.5 1

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
(e,e'p) PT [GeV/c] (e,e'p) PT [GeV/c]
1p0π 1p0π

Figure 6-19: Data-simulation comparisons for 12 C at 2.261 GeV showing the cross
section results as a function of 𝑃𝑇 for (top) all the events, (bottom left) events with
𝛿𝛼𝑇 < 45𝑜 dominated by QE interactions and no reinteractions, and (bottom right)
events with 135𝑜 < 𝛿𝛼𝑇 < 180𝑜 maximally affected by FSI and multi-hadron channels.
Colored lines show the contributions of different processes to the SuSAv2 GENIE
simulation: QE (blue), MEC (red), RES (green) and DIS (orange).

binding energy for 12 C, and 𝜖𝑁 = 0.024 GeV is the corresponding removal energy [225].
This approximation, refered to as 𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 , is compared to the true missing momentum
𝑃𝑀 𝑖𝑠𝑠 calculated as shown in equation 6.13,

𝑃𝑀 𝑖𝑠𝑠 = |⃗𝑞 − 𝑝⃗|, (6.13)

where ⃗𝑞 is the 3-vector for the momentum transfer based on the difference between
the kinematics of the incoming and the outgoing lepton, and 𝑝⃗ is the 3-vector of the
outgoing proton.
Figure 6-21 shows the data-simulation comparisons for 12
C at 2.261 GeV as a
function of (top left) the true missing momentum 𝑃𝑀 𝑖𝑠𝑠 and (top right) the missing

150
 12C @ 2.257 GeV, All events
×10−3
2.5 Data
SuSav2 (Total)
2 QE MEC
RES DIS

dδαT deg
dσ µ b
1.5 G2018

0.5

0
0 50 100 150
(e,e'p) δαT [deg]
1p0π

×10−3 ×10−3
12C @2.257 GeV, PT < 0.2 GeV/c 12C @2.257 GeV, PT > 0.4 GeV/c

2
dδαTdPT deg GeV/c

dδαTdPT deg GeV/c

3
µb

µb
1.5
2
1
d2σ

1 d2σ
0.5

0 0
0 50 100 150 0 50 100 150
(e,e'p)1p0π δαT [deg] (e,e'p)1p0π δαT [deg]

Figure 6-20: Data-simulation comparisons for 12 C at 2.261 GeV showing the cross
section results as a function of 𝛿𝛼𝑇 for (top) all the events, (bottom left) events
with 𝑃𝑇 < 0.2 GeV/c dominated by QE interactions and no reinteractions, and (bot-
tom right) events with 𝑃𝑇 > 0.4 GeV/c maximally affected by FSI and multi-hadron
channels. Colored lines show the contributions of different processes to the SuSAv2
GENIE simulation: QE (blue), MEC (red), RES (green) and DIS (orange).

momentum approximation 𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 commonly used by neutrino experiments. The bot-

tom panel shows the overlay of the two extracted data cross sections illustrating that
𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 fails to reproduce 𝑃𝑀 𝑖𝑠𝑠 . As can be seen in the interaction breakdown plots,
the differences are primarily driven by the RES interactions which are reconstructed
at lower values when the QE-like assumption deployed for 𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 is used.

Figures 6-22 and 6-23 also illustrate the equivalent of the multidimensional Mi-
croBooNE analysis using the 𝑃𝑇,𝑥 cross sections in slices of 𝑃𝑇,𝑦 (figure 6-22) and vice
versa (figure 6-23). As can be seen in the interaction breakdown plots, the regions
close to 0 for both kinematic variables (bottom left panels) are the ideal place to iso-
late QE events with small contributions from more complex events. Yet, even in this

151
 12C @2.257 GeV, All events 12C @2.257 GeV, All events
0.5
Data 0.5 Data
SuSav2 (Total) SuSav2 (Total)
0.4

dσ [ µ b ]
0.4

dPn,proxy GeV/c
QE MEC QE MEC
dσ [ µ b ]
dPMiss GeV/c
RES DIS RES DIS
0.3 G2018
0.3 G2018

0.2 0.2
0.1 0.1

0 0
0 0.5 1 1.5 0 0.5 1 1.5
(e,e'p) PMiss [GeV/c] (e,e'p) Pn,proxy [GeV/c]
1p0π 1p0π

12C @2.257 GeV, All events

0.25
dσ [ µ b ]
dPMiss GeV/c

0.2 PMiss
0.15 Pn,proxy

0.1
0.05
0
0 0.5 1 1.5
(e,e'p) PMiss [GeV/c]
1p0π

Figure 6-21: (Top) data-simulation comparisons on 12 C at 2.261 GeV showing the

cross section results as a function of (top left) the true missing momentum 𝑃𝑀 𝑖𝑠𝑠 and
(top right) the missing momentum approximation 𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 commonly used by neutrino
experiments. (Bottom) overlay of the two extracted data cross sections illustrating
the differences between 𝑃𝑀 𝑖𝑠𝑠 and 𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 .

very QE-dominated region, significant data-simulation differences are observed, most

likely due to the already-observed RES overestimation. On the other hand, events
occupying the phase-space outside that QE-rich band (bottom right) are dominated
primarily by most complicated interactions (RES and DIS) and QE events with strong
FSI effects, as indicated by the relevant pointy shapes. These populations result in
much more broader and smeared distributions.

Of particular interest is the asymmetric behavior observed in figure 6-23. As

discussed in section 3.2.1, this clearly-observed asymmetry is caused by multi-nucleon
and FSI effects. More precisely, the asymmetry in the QE-enhanced region (bottom
left) is entirely driven by the overpredicted RES events. In the non-QE-dominated
slice (bottom right), data-simulation disagreements by a factor of ∼ 2 are further

152
 12C @2.257 GeV, All events
0.5
Data
SuSav2 (Total)
0.4 QE MEC

dσ [ µ b ]
dPT,x GeV/c
RES DIS
0.3 G2018

0.2

0.1

0
−0.5 0 0.5
(e,e'p) PT,x [GeV/c]
1p0π

×10−3
12C @2.257 GeV, -0.15 < PT,y < 0.15 GeV/c 12C @2.257 GeV, PT,y < -0.15 GeV/c

25
1
dPT,xdPT,y GeV2/c2

dPT,xdPT,y GeV2/c2
20
µb

µb
0.8
15
0.6
10
d 2σ

0.4 d 2σ
0.2 5

0 0
−0.5 0 0.5 −0.5 0 0.5
(e,e'p) PT,x [GeV/c] (e,e'p) PT,x [GeV/c]
1p0π 1p0π

Figure 6-22: Data-simulation comparisons for 12 C at 2.261 GeV showing the cross
section results as a function of 𝑃𝑇,𝑥 for (top) all the events, (bottom left) events
with -0.15 < 𝑃𝑇,𝑦 < 0.15 GeV/c dominated by QE interactions and no reinteractions,
and (bottom right) events with 𝑃𝑇,𝑦 < -0.15 GeV/c maximally affected by FSI and
multi-hadron channels.

observed and are driven by RES/DIS events and FSI-enhanced QE interactions.

6.4 Electrons-For-Neutrinos Conclusions

In this electron-based analysis, the similarities between electron- and neutrino-nucleus

interactions were exploited, and electron scattering data with known beam energies
to test energy reconstruction methods and interaction models were used [208]. Even
in simple interactions where no pions are detected, only a small fraction of events was
found to reconstruct to the correct incident energy. More importantly, widely-used
interaction models reproduced the reconstructed energy distribution only qualita-
tively and the quality of the reproduction varied strongly with beam energy. This

153
 12C @2.257 GeV, All events

Data
0.4 SuSav2 (Total)
QE MEC

dσ [ µ b ]
dPT,y GeV/c
RES DIS
0.3 G2018

0.2

0.1

0
−1.5 −1 −0.5 0
(e,e'p) PT,y [GeV/c]
1p0π

×10−3
12C @2.257 GeV, -0.15 < PT,x < 0.15 GeV/c 12C @2.257 GeV, PT,x < -0.15 GeV/c

10
dPT,xdPT,y GeV2/c2

dPT,xdPT,y GeV2/c2
1
8
µb

µb
6
0.5 4
d 2σ

d 2σ
2

0 0
−0.5 0 0.5 −0.5 0 0.5
(e,e'p) PT,y [GeV/c] (e,e'p) PT,y [GeV/c]
1p0π 1p0π

Figure 6-23: Data-simulation comparisons for 12 C at 2.261 GeV showing the cross
section results as a function of 𝑃𝑇,𝑦 for (top) all the events, (bottom left) events
with -0.15 < 𝑃𝑇,𝑥 < 0.15 GeV/c dominated by QE interactions and no reinteractions,
and (bottom right) events with 𝑃𝑇,𝑥 < -0.15 GeV/c maximally affected by FSI and
multi-hadron channels.

shows both the need and the pathway to improve current models to meet the re-
quirements of next-generation, high-precision experiments such as DUNE [146] and
Hyper-Kamiokande (HK) [226].

6.5 Prospects With CLAS12

Jefferson Lab Experiment E12-17-006, “Electrons for Neutrinos: Addressing Critical

Neutrino–Nucleus Issues” (scientific rating: A) [227] has already taken further data on
more targets with a greater kinematical range using the upgraded CLAS12 detector
shown in figure 6-24. The approved experiment includes measurements on 4 He, C, Ar
and Sn with 1-, 2-, 4-, and 6-GeV electron beams, as well as measurements on O with

154
1- and 2-GeV electron beams. The 1- and 2-GeV measurements will be performed
with a minimum electron scattering angle of 5𝑜 , compared to a minimum CLAS angle
of about 15𝑜 . This will extend the measurements down to the much lower momentum
transfers, typical of some neutrino experiments, and to multi-hadron topologies. It
will therefore allow comparisons with the lower beam-energy data of T2K and HK.
The first part of the experiment ran in the second half of 2021 and the second one in
the beginning of 2022. The majority of the 2022 data sets have already been collected,
with the exception of the 1 GeV and 16
O data sets.

Cherenkov
Time-of-Flight
Tracker Calorimeter

Target

e– beam

BAND

Figure 6-24: Schematic view of the upgraded CLAS12 detector components.

155
156
Chapter 7

Summary

The findings presented in this thesis report on both neutrino and electron cross-
section modeling and analysis results in order to critically improve the understanding
of lepton-nucleus interactions. This knowledge will be used to significantly reduce the
cross-section related uncertainties of forthcoming experiments that aim to extract the
neutrino oscillation parameters with high-accuracy measurements.
The outlined work took a significant step towards this high-precision era with the
use of neutrino data sets from the MicroBooNE liquid argon time projection chamber
detector at Fermi National Laboratory. The reported neutrino analyses isolated parts
of the phase space where significant model improvements are required. Furthermore,
valuable kinematic variables were identified and established as tools to probe specific
nuclear effects with multi-differential measurements.
This thesis further realized the connections between electron and neutrino inter-
actions. This realization resulted in significant improvements of the modeling used
in neutrino oscillation analyses. Yet, these improved predictions failed to reproduce
exclusive electron scattering results from the CLAS detector at Thomas Jefferson
Laboratory using high-statistics data sets and monoenergetic beams. However, such
comparisons against electron scattering data sets can definitively constrain the vector
part and the nuclear effects in lepton-nucleus interactions in event generators that
will be used for the forthcoming oscillation analyses.

157
158
Chapter 8

Appendices

8.1 Total Struck Nucleon Momentum Derivation

The total momentum of the struck nucleon can be obtained following the derivation
outlined below for two-body interactions. We followed the approach introduced by
the Minerva collaboration to reconstruct the longitudinal and total nucleon momenta
in [132]. We focused on quasielastic-like (QE-like) CC1p0𝜋 processes,

𝜈𝜇 + 𝐴 → 𝜇− + 𝑝 + (𝐴 − 1), (8.1)

while using the formalism detailed in [228] for the nuclear masses,

𝑚𝐴 = 22 × 𝑀𝑛 + 18 × 𝑀𝑝 − 𝐵 [𝐺𝑒𝑉 ]
(8.2)
𝑚𝐴−1 = 𝑚𝐴 − 𝑀𝑛 + 𝜖𝑁 [𝐺𝑒𝑉 ].

Here 𝑀𝑝 and 𝑀𝑛 denote the proton and neutron masses, respectively, B = 0.34381
GeV is the argon binding energy (obtained from page 3 in [228] for 40 nucleons with
an average binding energy of 9 MeV) and 𝜖𝑁 = 0.0309 GeV is the removal energy,

159
with

𝜖𝑁 = 𝑆 𝑁 + 𝐸𝑥𝑁 + ⟨𝑇𝐴−1 ⟩ (8.3)

where 𝑆 𝑁 = 9.9 MeV is the neutron separation energy obtained for argon from
table 7 in [225], 𝐸𝑥𝑁 is the excitation energy, and ⟨𝑇𝐴−1 ⟩ is the average kinetic energy
of the remnant system, which is negligible.
In equation 8.1, the incident neutrino energy 𝐸𝜈 is unknown, but the dependence
of 𝛿⃗𝑝 (struck nucleon momentum before the interaction) on 𝐸𝜈 can be removed under
a QE-like approximation. This was achieved with the process detailed below.
First, we decomposed 𝛿⃗𝑝 into longitudinal and transverse components with respect
to the neutrino direction, used energy/momentum conservation equations and that
𝑝𝜈 = 𝐸 𝜈 ,

𝛿⃗𝑝 ≡ (𝛿⃗𝑝𝑇 , 𝛿𝑝𝐿 ) (8.4)

𝑝⃗𝜈 = 𝑝⃗𝜇 + 𝑝⃗𝑝 + 𝑝⃗𝐴−1 (8.5)

𝐸𝜈 + 𝛿𝑝𝐿 = 𝑝𝜇𝐿 + 𝑝𝑝𝐿 (8.6)

𝛿⃗𝑝𝑇 = 𝑝⃗𝜇𝑇 + 𝑝⃗𝑝𝐿 (8.7)

𝐸𝜈 + 𝑚𝐴 = 𝐸 𝜇 + 𝐸 𝑝 + 𝐸𝐴−1 (8.8)

where 𝑝⃗𝜇 and 𝑝⃗𝑝 are the muon and proton momenta, respectively, 𝐸𝜇 = 𝑝2𝜇 + 𝑚2𝜇
√︀

for the muon candidate and 𝐸𝑝 = 𝑝2𝑝 + 𝑚2𝑝 for the proton candidate, where the
√︀

corresponding momenta are obtained based on the particles’ ranges [137], and 𝐸𝐴−1
is the energy of remnant nuclear system.
Under the assumption that no final state interactions (FSI) take place,

𝑝⃗𝜈 + 𝛿⃗𝑝 = 𝑝⃗𝜇 + 𝑝⃗𝑝 . (8.9)

160
 If we combine equations 8.5 and 8.9, 𝛿⃗𝑝 gives the magnitude of its recoil momen-
tum, 𝛿⃗𝑝 = −⃗𝑝𝐴−1 and 𝛿𝑝 = 𝑝𝐴−1 . Combining equations 8.6 and 8.8 to eliminate 𝐸𝜈
yields

𝛿𝑝𝐿 = 𝑚𝐴 + 𝑝𝜇𝐿 + 𝑝𝑝𝐿 − 𝐸 𝜇 − 𝐸 𝑝 − 𝐸𝐴−1 . (8.10)

√︁
Using the fact that 𝐸𝐴−1 = 𝑚2𝐴−1 + 𝑝2𝐴−1 , 𝛿𝑝 = 𝑝𝐴−1 and the decomposition of
𝛿⃗𝑝 into longitudinal and transverse components in equation 8.4,

√︁
𝛿𝑝𝐿 = 𝑚𝐴 + 𝑝𝜇𝐿 + 𝑝𝑝𝐿 − 𝐸 𝜇 − 𝐸 𝑝 − 𝑚2𝐴−1 + 𝛿𝑝2𝑇 + 𝛿𝑝2𝐿 (8.11)

For simplicity, we defined

𝑅 ≡ 𝑚𝐴 + 𝑝𝜇𝐿 + 𝑝𝑝𝐿 − 𝐸 𝜇 − 𝐸 𝑝 (8.12)

Using R to simplify equation 8.11,

√︁
𝛿𝑝𝐿 = 𝑅 − 𝑚2𝐴−1 + 𝛿𝑝2𝑇 + 𝛿𝑝2𝐿 (8.13)

Rearranging the terms, squaring each side and solving for 𝛿𝑝𝐿 yields

1 𝑚2𝐴−1 + 𝛿𝑝2𝑇
𝛿𝑝𝐿 = 𝑅 − . (8.14)
2 2𝑅

Finally, combining the longitudinal and the transverse components, we obtain the
total struck nucleon momentum. Given that this momentum is an approximation
following the procedure and the assumptions mentioned above (QE-like scattering

161
and no FSI), we will be referring to it as 𝑝𝑛,𝑝𝑟𝑜𝑥𝑦 as opposed to 𝛿𝑝, where

√︁
𝑝𝑛,𝑝𝑟𝑜𝑥𝑦 = 𝛿𝑝2𝐿 + 𝛿𝑝2𝑇 (8.15)

162
8.2 Wiener SVD Regularization Technique

The procedure detailed below relies on [139]. We choose to work with a 𝜒2 (𝑠) metric.

𝜒2 (𝑠) = (m − 𝑟 · 𝑠)𝑇 𝐶𝑜𝑣 −1 (m − 𝑟 · 𝑠) (8.16)

Dimension
Notation Explanation
And Format
m measured spectrum in data (signal and background events) m × 1 vector
s free variable in 𝜒 (𝑠) function
2
n × 1 vector
𝑠ˆ estimator of true signal, obtained after minimizing 𝜒2 n × 1 vector
𝑠𝑡𝑟𝑢𝑒 true signal n × 1 vector
𝑠 nominal MC signal prediction n × 1 vector
r response matrix m × n matrix
symmetric covariance matrix
Cov m × m matrix
with statistical and systematic uncertainties

Note that we use 𝑠𝑡𝑟𝑢𝑒 to represent the true signal in order to differentiate from s
which is a variable in the function 𝜒2 (𝑠). We use 𝑠ˆ to represent the estimator of the
true signal 𝑠𝑡𝑟𝑢𝑒 , which is obtained after minimizing the 𝜒2 (𝑠) function. We further
restrict ourselves to the m ≥ n case.

Our objective is to minimize 𝜒2 (𝑠). Since the covariance matrix Cov is symmetric,
the inverse of it 𝐶𝑜𝑣 −1 is also symmetric. Hence, 𝐶𝑜𝑣 −1 can be decomposed with
Cholesky decomposition [229] into

𝐶𝑜𝑣 −1 = 𝑄𝑇 · 𝑄 (8.17)

where Q is a uniquely defined lower triangular matrix and 𝑄𝑇 is its transpose.

163
We further define

𝑀 ≡𝑄·m

𝑅≡𝑄·𝑟 (8.18)

(𝑝𝑟𝑒 − 𝑠𝑐𝑎𝑙𝑖𝑛𝑔)

Dimension
Notation Explanation
And Format
M measured spectrum after pre-scaling m × 1 vector
R response matrix after pre-scaling m × n matrix
Q Lower triangular matrix from Cholesky decomposition of 𝐶𝑜𝑣 −1 m × m matrix
𝑀 expectation spectrum after pre-scaling (R · 𝑠) m × 1 vector

Replacing the definitions in equation 8.18 into equation 8.16 yields

∑︁ ∑︁
𝜒2 (𝑠) = (𝑀 − 𝑅 · 𝑠)𝑇 (𝑀 − 𝑅 · 𝑠) = (𝑀𝑖 − 𝑅𝑖𝑗 · 𝑠𝑗 )2 (8.19)
𝑖 𝑗

Ideally, we want to achieve 𝜒2 (𝑠) = 0. That is obtained when

𝑀 − 𝑅 · 𝑠 = 0 ⇒ 𝑀 = 𝑅 · 𝑠 ⇒ 𝑅𝑇 · 𝑀 = 𝑅𝑇 · 𝑅 · 𝑠 (8.20)

We use the fact that 𝑅𝑇 · 𝑅 is a square n×n invertable matrix. Thus, the exact
solution 𝑠ˆ is uniquely defined as

𝑠ˆ = (𝑅𝑇 · 𝑅)−1 · 𝑅𝑇 · 𝑀 (8.21)

We decompose the measured spectrum into the signal (R · s) part and the noise
/ background (N).

𝑀 = 𝑅 · 𝑠𝑡𝑟𝑢𝑒 + 𝑁 (8.22)

164
 Substituting equation 8.22 into equation 8.21 yields

𝑠ˆ = (𝑅𝑇 · 𝑅)−1 · 𝑅𝑇 · (𝑅 · 𝑠𝑡𝑟𝑢𝑒 + 𝑁 ) (8.23)

with N representing the “noise” coming from uncertainties (statistical and system-
atic uncertainties associated with both m and r). Each term in the noise vector after
pre-scaling follows a normal distribution with 𝜇 = 0 and 𝜎 = 1, since the denominator
of the 𝜒2 function in equation 8.19 (i.e. square of error) is unity. Given the fact that
each term in the noise vector is independent (i.e. uncorrelated), we refer to the basis
in this domain as orthogonal.

The response matrix after pre-scaling (R) can be decomposed, using the singular
value decomposition (SVD) approach [230], as

𝑅 =𝑈 ·𝐷·𝑉𝑇 (8.24)

with both 𝑈𝑚×𝑚 and 𝑉𝑛×𝑛 being orthogonal matrices that satisfy 𝑈 𝑇 · 𝑈 = 𝑈 · 𝑈 𝑇
= 𝐼𝑚×𝑚 and 𝑉 𝑇 · 𝑉 = 𝑉 · 𝑉 𝑇 = 𝐼𝑛×𝑛 , with U,V and D being uniquely defined. I is the
identity matrix and the subscript represents the dimension. D is an m×n diagonal
matrix with positive definite diagonal elements (known as singular values) 𝐷𝑖𝑖 = 𝑑𝑖
arranged in descending order as i increases.

Dimension
Notation Explanation
And Format
𝑉𝑇 right orthogonal matrix from decomposition of R n × n vector
D diagonal matrix from decomposition of R m × n matrix
U left orthogonal matrix from decomposition of R m × m matrix

165
 Inserting equation 8.24 into 8.23, we have

𝑠ˆ = 𝑉 · 𝐷−1 · 𝑈 𝑇 · (𝑅 · 𝑠𝑡𝑟𝑢𝑒 + 𝑁 )

= 𝑉 · 𝐷−1 · (𝑅𝑈 · 𝑠𝑡𝑟𝑢𝑒 + 𝑁𝑈 ) (8.25)

= 𝑉 · 𝐷−1 · 𝑀𝑈

where 𝑅𝑈 ≡ 𝑈 𝑇 · 𝑅, 𝑁𝑈 ≡ 𝑈 𝑇 · 𝑁 , and 𝑀𝑈 ≡ 𝑈 𝑇 · 𝑀 are transformations of the

smearing matrix R, the noise N, and the measured signal M, respectively. Note that,
since U is an orthogonal matrix and the elements of the original noise vector N are
uncorrelated, the elements of the new noise vector 𝑁𝑈 are still uncorrelated. Each
element follows a normal distribution with 𝜇 = 0 and 𝜎 = 1. Thus, the basis in this
new domain is still orthogonal.
However, there are cases where the unbiased solution to an unfolding problem
via a direct inversion can have catastrophic oscillations via the introduction of huge
variances.
A proposed solution is the introduction of a trade-off term between the bias and
the variance to suppress the oscillations. We refer to that term as “regularization” and
is introduced in the form of an n×n matrix 𝐹 that is applied on the exact solution 𝑠ˆ.

𝑠ˆ = 𝐹 · 𝑉 · 𝐷−1 · (𝑅𝑈 · 𝑠𝑡𝑟𝑢𝑒 + 𝑁𝑈 )

= 𝐹 · 𝑉 · 𝐷−1 · 𝑈 𝑇 · 𝑀 (8.26)

= 𝐹 · 𝑉 · 𝐷−1 · 𝑀𝑈

Using equation 8.26 (focusing on the diagonal elements ii, using the fact that V
is orthogonal 𝑉 · 𝑉 𝑇 = 𝐼, and that D is diagonal 𝐷𝑖𝑖−1 = 1/𝑑𝑖 ),

𝑀𝑈,𝑖
(𝑉 𝑇 𝑠ˆ)𝑖 = 𝐹𝑖𝑖 ·
𝑑𝑖
(8.27)
ˆ 𝑀 (𝜔)
𝑆(𝜔) = 𝐹 (𝜔) ·
𝑅(𝜔)

166
 It is easy to see the similarities between the two lines in equation 8.27. Therefore,
following the same terminology as that for the signal processing, we refer to 𝑀𝑈 after
the SVD transformation as the measurement in the effective frequency domain in
analogy to the frequency domain in the signal processing.

The additional matrix F can be decomposed as

𝐹 = 𝑉 · 𝑊 · 𝑉 𝑇. (8.28)

At this point, the F and W matrices are still unknown.

Plugging equation 8.28 into equation 8.26 yields

𝑠ˆ = 𝑉 · 𝑊 · 𝐷−1 · (𝑅𝑈 · 𝑠𝑡𝑟𝑢𝑒 + 𝑁𝑈 ). (8.29)

We consider the expectation value of the signal in the effective frequency domain:

𝑀𝑈 = 𝑈 𝑇 · 𝑀 = 𝑈 𝑇 · 𝑅 · 𝑠 (8.30)

In general, the 𝑠𝑡𝑟𝑢𝑒 is unknown, so the expectation signal 𝑠 using the nominal
simulation prediction is used.
𝑅2 (𝜔)·𝑆 2 (𝜔)
The construction of W is based on the Wiener filter 𝑅2 (𝜔)·𝑆 2 (𝜔)+𝑁 2 (𝜔)
. Taking
equation 8.30 at bin i, we have

∑︁
2
⟨𝑅2 · 𝑆 2 ⟩ = 𝑀𝑈,𝑖 =𝐸𝑞. 8.27
𝑑2𝑖 · ( 𝑉𝑖𝑗𝑇 · 𝑠𝑗 )2 (8.31)
𝑗

⟨𝑁 2 ⟩ = 1 (8.32)

resulting in a Wiener filter of

𝑑2𝑖 · ( 𝑗 𝑉𝑖𝑗𝑇 · 𝑠𝑗 )2
∑︀
𝑊𝑖𝑘 = 2 ∑︀ 𝑇 · 𝛿𝑖𝑘 (8.33)
𝑑𝑖 · ( 𝑗 𝑉𝑖𝑗 · 𝑠𝑗 )2 + 1

At this point, W is uniquely defined. Here, equation 8.32 is obtained, since each

167
element of noise 𝑁𝑈 follows a normal distribution with 𝜇 = 0 and 𝜎 = 1. We have

𝑑𝑖 · ( 𝑗 𝑉𝑖𝑗𝑇 · 𝑠𝑗 )2
∑︀
−1
(𝑊 · 𝐷 )𝑖𝑗 = 2 ∑︀ 𝑇 · 𝛿𝑖𝑘 (8.34)
𝑑𝑖 · ( 𝑗 𝑉𝑖𝑗 · 𝑠𝑗 )2 + 1
The small value of 𝑑𝑖 is balanced by the finite value of the expectation value of
𝑁 2 ≡ 1 and, thus, equation 8.29 doesn’t suffer from catastrophic oscillations. From
equation 8.33, the construction of the Wiener filter takes into account the strengths
of both the signal and noise expectations and is independent of the regularization
strength 𝜏 used in traditional regularization techniques [231].
As shown in [231], the regularization can be applied on the curvature of the
spectrum instead of the strength of the spectrum, which involves an additional matrix
C. This is also be achieved in the Wiener-SVD approach:

𝑀 = 𝑅 · 𝐶 −1 · 𝐶 · 𝑠 (8.35)

by including an additional matrix C that has the commonly used regularization

forms, such as the first and second order derivatives. Since the effective frequency
domain is determined by the smearing matrix R, the inclusion of C would alter the
basis of the effective frequency domain. In this case, the SVD decomposition becomes
(just like shown in equation 8.24)

𝑅 · 𝐶 −1 = 𝑈𝐶 · 𝐷𝐶 · 𝑉𝐶𝑇 . (8.36)

Dimension
Notation Explanation
And Format
𝑉𝐶𝑇 right orthogonal matrix from decomposition of R ·𝐶 −1 n × n vector
𝐷𝐶 diagonal matrix from decomposition of R ·𝐶 −1
m × n matrix
𝑈𝐶 left orthogonal matrix from decomposition of R ·𝐶 −1 m × m matrix

The final solution of the regularisation becomes (just like shown in equation 8.25

168
by inserting it into equation 8.23)

𝑠ˆ = 𝐶 −1 · 𝑉𝐶 · 𝑊𝐶 · 𝑉𝐶𝑇 · 𝐶 · (𝑅𝑇 𝑅)−1 · 𝑅𝑇 · 𝑀 (8.37)

or, equivalently,
𝑠ˆ = 𝐴𝐶 · (𝑅𝑇 𝑅)−1 · 𝑅𝑇 · 𝑀 (8.38)

where

𝐴𝐶 = 𝐶 −1 · 𝑉𝐶 · 𝑊𝐶 · 𝑉𝐶𝑇 · 𝐶 (8.39)

The corresponding Wiener filter would be (once again, by realizing the similarities
between the two expressions in equation 8.27 and by using equation 8.39)

𝑑2𝐶𝑖 · ( 𝑗 𝑉𝐶𝑖𝑗
∑︀ 𝑇
· ( 𝑙 𝐶𝑗𝑙 𝑠𝑙 ))2
∑︀
𝑊𝐶,𝑖𝑖 = 2 ∑︀ 𝑇 ∑︀ (8.40)
𝑑𝐶𝑖 · ( 𝑗 𝑉𝐶𝑖𝑗 · ( 𝑙 𝐶𝑗𝑙 𝑠𝑙 ))2 + 1

where 𝐶𝑗𝑙 , 𝑉𝐶𝑖𝑗

𝑇
, and 𝑑𝐶𝑖 are matrix elements of matrices C, 𝑉𝐶 , and 𝐷𝐶 , respec-
tively.
Since the unfolded results are a linear transformation of the measurement, we
can easily evaluate the uncertainties associated with them. Equation 8.38 can be
rewritten as

𝑠ˆ = 𝑅𝑡𝑜𝑡 · 𝑚 (8.41)

with

𝑅𝑡𝑜𝑡 = 𝐴𝐶 · (𝑅𝑇 𝑅)−1 · 𝑅𝑇 · 𝑄. (8.42)

Then, the covariance matrix of 𝑠ˆ can be deduced from the covariance matrix of
m as

𝑇
𝐶𝑜𝑣𝑠^,m = 𝑅𝑡𝑜𝑡 · 𝐶𝑜𝑣m · 𝑅𝑡𝑜𝑡 . (8.43)

The variances of the unfolded data can also be easily calculated given that their

169
origin N in equation 8.37 is uniquely defined. Defining N(i) as a vector with the i-th
element being 1 and the rest of elements being 0, we can calculate the variance in s
due to i-th element in N as:

𝑇𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 (𝑖) = 𝐴𝐶 · (𝑅𝑇 𝑅)−1 · 𝑅𝑇 · 𝑁 (𝑖), (8.44)

with 𝑇𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛 (i) being a vector. The variance of the j-th element of 𝑇𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛,𝑗 can
thus be written as:

√︃
∑︁
𝑇𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛,𝑗 = 2
𝑇𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛,𝑗 (𝑖), (8.45)
𝑖

after summing the contribution from each independent noise source. The square
of 𝑇𝑑𝑒𝑣𝑖𝑎𝑡𝑖𝑜𝑛,𝑗 corresponds to the j-th diagonal element of the covariance matrix 𝐶𝑜𝑣𝑠
in equation 8.43.
Given equation 8.37, we can understand the entire process of unfolding as to
“remove” the effect of R through multiplying (𝑅𝑇 𝑅)−1 · 𝑅𝑇 and then replace it with
a new smearing matrix 𝐴𝐶 . Therefore, it is straightforward to estimate the bias on
the unfolded results:

𝑇𝑏𝑖𝑎𝑠 = (𝐴𝐶 − 𝐼) · (𝑅𝑇 𝑅)−1 · 𝑅𝑇 · 𝑀

(8.46)
= (𝐴𝐶 − 𝐼) · 𝑠

with I being identity matrix and 𝑠 being the expectation of the nominal MC signal
prediction.

170
8.3 Electrons-For-Neutrinos Fiducials
Fiducial cuts for e2a have been defined and used by several analyses [232–234] and
follow the same general procedure for all charged particle species. A series of event
selection cuts are applied to data to produce event samples suitable for defining the
fiducial regions of CLAS. From these event samples, regions of good acceptance are
found, defined as flat regions of 𝜑 space in bins of momentum and 𝜃, which are then
parameterised into functions. The specific procedures for each particle species for
which fiducial cuts have been defined are summarised in the following sections.

(a) (b)

Figure 8-1: The 𝜃 vs 𝜑 distributions for 𝜋 − , before (a) and after (b) the application
of fiducial cuts as defined in [235] for 2.2 GeV analysis of 4 He.

Electron fiducial cuts were defined in [236], using electron candidates identified by
the CLAS triggering requirements for e2a. These electron samples are then subjected
to geometric cuts in the 𝑢, 𝑣, and 𝑤 “views” corresponding to the orientation of the
scintillator layers of the electromagnetic calorimeter. This cut accounts for poorly
understood electron acceptance at the edges of the calorimeter, and the difference in
acceptance between the electromagnetic calorimeter and the Cerenkov Counter, which
can be seen in data as a characteristic “smile” feature in the uncut 𝜃 vs 𝜑 distributions
for electrons. Additionally, a cut is applied on the ratio of energy deposition by an
electron in the calorimeter to electron momentum as measured by the drift chambers,
𝐸𝑡𝑜𝑡 /𝑝𝑒′ . This ratio is known as the ‘sampling fraction’ and is fixed by the design of
the detector to 0.3𝑐. At the detector edges, this value can decrease due to shower

171
leakage, where the electron energy is not fully deposited in the calorimeter.
The electrons distributions for each beam energy and torus field setting are then
split into sectors and momentum bins 50 𝑀 𝑒𝑉 /𝑐 wide, spanning the electron momen-
tum range of the data. A plot of 𝜃𝑒 against 𝜑𝑒 is produced for each momentum bin,
and for each degree in 𝜃𝑒 , the plot was projected onto the 𝜑𝑒 axis and fitted with a
trapezoid function.

(a) (b)

Figure 8-2: The 𝜃 vs 𝜑 distributions for electrons, before (left) and after (right) the
application of fiducial cuts for 2.2 GeV analysis of 4 He.

Positively charged hadrons, i.e. protons and 𝜋 + , have their fiducial cuts defined as
a single species, under the assumption that their identical charge means their fiducial
regions will be the same. The positive hadron fiducial cuts were defined in [237].
Event samples are identified as tracks with good drift chamber status, a good hit
in the time-of-flight system, and particle identification via a 𝜒2 cut on the DCPB
bank
In order to eliminate protons in quasi-free reactions, which are not uniformly
distributed in 𝜑, an energy transfer cut is applied
The positively charged hadron distributions for each beam energy and torus field
setting are then split into sectors and momentum bins 50 𝑀 𝑒𝑉 /𝑐 wide, spanning the
momentum range of the data. A plot of 𝜃 against 𝜑 is produced for each momentum
bin, and for each degree in 𝜃, the plot was projected onto the 𝜑 axis and fitted with
a trapezoid function.

172
 (a) (b)

Figure 8-3: The 𝜃 vs 𝜑 distributions for protons, before (left) and after (right) the
application of fiducial cuts for 2.2 GeV analysis of 4 He.

(a) (b)

Figure 8-4: The 𝜃 vs 𝜑 distributions for 𝜋 + , before (left) and after (right) the appli-
cation of fiducial cuts for 2.2 GeV analysis of 4 He.

As neutral particles, the fiducial cuts for photons take the from of a cut on the
detection area of the electromagnetic calorimeter, with no momentum dependence.
Photon events are selected after applying the same 𝑢, 𝑣 and 𝑤 calorimeter cuts as
used for electrons, and the fiducial regions for each energy and target defined on
distributions of 𝑐𝑜𝑠𝜃 vs 𝜑. Two first order polynomials are used to describe the
outline of the sides of the sector and two second order polynomials to describe the
top and bottom edges, as shown in figure 8-5. Additionally, two ‘hot spots’ in the
corners of sector four were removed.
The originally defined fiducial cut functions for e2a were implemented in C++ and
FORTRAN versions for the ROOT and PAW data analysis frameworks respectively.

173
 (a) 4.4 GeV.

Figure 8-5: The cos 𝜃 vs 𝜑 distributions for photons, with fiducial cut outline indicated
by red for 4.4 GeV analysis for 3 He.

They were checked into the CLAS CVS repository for e2 analysis software, and have
been applied as part of the various analyses performed on this data since it was
collected in 1999.
The e4nu analyses of e2a data have used the C++/ROOT implementation of the
fiducial cuts, incorporating them into their analysis software, with minor modifica-
tions to ensure compatibility with contemporary software environments.
In the analysis of [235], 𝜋 − fiducial cuts were defined, initially by direct reuse of
the electron fiducial cut parameters to cut fiducial regions for 𝜋 − . This was considered
to offer a reasonable approximation for 𝜋 − , having the same charge and thus similar
behaviour in the torus field of CLAS. However, because the electron fiducial cuts are
only defined to a minimum momentum of 350 MeV/c, and the minimum momentum
threshold used for pion detection in CLAS is 150 MeV/c, a new set of cut parameters
had to be defined for the 𝜋 − in the momentum range 150 − 350 MeV/c. These used
50 MeV/c momentum bins on 2.2 GeV 12
C data, and follow the same procedure to
obtain the 𝜃 vs 𝜑 outline cuts and the 𝜃 gaps corresponding to malfunctioning TOF
paddles at this low momentum as described in [234] for the 𝑝 > 350 MeV/c region.
At 1.1 GeV beam energy, this procedure was used to define 𝜋 − fiducial cuts for
the whole momentum range, as the electron cuts offer minimal coverage for 𝜋 − .
The theta gap functions for dead channels in 𝜋 − were updated to account for
previously missed gaps, and to skip electron gaps from the CC, a detector not used

174
in determining the 𝜋 − . Several functions were refit to extend to the lower momentum
range of the pions, removing the need to apply them only in specific momentum
ranges, and allowing the clumsy box cuts previously used to be eliminated.

Figure 8-6: The 𝜃 vs momentum distributions for 𝜋 − in sector 3, after the application
of the original fiducial and theta gap cuts for 2.2 GeV analysis of 4 He. At low
momentum, the theta gap cuts fail, and have been replaced box cuts that do not
appropriately describe the gaps at low momenta.

Figure 8-7 shows the polar angle versus momentum distributions for 𝜋 − in each
sector of CLAS, after the application of fiducial cuts. Of the theta gap functions used
for the electrons, several are carried over unchanged, some are updated in order to
appropriately apply to the full momentum range of 𝜋 − , while gaps defined for the CC
are omitted, as this subsystem is not used to identify 𝜋 − .
Figure 8-8 shows the polar angle versus momentum distributions for 𝜋 − in each
sector of CLAS, after the application of fiducial cuts. As with the 1 GeV case, several
theta gap functions are carried over unchanged from electrons, some are updated in
order to appropriately apply to the full momentum range seen for 𝜋 − , and CC gaps
are omitted. Additionally at 2.2 and 4.4 GeV, the maximum polar angle condition
for 𝑝 > 350 MeV/c, imposed by the electron fiducial cut parameters, is removed.
At 1 GeV beam energy, dedicated fiducial cut parameters for 𝜋 − at 750 A torus
current were defined as part of the analysis of [235]. The 1500 A data was not used in

175
Figure 8-7: The 𝜃 vs momentum distributions for 𝜋 − , showing the 𝜋 − updated theta
gap cuts (green), retained (red) and not used (yellow) electron theta gap cuts for
1.1 GeV analysis of 3 He.

the analysis of [235], therefore no bespoke 𝜋 − fiducial cut parameters were produced
for this field setting. Figure 8-9 shows the 𝜃 vs momentum distributions for the 𝜋 −
at 1.1 GeV, after the application of the updated fiducial cuts.
Figure 8-10 shows the 𝜃 vs momentum distributions for the 𝜋 − at 2.2 GeV, after
the application of the updated fiducial cuts. The same cuts are used in 4.4 GeV
analyses.
As seen in the reuse of electron theta gap cuts on the 𝜋 − , where the gap functions
were not defined to the lower momentum of the pions, a similar effect is seen in 𝜋 + ,
with the gap functions defined for protons directly reused. This has resulted in similar
box cuts being applied at low momentum, as can be seen in figure 8-11, removing
good events rather than low and poorly understood acceptance regions.
Unlike the 𝜋 − , the positive hadron fiducial cut functions, defined for protons, are
for the most part valid at the lower pion momentum range. We have extended the
application of several of these functions to define cuts at lower 𝜋 + momenta, where
the functions continue to appropriately describe the gap. This is shown for sector 3

176
Figure 8-8: The 𝜃 vs momentum distributions for 𝜋 − , showing the 𝜋 − updated theta
gap cuts (green), retained (red) and not used (yellow) electron theta gap cuts for
2.2 GeV analysis of 3 He. A new parameterisation of the maximum polar angle cut is
also shown (black line). The same gap functions are applied at 4.4 GeV

(a)

Figure 8-9: The 𝜃 vs momentum distributions for 𝜋 − , after the application of the
updated theta gap cuts for 1.1 GeV analysis of 3 He.

177
 (a)

Figure 8-10: The 𝜃 vs momentum distributions for 𝜋 − , after the application of the
updated theta gap cuts for 2.2 GeV analysis of 3 He. The same cuts are applied in
the 4.4 GeV analysis.

Figure 8-11: The 𝜃 vs momentum distributions for 𝜋 + , showing the original theta
gap cuts for analysis of 2.2 GeV 4 He data. The same gap functions are applied at
4.4 GeV.

178
in figure 8-12.

(a) (b)

Figure 8-12: The 𝜃 vs momentum distributions for 𝜋 + in sector 3, showing the original
(a) and new (b) theta gap cuts for 2.2 GeV analysis of 3 He. The same gap functions
are applied at 4.4 GeV beam energy.

179
180
List of Figures

1-1 The Standard Model of particle physics illustrating the three genera-
tions of fermions, the gauge bosons, and the scalar Higgs boson. Figure
adapted from ScienceAlert [4]. . . . . . . . . . . . . . . . . . . . . . . 12
1-2 Graphic illustrating the left-handed (right-handed) nature of neutri-
nos (anti-neutrinos) via the orientation of the momentum-spin vectors.
Figure adapted from [6]. . . . . . . . . . . . . . . . . . . . . . . . . . 12
1-3 The experimental setup of an accelerator-based long-baseline neutrino
oscillation experiment. . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1-4 Neutrino energy spectra reconstruction depends on our ability to model
the interaction of neutrinos with atomic nuclei and the propagation
of particles through the atomic nucleus. This flow chart shows the
process, starting with an oscillated far-detector incident-energy spec-
trum (green), differentiating the physical neutrino interactions (green
arrows) from the experimental analysis (blue arrows), and ending up
with an inferred incident-energy spectrum that hopefully matches the
actual one. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1-5 The four main interaction processes for neutrino-nucleus scattering
events. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1-6 Neutrino cross-section evolution as a function of the neutrino energy
illustrating the energy range where each one of the four main processes
dominates. Figure adapted from [34]. . . . . . . . . . . . . . . . . . . 26
1-7 Feyman diagrams illustrating charged current (CC) and neutral current
(NC) processes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26

181
1-8 Nucleon momentum distribution options available in commonly used
neutrino event generators. Figure adapted from [35]. . . . . . . . . . 27

1-9 Schematic illustration the wealth of possible final state interactions

that the final state hadrons can undergo. Figure adapted from [39]. . 27

1-10 Quasielastic lepton-nucleus scattering where the outgoing nucleon does

not reinteract as it leaves the nucleus. . . . . . . . . . . . . . . . . . . 30

1-11 Graphic illustration of the three complementary approaches used in

this thesis to improve our understanding of lepton-nucleus interactions. 31

2-1 The Fermilab accelerator complex showing the accelerator components

and the different beams. Figure adapted from [45]. . . . . . . . . . . 34

2-2 The ingredients for a neutrino beam include the accelerated protons,
the target, the magnetic horn, the decay pipe, and the absorbers. Fig-
ure adapted from [45]. . . . . . . . . . . . . . . . . . . . . . . . . . . 35

2-3 The BNB neutrino flux prediction through the MicroBooNE detector
for 𝜈𝜇 , 𝜈¯𝜇 , 𝜈𝑒 , and 𝜈¯𝑒 . A TPC volume with dimensions 2.56 m × 2.33 m
× 10.37 m is assumed. Figure adapted from [46]. . . . . . . . . . . . . 35

2-4 Schematic illustration of the MicroBooNE detector and its dimensions.

Figure adapted from [55]. . . . . . . . . . . . . . . . . . . . . . . . . . 37

2-5 The MicroBooNE coordinate system. . . . . . . . . . . . . . . . . . . 37

2-6 The design of CRT planes as part of the MicroBooNE detector. Simula-
tion of cosmic rays crossing the CRT, the brown lines represent possible
cosmic ray trajectories. There are four CRT planes: top plane, bottom
plane, pipe side plane and feedthrough side plane. The beam direction
is along the z axis. Figure adapted from [56]. . . . . . . . . . . . . . . 38

2-7 Working principle of a LArTPC detector. Figure adapted from [61]. . 39

182
2-8 (Left) bipolar (U and V induction planes) and unipolar (Y collec-
tion plane) signal induction on the three MicroBooNE planes. Figure
adapted from [63]. (Right) schematic view of the wire planes. The
vertical collection Y wires are shown in pink, the induction U wires,
angled at +60𝑜 are shown in blue and the induction V wires, angled at
-60𝑜 , are shown in green. Figure adapted from [64]. . . . . . . . . . . 40
2-9 Entry/exit points of cosmic muon tracks with a signal from a muon
counter located outside of the cryostat. In the absence of space charge
effects, the points should be located along the TPC boundaries indi-
cated by the dashed lines. Figure adapted from [66]. . . . . . . . . . . 41
2-10 The production of scintillation light in liquid argon. A charged particle
can either excite or ionise the argon. Figure adapted from [6]. . . . . 42
2-11 The MicroBooNE light collection system with the 32 PMTs. Figure
adapted from [69]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
2-12 (Left) cosmic-induced event that was stored because of the coincidence
of a 1.6 𝜇s accelerator BNB signal and light detected by the PMTs.
(Right) neutrino-induced event where the light was coming from a neu-
trino interaction. Figure adapted from [70]. . . . . . . . . . . . . . . . 45
2-13 Optically reconstructed flash object recorded by the MicroBooNE PMT
light collection system. The dark orange regions represent a higher PE
yield. Figure adapted from [64]. . . . . . . . . . . . . . . . . . . . . . 46
2-14 Neutrino-induced tracks (black) are matched to the corresponding light
signals collected by PMTs (red circles) and are clearly separated from
the cosmic-induced ones (dimmed color tracks). Figure adapted from [72]. 47
2-15 Candidate neutrino event display from MicroBooNE data on one of the
induction planes. (a) The raw waveform image. (b) The image after
noise-filtering. (c) The image after 2D deconvolution. The image qual-
ity near the neutrino interaction vertex significantly improves after the
2D deconvolution and the latter leads to improvements in the pattern
recognition. Figure adapted from [74]. . . . . . . . . . . . . . . . . . 49

183
2-16 Illustration of the hierarchical structure of particles reconstructed for
a simulated charged current 𝜈𝜇 event in MicroBooNE. The interaction
includes a muon, proton and charged pion in the visible final state.
Figure adapted from [75]. . . . . . . . . . . . . . . . . . . . . . . . . . 50

2-17 Illustration of the measured dQ/dx vs dE/dx distribution with the

modified recombination model in the MC simulation with the Ar-
goNeuT parametrization. Figure adapted from [76]. . . . . . . . . . . 51

2-18 MicroBooNE event display of an event in a cosmic overlay sample.

A simulated neutrino event (orange box) is overlaid on top of cosmic
events (blue box). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52

3-1 Interaction breakdown of the cos𝜃𝜇 plot illustrating the dominance of

CCQE interactions after the application of our selection cuts. . . . . . 59

3-2 Interaction breakdown of the cos𝜃𝜇 plot illustrating the dominance of

RES interactions after the application of our selection cuts for the non-
CC1p0𝜋 background part of the MC sample. . . . . . . . . . . . . . . 60

3-3 Vertex 𝑧 distribution for the measured events, after the beam related
MC background has been subtracted, before (left) and after (right)
detection efficiency corrections. No small-𝑧 enhancement is observed
and, with efficiency corrections, the measured distribution is consistent
with that of a uniform neutrino interaction vertex. . . . . . . . . . . . 62

3-4 The flux integrated single differential CC1p0𝜋 cross sections as a func-
tion of the cosine of the measured muon scattering angle. Inner and
outer error bars show the statistical and total (statistical and system-
atic) uncertainty at the 1𝜎, or 68%, confidence level. Colored lines
show the results of theoretical absolute cross section calculations using
different event generators (without passing through a detector simula-
tion). The blue band shows the extracted cross section obtained from
analyzing MC events propagated through our full detector simulation.
The width of the band denotes the simulation statistical uncertainty. 64

184
3-5 As figure 3-4, but for the differential cross sections as a function of
measured muon momentum (left) and measured proton scattering an-
gle (middle) and momentum (right). Cross sections are shown for the
full measured phase-space (top) and for events with cos(𝜃𝜇 ) < 0.8 (bot-
tom). Inner and outer error bars show the statistical and total (statis-
tical and systematic) uncertainty at the 1𝜎, or 68%, confidence level.
Colored lines show the results of theoretical absolute cross section cal-
culations using different event generators (without passing through a
detector simulation). The blue band shows the extracted cross section
obtained from analyzing MC events passed through our full detector
simulation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
3-6 The flux integrated single differential CC1p0𝜋 cross sections as a func-
tion of 𝑄2𝐶𝐶𝑄𝐸 = (𝐸𝜈𝑐𝑎𝑙 − 𝐸𝜇 )2 − (⃗𝑝𝜈 − 𝑝⃗𝜇 )2 and 𝐸𝜈𝑐𝑎𝑙 = 𝐸𝜇 + 𝑇𝑝 + 𝐵𝐸,
where 𝐵𝐸 = 40 MeV and 𝑝⃗𝜈 = (0, 0, 𝐸𝜈𝑐𝑎𝑙 ). Colored lines show the
results of theoretical absolute cross section calculations using different
event generators (without passing through a detector simulation). The
blue band shows the extracted cross section obtained from analyzing
MC events passed through our full detector simulation. . . . . . . . . 67
3-7 Muon angular distribution after the implementation of the analysis
framework improvements. No data-MC disagreement is observed in
the forward direction. . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
3-8 Schematic illustration of the single transverse variables 𝛿𝑝𝑇 , 𝛿𝛼𝑇 and
𝛿𝜑𝑇 . Figure adapted from [134]. . . . . . . . . . . . . . . . . . . . . . 70
3-9 Schematic illustration of 𝛿𝑝𝑇 𝑥 and 𝛿𝑝𝑇 𝑦 . Figure adapted from [135]. . 72
3-10 Interaction breakdown of the CC1p0𝜋 events as a function of 𝛿𝑝𝑇 𝑥 (left)
and 𝛿𝑝𝑇 𝑦 (right). The data correspond to the “combined” MicroBooNE
runs 1-3. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
3-11 Topological (left) and interaction (right) breakdown after the applica-
tion of the event selection for 𝛿𝑝𝑇 . . . . . . . . . . . . . . . . . . . . . 74
3-12 Response matrices of 𝛿𝑝𝑇 using the selected CC1p0𝜋 MC events. . . . 75

185
3-13 Total covariance matrix for 𝛿𝑝𝑇 . . . . . . . . . . . . . . . . . . . . . . 77

3-14 Additional smearing matrix 𝐴𝑐 for 𝛿𝑝𝑇 . . . . . . . . . . . . . . . . . . 78

3-15 The flux-integrated single- (top) and double- in 𝛿𝛼𝑇 bins (bottom) dif-
ferential CC1p0𝜋 cross sections as a function of the transverse missing
momentum 𝛿𝑝𝑇 . Inner and outer error bars show the statistical and
total (statistical and shape systematic) uncertainty at the 1𝜎, or 68%,
confidence level. The gray band shows the normalization systematic
uncertainty. Colored lines show the results of theoretical absolute cross
section calculations with and without FSI based on the GENIE and
GiBUU event generators. . . . . . . . . . . . . . . . . . . . . . . . . . 80

3-16 The flux-integrated single- (top) and double- in 𝛿𝑝𝑇 bins (bottom)
differential CC1p0𝜋 cross sections as a function of the angle 𝛿𝛼𝑇 . Inner
and outer error bars show the statistical and total (statistical and shape
systematic) uncertainty at the 1𝜎, or 68%, confidence level. The gray
band shows the normalization systematic uncertainty. Colored lines
show the results of theoretical absolute cross section calculations with
a number of FSI modeling options based on the GENIE event generator. 81

3-17 The flux-integrated single- (top) and double- in 𝛿𝑝𝑇,𝑦 bins (bottom)
differential CC1p0𝜋 cross sections as a function of the angle 𝛿𝑝𝑇,𝑥 .
Inner and outer error bars show the statistical and total (statistical and
shape systematic) uncertainty at the 1𝜎, or 68%, confidence level. The
gray band shows the normalization systematic uncertainty. Colored
lines show the results of theoretical absolute cross section calculations
with a number of event generators. . . . . . . . . . . . . . . . . . . . 82

3-18 Cross-section interaction breakdown for all the selected events. The
breakdown is shown for (top left) the G18 configuration with FSI ef-
fects, (top right) the G18 configuration without FSI effects, (bottom
left) GiB with FSI effects, and (bottom right) GiB without FSI effects. 83

186
3-19 Cross-section interaction breakdown for events with 𝛿𝛼𝑇 < 45𝑜 . The
breakdown is shown for (top left) the G18 configuration with FSI ef-
fects, (top right) the G18 configuration without FSI effects, (bottom
left) GiB with FSI effects, and (bottom right) GiB without FSI effects. 83

3-20 Cross-section interaction breakdown for events with 135𝑜 < 𝛿𝛼𝑇 <
180𝑜 . The breakdown is shown for (top left) the G18 configuration
with FSI effects, (top right) the G18 configuration without FSI effects,
(bottom left) GiB with FSI effects, and (bottom right) GiB without
FSI effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

3-21 Cross-section interaction breakdown for all the selected events. The
breakdown is shown for (top left) the G21 hA configuration with the
hA2018 FSI model, (top right) the G21 hN configuration with the hN
FSI model, (bottom left) the G21 G4 configuration with the G4 FSI
model, and (bottom right) the G21 No FSI configuration without FSI
effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84

3-22 Cross-section interaction breakdown for events with 𝛿𝑝𝑇 < 0.2 GeV/c.
The breakdown is shown for (top left) the G21 hA configuration with
the hA2018 FSI model, (top right) the G21 hN configuration with the
hN FSI model, (bottom left) the G21 G4 configuration with the G4
FSI model, and (bottom right) the G21 No FSI configuration without
FSI effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

3-23 Cross-section interaction breakdown for events with 𝛿𝑝𝑇 > 0.4 GeV/c.
The breakdown is shown for (top left) the G21 hA configuration with
the hA2018 FSI model, (top right) the G21 hN configuration with the
hN FSI model, (bottom left) the G21 G4 configuration with the G4
FSI model, and (bottom right) the G21 No FSI configuration without
FSI effects. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

187
3-24 Cross-section interaction breakdown for all the selected events. The
breakdown is shown for (top left) the G18 LFG configuration, (top
right) the G18 RFG configuration, (bottom left) the G18 EffSF con-
figuration, and (bottom right) the G18 No RPA configuration. . . . 86

3-25 Cross-section interaction breakdown for events with 𝛿𝑝𝑇,𝑦 < -0.15 GeV/c.
The breakdown is shown for (top left) the G18 LFG configuration, (top
right) the G18 RFG configuration, (bottom left) the G18 EffSF con-
figuration, and (bottom right) the G18 No RPA configuration. . . . 86

3-26 Cross-section interaction breakdown for events with -0.15 < 𝛿𝑝𝑇,𝑦 <
0.15 GeV/c. The breakdown is shown for (top left) the G18 LFG con-
figuration, (top right) the G18 RFG configuration, (bottom left) the
G18 EffSF configuration, and (bottom right) the G18 No RPA config-
uration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

3-27 MicroBooNE total Protons on Target (POT) collected with the Booster
Neutrino Beam (BNB) during the five run periods. In this thesis, the
first three run periods were used. . . . . . . . . . . . . . . . . . . . . 88

4-1 Charged-current cross sections as a function of neutrino energy ob-

tained using GENIE for muon neutrino scattering using the DUNE near
detector (left) and far detector (right) oscillated fluxes. The shaded
bands show the fractional contribution for each interaction mechanism,
quasielastic scattering (QE), meson-exchange currents (MEC), reso-
nance excitation (RES), and deep inelastic scattering (DIS). See text
for details of the interaction mechanisms. The numbers in parenthe-
ses indicate the percentage of the cross section due to each interaction
mechanism. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92

4-2 (Left) electron-nucleus inclusive scattering via one-photon exchange

and (right) charged current neutrino-nucleus inclusive scattering via
𝑊 exchange with a final state charged lepton. . . . . . . . . . . . . . 93

188
4-3 Reaction mechanisms for lepton-nucleus scattering (a) quasielastic scat-
tering (QE) where one nucleon is knocked out of the nucleus, (b) 2p2h
where two nucleons are knocked out of the nucleus, (c) RES resonance
production where a nucleon is excited to a resonance which decays to
a nucleon plus meson(s), and (d) DIS where the lepton interacts with
a quark in the nucleon. . . . . . . . . . . . . . . . . . . . . . . . . . . 93

4-4 Comparison between GENIE v2 and v3 descriptions of inclusive C(𝑒, 𝑒′ )

scattering cross sections at 𝐸0 = 0.56 GeV, 𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.24
GeV2 . Black points show the data, solid black line shows the GENIE
v3 results and dashed black line shows the GENIE v2 results. . . . . 95

4-5 Comparison of inclusive C(𝑒, 𝑒′ ) scattering cross sections for data and
for GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top)
𝐸0 = 0.24 GeV, 𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.05 GeV2 (middle) 𝐸0 = 0.56
GeV, 𝜃𝑒 = 36∘ and 𝑄2𝑄𝐸 ≈ 0.11 GeV2 , and (bottom) 𝐸0 = 0.56 GeV,
𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.24 GeV2 . Black points show the data, solid
black lines show the total GENIE prediction, colored lines show the
contribution of the different reaction mechanisms: (blue) QE, (red)
MEC, (green) RES and (orange) DIS. . . . . . . . . . . . . . . . . . . 102

4-6 Comparison of inclusive C(𝑒, 𝑒′ ) scattering cross sections for data and
for GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top)
𝐸0 = 0.96 GeV, 𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.32 GeV2 , (middle) 𝐸0 = 1.30
GeV, 𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.54 GeV2 , and (bottom) 𝐸0 = 2.22 GeV,
𝜃𝑒 = 15.5∘ and 𝑄2𝑄𝐸 ≈ 0.33 GeV2 . Black points show the data, solid
black lines show the total GENIE prediction, colored lines show the
contribution of the different reaction mechanisms: (blue) QE, (red)
MEC, (green) RES and (orange) DIS. . . . . . . . . . . . . . . . . . . 103

189
4-7 Comparison of inclusive C(𝑒, 𝑒′ ) scattering cross sections for data and
for GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top)
𝐸0 = 1.501 GeV, 𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.92 GeV2 , (middle) 𝐸0 =
3.595 GeV, 𝜃𝑒 = 16∘ and 𝑄2𝑄𝐸 ≈ 1.04 GeV2 , and (bottom) 𝐸0 = 3.595
GeV, 𝜃𝑒 = 20∘ and 𝑄2𝑄𝐸 ≈ 1.3 GeV2 . Black points show the data,
solid black lines show the total GENIE prediction, colored lines show
the contribution of the different reaction mechanisms: (blue) QE, (red)
MEC, (green) RES and (orange) DIS. . . . . . . . . . . . . . . . . . . 104

4-8 Comparison of inclusive Fe(𝑒, 𝑒′ ) scattering cross sections for data and
for GENIE. (left) data vs GSuSAv2 and (right) data vs G2018. (top)
Fe(𝑒, 𝑒′ ), 𝐸0 = 0.56 GeV, 𝜃𝑒 = 60∘ and 𝑄2𝑄𝐸 ≈ 0.24 GeV2 , (middle)
Fe(𝑒, 𝑒′ ), 𝐸0 = 0.96 GeV, 𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.32 GeV2 , (bot-
tom) Fe(𝑒, 𝑒′ ), 𝐸0 = 1.30 GeV, 𝜃𝑒 = 37.5∘ and 𝑄2𝑄𝐸 ≈ 0.54 GeV2 .
Black points show the data, solid black lines show the total GENIE
prediction, colored lines show the contribution of the different reaction
mechanisms: (blue) QE, (red) MEC, (green) RES and (orange) DIS. . 105

4-9 Comparison of inclusive Ar(𝑒, 𝑒′ ) scattering cross sections for data and
for GENIE at 𝐸0 = 2.22 GeV, 𝜃𝑒 = 15.5∘ and 𝑄2𝑄𝐸 ≈ 0.33 GeV2 . (left)
data vs GSuSAv2 and (right) data vs G2018. Black points show the
data, solid black lines show the total GENIE prediction, colored lines
show the contribution of the different reaction mechanisms: (blue) QE,
(red) MEC, (green) RES and (orange) DIS. . . . . . . . . . . . . . . 106

4-10 Comparison of inclusive proton (left) and deuterium (right) (𝑒, 𝑒′ ) scat-
tering cross sections for data and for GENIE using G2018. (top) 𝐸0 =
2.445 GeV and 𝜃𝑒 = 20∘ , (middle) 𝐸0 = 3.245 GeV and 𝜃𝑒 = 26.98∘ ,
and (bottom) 𝐸0 = 5.5 GeV and 𝜃𝑒 = 41∘ . Black points show the
data, solid black lines show the total GENIE prediction, colored lines
show the contribution of the different reaction mechanisms: (green)
RES and (orange) DIS. The first peak at lowest energy transfer is the
∆(1232) resonance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107

190
4-11 Comparison of inclusive proton (left) and deuterium (right) (𝑒, 𝑒′ ) scat-
tering cross sections for data and for GENIE using G2018. (top)
𝐸0 = 4.499 GeV and 𝜃𝑒 = 4∘ , (middle) 𝐸0 = 6.699 GeV and 𝜃𝑒 = 4∘ ,
and (bottom) 𝐸0 = 9.993 GeV and 𝜃𝑒 = 4∘ . Black points show the
data, solid black lines show the total GENIE prediction, colored lines
show the contribution of the different reaction mechanisms: (green)
RES and (orange) DIS. The first peak at lowest energy transfer is the
∆(1232) resonance. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108

4-12 Comparison of semi-exclusive 1.16 GeV lepton-carbon scattering for

𝑄2 ≥ 0.1 GeV2 . The number of generated events is plotted versus en-
ergy transfer (left) and 4-momuntm transfer squared (right) for events
with exactly one proton with 𝑃𝑝 ≥ 300 MeV/c, no charged pions with
𝑃𝜋 ≥ 70 MeV/c and no neutral pions or photons of any momentum for
eGENIE electrons (orange) and GENIE CC 𝜈𝜇 (blue). The electron
events have been weighted by 𝑄4 . Both curves are area normalized. . 110

4-13 Number of simulated events for QE scattering on 12 C at 1.161 GeV with

𝑄2 ≥ 0.1 shown as a function of the energy transfer 𝜔 and the momen-
tum transfer 𝑞3 = |⃗𝑞 | for all the available nuclear models in GENIE for
neutrinos (top) and for electrons (bottom). (left) the GSuSAv2 model
which uses a Relativistic Mean Field momentum distribution, (middle)
the Nieves or Rosenbluth cross section with the Local Fermi Gas mo-
mentum distribution, and (right) the Llewellyn-Smith or Rosenbluth
cross section with the Relativistic Fermi Gas momentum distribution.
The electron events have been weighted by 𝑄4 . . . . . . . . . . . . . . 111

4-14 Initial momentum distribution of protons in simulated QE C(𝑒, 𝑒′ 𝑝)

events at 𝐸 = 1.161 GeV for the Local Fermi Gas (solid histogram)
and Relativistic Fermi Gas (dotted histogram) models. The two curves
are normalized to have the same area. . . . . . . . . . . . . . . . . . . 112

191
4-15 Number of simulated events as a function of the energy transfer 𝜔 and
of the momentum transfer 𝑞3 = |⃗𝑞 | for neutrinos (left) and for electrons
(right) using GSuSav2 for MEC interactions. The electron events have
been scaled by 𝑄4 and all the samples have been generated with 𝑄2 ≥
0.1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112

5-1 Schematic view of the accelerator facility and the experimental halls
at Jefferson Lab. Figure adapted from [196]. . . . . . . . . . . . . . . 116
5-2 Drawing of the CLAS detector showing the sector structure and the
different detectors. The beam enters from the upper left side. The
target is located at the center of the detector. . . . . . . . . . . . . . 117
5-3 The CLAS superconducting toroidal magnet. Figure adapted from [197].118
5-4 (Left) Illustration of the region 3 drift chamber structure for one of
the CLAS sectors. (Right) Schematic representation of the thee drift
chamber regions. Figures adapted from [198]. . . . . . . . . . . . . . 118
5-5 Schematic view of a CLAS electromagnetic calorimeter module. Fig-
ures adapted from [199]. . . . . . . . . . . . . . . . . . . . . . . . . . 120
5-6 Optical arrangement of one of the optical modules of the CLAS Cherenkov
detector, showing the optical and light collection components. Figures
adapted from [200]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
5-7 Schematic view of the TOF counters in one sector illustrating the
grouping into four panels. Figure adapted from [201]. . . . . . . . . . 122

6-1 The expected energy distribution of different 𝜈𝜇 beams, (left) before

oscillation at the near detector and (right) after oscillation at the far
detector. The vertical lines show the three electron beam energies of
this measurement. The NO𝜈A far-detector beam flux is calculated
using the near detector flux and the neutrino oscillation parameters
from [1]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
6-2 ∆𝜃𝛾,𝑒′ as a function of ∆𝜑𝛾,𝑒′ . The red box indicated the region with
radiated photons which was removed in our analysis. . . . . . . . . . 129

192
6-3 Number of events vs 𝐸𝑐𝑎𝑙 = 𝐸𝑒′ + 𝑇𝑝 the scattered electron energy plus
proton kinetic energy for 4.32 GeV H(𝑒, 𝑒′ 𝑝). Black points are data,
the blue histogram shows the unradiated GENIE prediction and the
black histogram shows the GENIE prediction with electron radiation.
The GENIE calculations have been scaled to have the same integral as
the data. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130

6-4 Schematic illustration of the data driven background correction using

detected (e,e’p𝜋) events. . . . . . . . . . . . . . . . . . . . . . . . . . 132

6-5 The effect of undetected pion subtraction. The number of weighted

events as a function of reconstructed energy 𝐸𝑄𝐸 for 4.453 GeV Fe(𝑒, 𝑒′ )
events for (left) events with a detected 𝜋 ± or photon (blue), events with
one (red) or two (light brown) undetected 𝜋 ± or photons and (right)
all (𝑒, 𝑒′ 𝑋) events with detected or undetected 𝜋 ± or photon (blue),
(𝑒, 𝑒′ ) events with no detected 𝜋 ± or photon (red), and (𝑒, 𝑒′ ) events
after subtraction for undetected 𝜋 ± or photon (light brown). . . . . . 133

6-6 Illustration of the successful closure test of the data driven correc-
tion for undetected particles as a function of 𝐸 𝐶𝑎𝑙 using the (e,e’p)1𝑝0𝜋
channel on 12
C at 𝐸𝑏𝑒𝑎𝑚 = 2.257 GeV. The contribution of the un-
subtracted (e,e’p)1𝑝0𝜋 spectrum (black) is reduced to the subtracted
(e,e’p)1𝑝0𝜋 spectrum (magenta), which is in reasonable agreement with
the true (e,e’p)1𝑝0𝜋 spectrum (green). . . . . . . . . . . . . . . . . . . 134

6-7 The proton (black) and charged pion (blue) multiplicities for data
(points), SuSav2 (solid histogram) and G2018 (dashed histogram) for
2.257 GeV carbon. Error bars show the 68% (1𝜎) confidence limits
for the statistical and point-to-point systematic uncertainties added in
quadrature. Error bars are not shown when they are smaller than the
size of the data point. Normalization uncertainties of 3% not shown. . 134

193
6-8 (Top row) Acceptance correction factors, (middle row) acceptance cor-
rection factor uncertainties, and (bottom row) electron radiation cor-
rection factors plotted vs E𝑐𝑎𝑙 for the three incident beam energies.
Results for carbon are shown in black, helium in green and iron in ma-
genta. The left column (a,d,g) shows the 1.1 GeV results, the middle
column (b,e,h) shows the 2.2 GeV results and the right column (c,f,i)
shows the 4.4 GeV results. . . . . . . . . . . . . . . . . . . . . . . . . 135

6-9 Electron acceptance maps for 12

C at 𝐸𝑏𝑒𝑎𝑚 = 1.159 GeV as a function
of (left) cos𝜃𝑒 vs 𝜑𝑒 and (right) cos𝜃𝑒 vs momentum 𝑝𝑒 . . . . . . . . . 135

6-10 Comparison between the inclusive C(𝑒, 𝑒′ ) cross sections measured at

37.5∘ for data (points) and SuSav2 (lines) for the 0.961 and 1.299 GeV
SLAC data and our 1.159 GeV CLAS data. . . . . . . . . . . . . . . . 137

6-11 The 1.159 GeV C(𝑒, 𝑒′ )0𝜋 cross section plotted as a function of the re-
constructed energy 𝐸𝑄𝐸 for data (black points), GENIE SuSAv2 (solid
black curve) and GENIE G2018 (dotted black curve). The colored lines
show the contributions of different processes to the GENIE SuSAv2
cross section: QE (blue), MEC (red), RES (green) and DIS (orange).
Error bars show the 68% (1𝜎) confidence limits for the statistical and
point-to-point systematic uncertainties added in quadrature. Error
bars are not shown when they are smaller than the size of the data
point. Normalization uncertainty of 3% not shown. . . . . . . . . . . 140

194
6-12 The 𝐴(𝑒, 𝑒′ 𝑝)0𝜋 cross section plotted as a function of the reconstructed
quasielastic energy 𝐸𝑄𝐸 for data (black points), SuSAv2 (black solid
curve) and G2018 (black dotted curve). Different panels show results
for different beam energy and target nucleus combinations: (top row)
Carbon target at (left to right) 1.159, 2.257 and 4.453 GeV, and (bot-
tom) Iron target at (left) 2.257 and (right) 4.453 GeV incident beam.
The 1.159 GeV yields have been scaled by 1/2 and the 4.453 GeV yields
have been scaled by 5 to have the same vertical scale. Colored lines
show the contributions of different processes to the SuSAv2 GENIE
simulation: QE (blue), MEC (red), RES (green) and DIS (orange).
Error bars show the 68% (1𝜎) confidence limits for the statistical and
point-to-point systematic uncertainties added in quadrature. Error
bars are not shown when they are smaller than the size of the data
point. Normalization uncertainties of 3% not shown. . . . . . . . . . 141

6-13 The 𝐴(𝑒, 𝑒′ 𝑝)1𝑝0𝜋 cross section plotted as a function of the reconstructed
calorimetric energy 𝐸𝑐𝑎𝑙 for data (black points), SuSAv2 (black solid
curve) and G2018 (black dotted curve). Different panels show results
for different beam energy and target nucleus combinations: (top row)
Carbon target at (left to right) 1.159, 2.257 and 4.453 GeV, and (bot-
tom) Iron target at (left) 2.257 and (right) 4.453 GeV incident beam.
The 1.159 GeV yields have been scaled by 1/2 and the 4.453 GeV yields
have been scaled by 5 to have the same vertical scale. The insets show
the cross sections with the same horizontal scale and an expanded verti-
cal scale. Colored lines show the contributions of different processes to
the SuSAv2 GENIE simulation: QE (blue), MEC (red), RES (green)
and DIS (orange). Error bars show the 68% (1𝜎) confidence limits
for the statistical and point-to-point systematic uncertainties added in
quadrature. Error bars are not shown when they are smaller than the
size of the data point. Normalization uncertainties of 3% not shown. 143

195
6-14 Energy feed-down cross-sections (𝐸𝑟𝑒𝑐 − 𝐸𝑡𝑟𝑢𝑒 )/𝐸𝑡𝑟𝑢𝑒 for data (points)
and SuSav2 (lines) for 1.159 GeV (red triangles and dotted lines),
2.257 GeV (green squares and dashed lines) and 4.453 GeV (blue dots
and solid lines) on carbon for (a) 𝐸 𝑐𝑎𝑙 , and (b) 𝐸 𝑄𝐸 . . . . . . . . . . 144

6-15 Energy feed-down cross-sections (𝐸𝑟𝑒𝑐 − 𝐸𝑡𝑟𝑢𝑒 )/𝐸𝑡𝑟𝑢𝑒 for data (points)
and SuSav2 (lines) for 1.159 GeV (red triangles and dotted lines),
2.257 GeV (green squares and dashed lines) and 4.453 GeV (blue dots
and solid lines) on iron for (c) 𝐸 𝑐𝑎𝑙 , and (d) Fe 𝐸 𝑄𝐸 . . . . . . . . . . 145

6-16 (Left) the 2.257 GeV C(𝑒, 𝑒′ 𝑝)1𝑝0𝜋 cross section plotted versus missing
transverse momentum, 𝑃𝑇 , for data (black points), SuSav2 (black solid
line) and G2018 (black dashed line). The vertical lines at 200 MeV/𝑐
and at 400 MeV/𝑐 separate the three bins in 𝑃𝑇 . Colored lines show
the contributions of different processes to the SuSAv2 GENIE simula-
tion: QE (blue), MEC (red), RES (green) and DIS (orange). (Right)
The cross section plotted versus the calorimetric energy 𝐸𝑐𝑎𝑙 for dif-
ferent bins in 𝑃𝑇 : (top) 𝑃𝑇 < 200 MeV/𝑐, (middle) 200 MeV/𝑐 ≤ 𝑃𝑇 ≤
400 MeV/𝑐, and (bottom) 𝑃𝑇 > 400 MeV/𝑐. . . . . . . . . . . . . . . 147

6-17 The cross section plotted vs transverse missing momentum 𝑃𝑇 for data
(black points), SuSAv2 (black solid curve) and G2018 (black dotted
curve). Different panels show results for different beam energy and
target nucleus combinations: (top row) Carbon target at (left to right)
1.159, 2.257 and 4.453 GeV, and (bottom) Iron target at (left) 2.257
and (right) 4.453 GeV. The 4.453 GeV yields have been scaled by four
to have the same vertical scale. Colored lines show the contributions
of different processes to the SuSAv2 GENIE simulation: QE (blue),
MEC (red), RES (green) and DIS (orange). . . . . . . . . . . . . . . 148

196
6-18 The cross section plotted vs 𝛿𝛼𝑇 (a-e) and vs 𝛿𝜑𝑇 (f-j) for data (black
points), SuSAv2 (black solid curve) and G2018 (black dotted curve).
Different panels show results for different beam energy and target nu-
cleus combinations: (top row) Carbon target at (left to right) 1.159,
2.257 and 4.453 GeV, and (bottom) Iron target at (left) 2.257 and
(right) 4.453 GeV. The 4.453 GeV yields have been scaled by two to
have the same vertical scale. Colored lines show the contributions of
different processes to the SuSAv2 GENIE simulation: QE (blue), MEC
(red), RES (green) and DIS (orange). . . . . . . . . . . . . . . . . . . 149

6-19 Data-simulation comparisons for 12

C at 2.261 GeV showing the cross
section results as a function of 𝑃𝑇 for (top) all the events, (bottom
left) events with 𝛿𝛼𝑇 < 45𝑜 dominated by QE interactions and no
reinteractions, and (bottom right) events with 135𝑜 < 𝛿𝛼𝑇 < 180𝑜
maximally affected by FSI and multi-hadron channels. Colored lines
show the contributions of different processes to the SuSAv2 GENIE
simulation: QE (blue), MEC (red), RES (green) and DIS (orange). . 150

6-20 Data-simulation comparisons for 12

C at 2.261 GeV showing the cross
section results as a function of 𝛿𝛼𝑇 for (top) all the events, (bottom
left) events with 𝑃𝑇 < 0.2 GeV/c dominated by QE interactions and no
reinteractions, and (bottom right) events with 𝑃𝑇 > 0.4 GeV/c maxi-
mally affected by FSI and multi-hadron channels. Colored lines show
the contributions of different processes to the SuSAv2 GENIE simula-
tion: QE (blue), MEC (red), RES (green) and DIS (orange). . . . . 151

6-21 (Top) data-simulation comparisons on 12

C at 2.261 GeV showing the
cross section results as a function of (top left) the true missing mo-
mentum 𝑃𝑀 𝑖𝑠𝑠 and (top right) the missing momentum approximation
𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 commonly used by neutrino experiments. (Bottom) overlay of
the two extracted data cross sections illustrating the differences be-
tween 𝑃𝑀 𝑖𝑠𝑠 and 𝑃𝑛,𝑝𝑟𝑜𝑥𝑦 . . . . . . . . . . . . . . . . . . . . . . . . . 152

197
6-22 Data-simulation comparisons for 12
C at 2.261 GeV showing the cross
section results as a function of 𝑃𝑇,𝑥 for (top) all the events, (bottom
left) events with -0.15 < 𝑃𝑇,𝑦 < 0.15 GeV/c dominated by QE inter-
actions and no reinteractions, and (bottom right) events with 𝑃𝑇,𝑦 <
-0.15 GeV/c maximally affected by FSI and multi-hadron channels. . 153
6-23 Data-simulation comparisons for 12
C at 2.261 GeV showing the cross
section results as a function of 𝑃𝑇,𝑦 for (top) all the events, (bottom
left) events with -0.15 < 𝑃𝑇,𝑥 < 0.15 GeV/c dominated by QE inter-
actions and no reinteractions, and (bottom right) events with 𝑃𝑇,𝑥 <
-0.15 GeV/c maximally affected by FSI and multi-hadron channels. . 154
6-24 Schematic view of the upgraded CLAS12 detector components. . . . . 155

8-1 The 𝜃 vs 𝜑 distributions for 𝜋 − at 2.2 GeV. . . . . . . . . . . . . . . . 171

8-2 The 𝜃 vs 𝜑 distributions for electrons at 2.2 GeV. . . . . . . . . . . . 172
8-3 The 𝜃 vs 𝜑 distributions for protons. . . . . . . . . . . . . . . . . . . 173
8-4 The 𝜃 vs 𝜑 distributions for 𝜋 + . . . . . . . . . . . . . . . . . . . . . . 173
8-5 The cos 𝜃 vs 𝜑 distributions for photons. . . . . . . . . . . . . . . . . 174
8-6 The 𝜃 vs momentum distributions for 𝜋 − . . . . . . . . . . . . . . . . . 175
8-7 The 𝜃 vs momentum distributions for 𝜋 − . . . . . . . . . . . . . . . . . 176
8-8 The 𝜃 vs momentum distributions for 𝜋 − . . . . . . . . . . . . . . . . . 177
8-9 The 𝜃 vs momentum distributions for 𝜋 − . . . . . . . . . . . . . . . . . 177
8-10 The 𝜃 vs momentum distributions for 𝜋 − . . . . . . . . . . . . . . . . . 178
8-11 The 𝜃 vs momentum distributions for 𝜋 − . . . . . . . . . . . . . . . . . 178
8-12 The 𝜃 vs momentum distributions for 𝜋 + . . . . . . . . . . . . . . . . . 179

198
List of Tables

1.1 Summary table of the 3𝜈 oscillation parameters. The 1𝜎 intervals for

both the case normal are inverse ordering are shown. Table adapted
from [1]. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

3.1 Interaction breakdown after the application of our selection cuts. . . . 59

3.2 Integrated cross section values and 𝜒2 values for the agreement be-
tween the measured cross sections and various event generators. Re-
sults are listed for the full measured phase space and for a limited one
of cos(𝜃𝜇 ) < 0.8. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66

5.1 Target areal densities and integrated charges for the 𝑒4𝜈 data sets. . . 122

6.1 Summary of the total systematic uncertainties used in the e4𝜈 analysis. 138
6.2 (𝑒, 𝑒′ 𝑝)1𝑝0𝜋 events reconstructed to the correct beam energy. Peak Frac-
tion refers to the fraction of events reconstructed to the correct beam
energy and Peak Sum refers to the integrated weighted cross section (as
shown in Fig. 6-13) reconstructed to the correct beam energy. The peak
integration windows are 1.1 ≤ 𝐸𝑐𝑎𝑙 ≤ 1.22 GeV, 2.19 ≤ 𝐸𝑐𝑎𝑙 ≤ 2.34
GeV, and 4.35 ≤ 𝐸𝑐𝑎𝑙 ≤ 4.60 GeV, respectively, for the three incident
beam energies. SuSAv2 is not intended to model nuclei lighter than 12
C.142

199
200
Bibliography

[1] M. Tanabashi et al. Review of particle physics. Phys. Rev. D, 98:030001, Aug
2018.

[2] Y. Fukuda et al. Evidence for oscillation of atmospheric neutrinos. Phys. Rev.
Lett., 81:1562–1567, Aug 1998.

[3] Q. R. Ahmad et al. Direct evidence for neutrino flavor transformation from
neutral current interactions in the Sudbury Neutrino Observatory. Phys. Rev.
Lett., 89:011301, 2002.

[4] What Is The Standard Model of Particle Physics? https://www.

sciencealert.com/the-standard-model.

[5] M. Goldhaber, L. Grodzins, and A. W. Sunyar. Helicity of neutrinos. Phys.

Rev., 109:1015–1017, Feb 1958.

[6] Wouter Van De Pontseele. Search for Electron Neutrino Anomalies with the
MicroBooNE Detector. PhD thesis, Oxford U., 2020.

[7] M. Ahmed et al. Search for the lepton-family-number nonconserving decay

𝜇+ → 𝑒+ 𝛾. Phys. Rev. D, 65:112002, Jun 2002.

[8] M. Aker et al. Direct neutrino-mass measurement with sub-electronvolt sensi-

tivity. Nat. Phys., 18:160–166, 2022.

[9] E. Majorana. Teoria simmetrica dell’elettrone e del positrone. Nuovo Cim,

14:171, 1937.

[10] E.Kh. Akhmedov, G.C. Branco, and M.N. Rebelo. Seesaw mechanism and
structure of neutrino mass matrix. Physics Letters B, 478(1):215–223, 2000.

[11] Martin Freund. Analytic approximations for three neutrino oscillation param-
eters and probabilities in matter. Phys. Rev. D, 64:053003, Jul 2001.

[12] A. Cervera, A. Donini, M.B. Gavela, J.J. [Gomez Cádenas], P. Hernández,

O. Mena, and S. Rigolin. Golden measurements at a neutrino factory. Nuclear
Physics B, 579(1):17 – 55, 2000.

201
[13] A. Cervera, A. Donini, M.B. Gavela, J.J. Gomez Cádenas, P. Hernández,
O. Mena, and S. Rigolin. Erratum to “golden measurements at a neutrino
factory”: [nucl. phys. b 579 (2000) 17]. Nuclear Physics B, 593(3):731 – 732,
2001.

[14] C. Giunti and C. W. Kim. Fundamentals of neutrino physics and astrophysics.

University Press, 2007.

[15] ALEPH, DELPHI, L3, OPAL, and SLD Collaborations. Precision electroweak
measurements on the z resonance. Physics Reports, 427(5):257–454, 2006.

[16] A. Yu. Smirnov. The msw effect and matter effects in neutrino oscillations.
Phys. Scr. 2005 57, 2005.

[17] Pablo F. de Salas, Stefano Gariazzo, Olga Mena, Christoph A. Ternes, and
Mariam Tórtola. Neutrino mass ordering from oscillations and beyond: 2018
status and future prospects. Frontiers in Astronomy and Space Sciences, 5:36,
2018.

[18] M. G. Aartsen et al. Determining neutrino oscillation parameters from atmo-

spheric muon neutrino disappearance with three years of icecube deepcore data.
Phys. Rev. D, 91:072004, Apr 2015.

[19] A.A. Aguilar-Arevalo et al. The Neutrino Flux prediction at MiniBooNE. Phys.
Rev. D, 79:072002, 2009.

[20] B. Abi et al. Long-baseline neutrino oscillation physics potential of the DUNE
experiment. Eur. Phys. J. C, 80(10):978, 2020.

[21] M. Fukugita and T. Yanagida. Baryogenesis without grand unification. Phys.

Lett. B, 174:45, 1986.

[22] K. Abe et al. Constraint on the matter–antimatter symmetry-violating phase

in neutrino oscillations. Nature, 580:339, 2020.

[23] K. Abe et al. Search for CP Violation in Neutrino and Antineutrino Oscillations
by the T2K Experiment with 2.2 × 1021 Protons on Target. Phys. Rev. Lett.,
121(17):171802, 2018.

[24] L. Alvarez-Ruso et al. NuSTEC White Paper: Status and challenges of neu-
trino–nucleus scattering. Prog. Part. Nucl. Phys., 100:1–68, 2018.

[25] M.A. Acero et al. New constraints on oscillation parameters from 𝜈𝑒 appearance
and 𝜈𝜇 disappearance in the NOvA experiment. Phys. Rev. D, 98:032012, 2018.

[26] Artur M. Ankowski, Pilar Coloma, Patrick Huber, Camillo Mariani, and Erica
Vagnoni. Missing energy and the measurement of the CP-violating phase in
neutrino oscillations. Phys. Rev. D, 92(9):091301, 2015.

202
[27] Noemi Rocco. Ab initio calculations of lepton-nucleus scattering. Frontiers in
Physics, 8:116, 2020.

[28] S. Dolan, G. D. Megias, and S. Bolognesi. Implementation of the susav2-meson

exchange current 1p1h and 2p2h models in genie and analysis of nuclear effects
in t2k measurements. Phys. Rev. D, 101:033003, Feb 2020.

[29] Noemi Rocco, Alessandro Lovato, and Omar Benhar. Unified description of
electron-nucleus scattering within the spectral function formalism. Phys. Rev.
Lett., 116:192501, May 2016.

[30] L. Aliaga et al. Neutrino flux predictions for the numi beam. Phys. Rev. D,
94:092005, Nov 2016.

[31] Kuldeep K. Maan. Constraints on the Neutrino Flux in NOvA using the Near
Detector Data. PoS, ICHEP2016:931, 2016.

[32] L. Haegel. T2K near detector constraints for oscillation results. In 18th Interna-
tional Workshop on Neutrino Factories and Future Neutrino Facilities Search,
1 2017.

[33] C. Andreopoulos, A. Bell, D. Bhattacharya, F. Cavanna, J. Dobson, S. Dytman,

H. Gallagher, P. Guzowski, R. Hatcher, P. Kehayias, A. Meregaglia, D. Naples,
G. Pearce, A. Rubbia, M. Whalley, and T. Yang. The genie neutrino monte
carlo generator. Nucl. Instrum. Methods Phys. Res., Sect. A, 614(1):87 – 104,
2010.

[34] J. A. Formaggio and G. P. Zeller. From ev to eev: Neutrino cross sections across
energy scales. Rev. Mod. Phys., 84:1307–1341, Sep 2012.

[35] M. D. Tutto. First measurements of inclusive muon neutrino charged current

differential cross sections on argon at 0.8 gev average neutrino energy with the
microboone detector. PhD thesis, University of Oxford, 2019.

[36] A. Bodek and J. L. Ritchie. Fermi-motion effects in deep-inelastic lepton scat-

tering from nuclear targets. Phys. Rev. D, 23:1070–1091, Mar 1981.

[37] R.C. Carrasco and E. Oset. Interaction of Real Photons With Nuclei From
100-MeV to 500-MeV. Nucl. Phys. A, 536:445–508, 1992.

[38] Artur M. Ankowski and Jan T. Sobczyk. Argon spectral function and neutrino
interactions. Phys. Rev. C, 74:054316, Nov 2006.

[39] Lars Bathe-Peters. Studies of single transverse kinematic variables for neutrino
interactions on argon. Masters thesis, Harvard University, 2020.

[40] Teppei Katori and Marco Martini. Neutrino–nucleus cross sections for os-
cillation experiments. Journal of Physics G: Nuclear and Particle Physics,
45(1):013001, dec 2017.

203
[41] J E Amaro, M B Barbaro, J A Caballero, R González-Jiménez, G D Megias, and
I Ruiz Simo. Electron- versus neutrino-nucleus scattering. Journal of Physics
G: Nuclear and Particle Physics, 47(12):124001, nov 2020.

[42] A. Papadopoulou, A. Ashkenazi, S. Gardiner, M. Betancourt, S. Dytman, L. B.

Weinstein, E. Piasetzky, F. Hauenstein, M. Khachatryan, S. Dolan, G. D.
Megias, and O. Hen. Inclusive electron scattering and the genie neutrino event
generator. Phys. Rev. D, 103:113003, Jun 2021.

[43] Sacha E. Kopp. Accelerator neutrino beams. Physics Reports, 439(3):101–159,

2007.

[44] M. Szleper A. Para. Neutrino oscillations experiments using off-axis numi beam.
arXiv:0110032v1, 2001.

[45] Fermilab’s accelerator complex. https://www.fnal.gov/pub/science/particle-

accelerators/accelerator-complex.html.

[46] Aguilar-Arevalo et al. Neutrino flux prediction at miniboone. Phys. Rev. D,

79:072002, Apr 2009.

[47] P. Adamson et al. The numi neutrino beam. Nuclear Instruments and Meth-
ods in Physics Research Section A: Accelerators, Spectrometers, Detectors and
Associated Equipment, 806:279–306, 2016.

[48] J. Evans et al. The minos experiment: Results and prospects. Adv. High Energy
Phys., page 182537, 2013.

[49] Tomasz Golan, Leonidas Aliaga, and Mike Kordosky. MINERvA’s Flux Predic-
tion, chapter 12, page 8. JPS, 2016.

[50] R. Acciarri et al. Demonstration of mev-scale physics in liquid argon time

projection chambers using argoneut. Phys. Rev. D, 99:012002, Jan 2019.

[51] M. A. Acero et al. First measurement of neutrino oscillation parameters using

neutrinos and antineutrinos by nova. Phys. Rev. Lett., 123:151803, Oct 2019.

[52] P. Adamson et al. Search for sterile neutrinos in minos and minos+ using a
two-detector fit. Phys. Rev. Lett., 122:091803, Mar 2019.

[53] G. Karagiorgi, Z. Djurcic, J. M. Conrad, M. H. Shaevitz, and M. Sorel. Viability

of ∆𝑚2 ∼ 1 ev2 sterile neutrino mixing models in light of miniboone electron
neutrino and antineutrino data from the booster and numi beamlines. Phys.
Rev. D, 80:073001, Oct 2009.

[54] A. Aguilar-Arevalo et al. Significant excess of electronlike events in the mini-

boone short-baseline neutrino experiment. Phys. Rev. Lett., 121:221801, Nov
2018.

204
[55] C. Adams et al. Rejecting cosmic background for exclusive charged current quasi
elastic neutrino interaction studies with Liquid Argon TPCs; a case study with
the MicroBooNE detector. Eur. Phys. J. C, 79(8):673, 2019.
[56] C. Adams et al. Design and construction of the MicroBooNE Cosmic Ray
Tagger system. arXiv:1901.02862, 2019.
[57] D. Heck et al. Corsika: A monte carlo code to simulate extensive air showers.
Forschungszentrum Karlsruhe Report FZKA, 1998.
[58] S. Agostinelli et al. Geant4—a simulation toolkit. Nucl. Instrum. Meth., A 506,
2003.
[59] Jay N. Marx and David R. Nygren. The time projection chamber, 1978.
[60] C. Rubbia. The liquid argon time projection chamber: A new concept for
neutrino detectors, 1977.
[61] F. Cavanna, A. Ereditato, and B.T. Fleming. Advances in liquid argon de-
tectors. Nuclear Instruments and Methods in Physics Research Section A: Ac-
celerators, Spectrometers, Detectors and Associated Equipment, 907:1–8, 2018.
Advances in Instrumentation and Experimental Methods (Special Issue in Hon-
our of Kai Siegbahn).
[62] O. Bunemann, T. E. Cranshaw, and J. A. Harvey. Design of grid ionization
chambers. Canadian Journal of Research 27a.5, page 191–206, 1949.
[63] R. Acciarri et al. Noise characterization and filtering in the microboone liquid
argon tpc. Journal of Instrumentation, 12(08):P08003–P08003, aug 2017.
[64] MicroBooNE Collaboration. Design and construction of the microboone detec-
tor. JINST 12.P02017, 2017.
[65] H. Chen, G. De Geronimo, F. Lanni, D. Lissauer, D. Makowiecki, V. Radeka,
S. Rescia, C. Thorn, and B. Yu. Front End Readout Electronics of the Micro-
BooNE Experiment. Phys. Procedia, 37:1287–1294, 2012.
[66] P. Abratenko et al. Measurement of space charge effects in the MicroBooNE
LArTPC using cosmic muons. Journal of Instrumentation, 15(12):P12037–
P12037, dec 2020.
[67] Robert S. Mulliken. Potential curves of diatomic rare-gas molecules and their
ions, with particular reference to xe2. The Journal of Chemical Physics,
52(10):5170–5180, 1970.
[68] MicroBooNE Collaboration. Pmt gain calibration in microboone.
MICROBOONE-NOTE-1064-TECH, 2019.
[69] Microboone photomultiplier. https://news.fnal.gov/2015/07/microboone-
photomultiplier/.

205
[70] M. Del Tutto. First measurements of inclusive muon neutrino charged current
differential cross sections on argon at 0.8 GeV average neutrino energy with the
MicroBooNE detector. PhD thesis, University of Oxford, 2019.

[71] P. Abratenko et al. Neutrino event selection in the MicroBooNE liquid argon
time projection chamber using wire-cell 3d imaging, clustering, and charge-light
matching. Journal of Instrumentation, 16(06):P06043, jun 2021.

[72] Spotting accelerator-produced neutrinos in a cosmic haystack.

https://science.osti.gov/hep/Highlights/2022/HEP-2022-02-a.

[73] C. Adams et al. Ionization electron signal processing in single phase LArTPCs.
part i. algorithm description and quantitative evaluation with MicroBooNE
simulation. Journal of Instrumentation, 13(07):P07006–P07006, jul 2018.

[74] C. Adams et al. Ionization electron signal processing in single phase LArT-
PCs. Part II. Data/simulation comparison and performance in MicroBooNE.
J. Instrum., 13(07):P07007, 2018.

[75] R. Acciarri et al. The Pandora multi-algorithm approach to automated pat-

tern recognition of cosmic-ray muon and neutrino events in the MicroBooNE
detector. Eur. Phys. J. C, 78(1):82, 2018.

[76] R Acciarri et al. A study of electron recombination using highly ionizing par-
ticles in the ArgoNeuT liquid argon TPC. JINST, 8(08):P08005–P08005, aug
2013.

[77] C Adams et al. Reconstruction and measurement of o(100) mev energy

electromagnetic activity from neutral pion decays in the microboone lartpc.
arXiv:1910.02166, 2019.

[78] MicroBooNE Collaboration. Reconstruction performance studies with micro-

boone data in support of summer 2018 analyses. MICROBOONE-NOTE-1049-
PUB, 2018.

[79] GENIE Collaboration. Recent highlights from genie v3. Eur. Phys. J. Spec.
Top., 2021.

[80] C. Anderson et al. First Measurements of Inclusive Muon Neutrino Charged

Current Differential Cross Sections on Argon. Phys. Rev. Lett., 108(10):161802,
2012.

[81] Y. Nakajima et al. Measurement of Inclusive Charged Current Interactions on

Carbon in a Few-GeV Neutrino Beam. Phys. Rev. D, 83:012005, 2011.

[82] A.A. Aguilar-Arevalo et al. First measurement of the muon antineutrino

double-differential charged-current quasielastic cross section. Phys. Rev. D,
88(3):032001, 2013.

206
[83] K. Abe et al. Measurement of the 𝜈𝜇 charged-current quasielastic cross section
on carbon with the ND280 detector at T2K. Phys. Rev. D, 92(11):112003, 2015.

[84] M.F. Carneiro et al. High-Statistics Measurement of Neutrino Quasielastic-

Like Scattering at 𝐸𝜈 ∼ 6 GeV on a Hydrocarbon Target. Phys. Rev. Lett.,
124(12):121801, 2020.

[85] P. Abratenko et al. First Measurement of Inclusive Muon Neutrino Charged

Current Differential Cross Sections on Argon at 𝐸𝜈 ∼0.8 GeV with the Micro-
BooNE Detector. Phys. Rev. Lett., 123(13):131801, 2019.

[86] G.A. Fiorentini et al. Measurement of Muon Neutrino Quasielastic Scattering

on a Hydrocarbon Target at 𝐸𝜈 ∼ 3.5 GeV. Phys. Rev. Lett., 111:022502, 2013.

[87] M. Betancourt et al. Direct Measurement of Nuclear Dependence of Charged

Current Quasielasticlike Neutrino Interactions Using MINER𝜈A. Phys. Rev.
Lett., 119(8):082001, 2017.

[88] T. Walton et al. Measurement of muon plus proton final states in 𝜈𝜇 interactions
on hydrocarbon at ⟨𝐸𝜈 ⟩ = 4.2 GeV. Phys. Rev. D, 91(7):071301, 2015.

[89] K. Abe et al. Characterization of nuclear effects in muon-neutrino scattering

on hydrocarbon with a measurement of final-state kinematics and correlations
in charged-current pionless interactions at T2K. Phys. Rev. D, 98(3):032003,
2018.

[90] U. Mosel et al. Energy reconstruction in the Long-Baseline Neutrino Experi-

ment. Phys. Rev. Lett., 112:151802, 2014.

[91] J.A. Formaggio and G.P. Zeller. From eV to EeV: Neutrino Cross Sections
Across Energy Scales. Rev. Mod. Phys., 84:1307–1341, 2012.

[92] P. Abratenko et al. First measurement of differential charged current quasielas-

ticlike 𝜈𝜇 -argon scattering cross sections with the microboone detector. Phys.
Rev. Lett., 125:201803, Nov 2020.

[93] R. Acciarri et al. Design and Construction of the MicroBooNE Detector. J.

Instrum., 12(02):P02017, 2017.

[94] M. Antonello et al. A Proposal for a Three Detector Short-Baseline Neutrino

Oscillation Program in the Fermilab Booster Neutrino Beam. arXiv:1503.01520,
2015.

[95] F. Tortorici, V. Bellini, and C.M. Sutera. Upgrade of the ICARUS T600 Time
Projection Chamber. J. Phys. Conf. Ser., 1056(1):012057, 2018.

[96] Babak Abi et al. Deep Underground Neutrino Experiment (DUNE),

Far Detector Technical Design Report, Volume I Introduction to DUNE.
arXiv:2002.02967, 2 2020.

207
 [97] Babak Abi et al. Deep Underground Neutrino Experiment (DUNE), Far De-
tector Technical Design Report, Volume II DUNE Physics. arXiv:2002.03005,
2 2020.

[98] Babak Abi et al. Deep Underground Neutrino Experiment (DUNE), Far De-
tector Technical Design Report, Volume III DUNE Far Detector Technical Co-
ordination. arXiv:2002.03008, 2 2020.

[99] Babak Abi et al. Deep Underground Neutrino Experiment (DUNE), Far Detec-
tor Technical Design Report, Volume IV Far Detector Single-phase Technology.
arXiv:2002.03010, 2 2020.

[100] P. Abratenko et al. Determination of muon momentum in the MicroBooNE

LArTPC using an improved model of multiple Coulomb scattering. J. Instrum.,
12(10):P10010, 2017.

[101] D. Kaleko et al. PMT Triggering and Readout for the MicroBooNE Experiment.
J. Instrum., 8:C09009, 2013.

[102] C. Adams et al. Calibration of the charge and energy loss per unit length of the
MicroBooNE liquid argon time projection chamber using muons and protons.
J. Instrum., 15(03):P03022, 2020.

[103] C. Andreopoulos et al. The genie neutrino monte carlo generator. Nucl. Instrum.
Meth. A, 614:87–104, 2010.

[104] Costas Andreopoulos, Christopher Barry, Steve Dytman, Hugh Gallagher,

Tomasz Golan, Robert Hatcher, Gabriel Perdue, and Julia Yarba. The GENIE
Neutrino Monte Carlo Generator: Physics and User Manual. arXiv:1510.05494,
2015.

[105] B.P. Roe. Statistical errors in Monte Carlo estimates of systematic errors. Nucl.
Instrum. Meth. A, 570:159–164, 2007.

[106] R. Pordes and E. Snider. The Liquid Argon Software Toolkit (LArSoft): Goals,
Status and Plan. PoS, ICHEP2016:182, 2016.

[107] E. Snider and G. Petrillo. LArSoft: Toolkit for Simulation, Reconstruction

and Analysis of Liquid Argon TPC Neutrino Detectors. J. Phys. Conf. Ser.,
898(4):042057, 2017.

[108] C.H. Llewellyn Smith. Neutrino Reactions at Accelerator Energies. Phys. Rep.,
3:261–379, 1972.

[109] Teppei Katori. Meson Exchange Current (MEC) Models in Neutrino Interaction
Generators. AIP Conf. Proc., 2015.

[110] D. Rein and L. Sehgal. Neutrino Excitation of Baryon Resonances and Single
Pion Production. Ann. Phys. (N.Y.), 133:79–153, 1981.

208
[111] S.G. Mashnik et al. CEM03 and LAQGSM03: New modeling tools for nuclear
applications. J. Phys. Conf. Ser., 41:340–351, 2006.
[112] P. Stowell, C. Wret, C. Wilkinson, L. Pickering, S. Cartwright, Y. Hayato,
K. Mahn, K.S. McFarland, J. Sobczyk, R. Terri, L. Thompson, M.O. Wascko,
and Y. Uchida. NUISANCE: a neutrino cross-section generator tuning and
comparison framework. Journal of Instrumentation, 12(01):P01016–P01016,
jan 2017.
[113] T. Golan et al. NuWro: the Wroclaw Monte Carlo Generator of Neutrino
Interactions. Nucl.Phys.Proc.Suppl., 499:229–232, 2012.
[114] Y. Hayato. A neutrino interaction simulation program library NEUT. Acta
Phys. Polon., B40:2477, 2009.
[115] J. Nieves, J. E. Amaro, and M. Valverde. Inclusive quasielastic charged-current
neutrino-nucleus reactions. Phys. Rev. C, 70:055503, Nov 2004.
[116] Jonathan Engel. Approximate treatment of lepton distortion in charged current
neutrino scattering from nuclei. Phys. Rev. C, 57:2004–2009, 1998.
[117] U. Mosel. Neutrino event generators: foundation, status and future. Phys. Rev.
G, 2019.
[118] A. Bodek et al. Neutrino Quasielastic Scattering on Nuclear Targets:
Parametrizing Transverse Enhancement (Meson Exchange Currents). Eur.
Phys. J. C, 71:1726, 2011.
[119] K.M. Graczyk et al. C(5)**A axial form factor from bubble chamber experi-
ments. Phys. Rev. D, 80:093001, 2009.
[120] C. Berger and L. Sehgal. PCAC and coherent pion production by low energy
neutrinos. Phys. Rev. D, 79:053003, 2009.
[121] J. Nieves, F. Sanchez, I. Ruiz Simo, and M.J. Vicente Vacas. Neutrino Energy
Reconstruction and the Shape of the CCQE-like Total Cross Section. Phys.
Rev. D, 85:113008, 2012.
[122] J. Schwehr, D. Cherdack, and R. Gran. GENIE implementation of IFIC Valen-
cia model for QE-like 2p2h neutrino-nucleus cross section. arXiv, 1 2016.
[123] J. A. Nowak. Four Momentum Transfer Discrepancy in the Charged Current 𝜋 +
Production in the MiniBooNE: Data vs. Theory. AIP Conf. Proc., 1189(1):243–
248, 2009.
[124] K. Kuzmin et al. Lepton polarization in neutrino nucleon interactions. Phys.
Part. Nucl., 35:S133–S138, 2004.
[125] Ch. Berger and L.M. Sehgal. Lepton mass effects in single pion production by
neutrinos. Phys. Rev. D, 76:113004, 2007.

209
[126] K. M. Graczyk and J. T. Sobczyk. Form Factors in the Quark Resonance Model.
Phys. Rev. D, 77:053001, 2008. [Erratum: Phys.Rev.D 79, 079903 (2009)].

[127] Tina Leitner, L. Alvarez-Ruso, and U. Mosel. Charged current neutrino nucleus
interactions at intermediate energies. Phys. Rev. C, 73:065502, 2006.

[128] Ulrich Mosel. Neutrino event generators: foundation, status and future. J.
Phys. G, 46(11):113001, 2019.

[129] Torbjorn Sjostrand, Stephen Mrenna, and Peter Z. Skands. PYTHIA 6.4
Physics and Manual. JHEP, 05:026, 2006.

[130] MicroBooNE Collaboration. Novel approach for evaluating detector-related

uncertainties in a lartpc using microboone data. arXiv:2111.03556, 2021.

[131] P. Abratenko et al. New CC0𝜋 genie model tune for microboone. Phys. Rev.
D, 105:072001, Apr 2022.

[132] X.-G. Lu et al. Measurement of final-state correlations in neutrino muon-proton

mesonless production on hydrocarbon at ev=3gev. Phys. Rev. Lett., 121:022504,
Jul 2018.

[133] K. et al Abe. First t2k measurement of transverse kinematic imbalance in

the muon-neutrino charged-current single-𝜋 + production channel containing at
least one proton. Phys. Rev. D, 103:112009, Jun 2021.

[134] S. Dolan. Exploring nuclear effects with transverse imbalances.

https://indico.ectstar.eu/event/19/contributions/409/attachments/313/414/sdolanTalk.pdf.

[135] T. et al Cai. Nucleon binding energy and transverse momentum imbalance in

neutrino-nucleus reactions. Phys. Rev. D, 101:092001, May 2020.

[136] Table 289: muons in liquid argon. http://pdg.lbl.gov/2012/

AtomicNuclearProperties/MUON_ELOSS_TABLES/muonloss_289.pdf.

[137] Stopping powers and ranges for protons and al-

pha particles. https://www.nist.gov/pml/
stopping-power-range-tables-electrons-protons-and-helium-ions.

[138] Classification of track-like particles in MicroBooNE - Paper Draft,

https://arxiv.org/abs/2109.02460.

[139] W. Tang, X. Li, X. Qian, H. Wei, and C. Zhang. Data unfolding with wiener-svd
method. Journal of Instrumentation, 12(10):P10002–P10002, Oct 2017.

[140] Andreas Höcker and Vakhtang Kartvelishvili. Svd approach to data unfolding.
Nuclear Instruments and Methods in Physics Research Section A: Accelerators,
Spectrometers, Detectors and Associated Equipment, 372(3):469–481, Apr 1996.

210
[141] S Schmitt. Tunfold, an algorithm for correcting migration effects in high energy
physics. Journal of Instrumentation, 7(10):T10003–T10003, Oct 2012.

[142] BaBar Statistics Working Group. Recommended statistical procedures for

babar. BABAR Analysis Document 318, 2002.

[143] D. Ashery, I. Navon, G. Azuelos, H.K. Walter, H.J. Pfeiffer, and F.W. Schleputz.
True Absorption and Scattering of Pions on Nuclei. Phys. Rev. C, 23:2173–2185,
1981.

[144] L. A. Harewood and R. Gran. Elastic hadron-nucleus scattering in neutrino-

nucleus reactions and transverse kinematics measurements. arXiv:1906.10576,
2019.

[145] D. H. Wright and M. H. Kelsey. The Geant4 Bertini Cascade. Nucl. Instrum.
Meth. A, 804:175–188, 2015.

[146] K. Abe et al. The DUNE Far Detector Interim Design Report Volume 1:
Physics, Technology and Strategies. arXiv:1807.10334, 2018.

[147] M. A. Acero et al. First measurement of neutrino oscillation parameters using

neutrinos and antineutrinos by nova. Phys. Rev. Lett., 123:151803, Oct 2019.

[148] Babak Abi et al. https://home.fnal.gov/ ljf26/DUNEFluxes/.

[149] Artur M. Ankowski and Alexander Friedland. Assessing the accuracy of the GE-
NIE event generator with electron-scattering data. Phys. Rev. D, 102(5):053001,
2020.

[150] J E Amaro, M B Barbaro, J A Caballero, R González-Jiménez, G D Megias,

and I Ruiz Simo. Electron- versus neutrino-nucleus scattering. J. Phys. G,
47(12):124001, nov 2020.

[151] P. Barreau et al. Deep Inelastic electron Scattering from Carbon. Nucl. Phys.
A402, pages 515–540, 1983.

[152] P.E. Bosted and V. Mamyan. Empirical Fit to electron-nucleus scattering.

arXiv:1203.2262, 2012.

[153] Steven Dytman, Yoshinari Hayato, Roland Raboanary, Jan Sobczyk, Julia
Tena Vidal, and Narisoa Vololoniaina. Comparison of Validation Methods of
Simulations for Final State Interactions in Hadron Production Experiments.
arXiv2103.07535, 2021.

[154] O. Hen, L. B. Weinstein, E. Piasetzky, G. A. Miller, M. M. Sargsian, and

Y. Sagi. Correlated fermions in nuclei and ultracold atomic gases. Phys. Rev.
C, 92:045205, Oct 2015.

211
[155] O. Hen, G. A. Miller, E. Piasetzky, and L. B. Weinstein. Nucleon-Nucleon
Correlations, Short-lived Excitations, and the Quarks Within. Rev. Mod. Phys.,
89(4):045002, 2017.

[156] A Bodek and U K Yang. Higher twist, 𝜉𝑤 scaling, and effective LO PDFs for
lepton scattering in the few GeV region. J. Phys. G, 29(8):1899–1905, jul 2003.

[157] R. Bradford, A. Bodek, Howard Scott Budd, and J. Arrington. A New pa-
rameterization of the nucleon elastic form-factors. Nucl. Phys. B, Proc. Suppl.,
159:127–132, 2006.

[158] J. Nieves, I. Ruiz Simo, and M. J. Vicente Vacas. Inclusive charged-current

neutrino-nucleus reactions. Phys. Rev. C, 83:045501, Apr 2011.

[159] G. D. Megias, J. E. Amaro, M. B. Barbaro, J. A. Caballero, and T. W. Don-

nelly. Inclusive electron scattering within the susav2 meson-exchange current
approach. Phys. Rev. D, 94:013012, Jul 2016.

[160] G.D. Megias, J.E. Amaro, M.B. Barbaro, J.A. Caballero, T.W. Donnelly, and
I. Ruiz Simo. Charged-current neutrino-nucleus reactions within the superscal-
ing meson-exchange current approach. Phys. Rev. D, 94(9):093004, 2016.

[161] J.A. Caballero. General study of superscaling in quasielastic (e,e’) and (nu, mu)
reactions using the relativistic impulse approximation. Phys. Rev. C, 74:015502,
2006.

[162] R. González-Jiménez, A. Nikolakopoulos, N. Jachowicz, and J.M. Udías. Nu-

clear effects in electron-nucleus and neutrino-nucleus scattering within a rela-
tivistic quantum mechanical framework. Phys. Rev. C, 100(4):045501, 2019.

[163] R. González-Jiménez, M.B. Barbaro, J.A. Caballero, T.W. Donnelly, N. Ja-

chowicz, G.D. Megias, K. Niewczas, A. Nikolakopoulos, and J.M. Udías. Con-
straints in modeling the quasielastic response in inclusive lepton-nucleus scat-
tering. Phys. Rev. C, 101(1):015503, 2020.

[164] J. Schwehr, D. Cherdack, and R. Gran. GENIE implementation of IFIC Valen-

cia model for QE-like 2p2h neutrino-nucleus cross section. arXiv:1601.02038,
2016.

[165] I Ruiz Simo, J E Amaro, M B Barbaro, A De Pace, J A Caballero, and T W

Donnelly. Relativistic model of 2p-2h meson exchange currents in (anti)neutrino
scattering. Journal of Physics G: Nuclear and Particle Physics, 44(6):065105,
apr 2017.

[166] I. Ruiz Simo, J.E. Amaro, M.B. Barbaro, A. De Pace, J.A. Caballero, G.D.
Megias, and T.W. Donnelly. Emission of neutron–proton and proton–proton
pairs in neutrino scattering. Physics Letters B, 762:124–130, 2016.

212
[167] T. W. Donnelly and Ingo Sick. Superscaling in inclusive electron-nucleus scat-
tering. Phys. Rev. Lett., 82:3212–3215, Apr 1999.

[168] T. W. Donnelly and Ingo Sick. Superscaling of inclusive electron scattering

from nuclei. Phys. Rev. C, 60:065502, Nov 1999.

[169] M. B. Barbaro, J. A. Caballero, A. De Pace, T. W. Donnelly, R. González-

Jiménez, and G. D. Megias. Mean-field and two-body nuclear effects in inclusive
electron scattering on argon, carbon, and titanium: The superscaling approach.
Phys. Rev. C, 99:042501, Apr 2019.

[170] G. D. Megias, J. E. Amaro, M. B. Barbaro, J. A. Caballero, and T. W. Don-

nelly. Inclusive electron scattering within the susav2 meson-exchange current
approach. Phys. Rev. D, 94:013012, Jul 2016.

[171] G. D. Megias, J. E. Amaro, M. B. Barbaro, J. A. Caballero, T. W. Donnelly,

and I. Ruiz Simo. Charged-current neutrino-nucleus reactions within the su-
perscaling meson-exchange current approach. Phys. Rev. D, 94:093004, Nov
2016.

[172] G D Megias, M B Barbaro, J A Caballero, J E Amaro, T W Donnelly, I Ruiz

Simo, and J W Van Orden. Neutrino–oxygen CC0π scattering in the SuSAv2-
MEC model. Journal of Physics G: Nuclear and Particle Physics, 46(1):015104,
dec 2018.

[173] G. D. Megias, M. B. Barbaro, J. A. Caballero, and S. Dolan. Analysis of the

minerva antineutrino double-differential cross sections within the susav2 model
including meson-exchange currents. Phys. Rev. D, 99:113002, Jun 2019.

[174] G.D. Megias et al. Meson-exchange currents and quasielastic predictions for
charged-current neutrino-12 𝐶 scattering in the superscaling approach. Phys.
Rev. D, 91(7):073004, 2015.

[175] J. E. Amaro, M. B. Barbaro, J. A. Caballero, A. De Pace, T. W. Donnelly,

G. D. Megias, and I. Ruiz Simo. Density dependence of 2p-2h meson-exchange
currents. Phys. Rev. C, 95:065502, Jun 2017.

[176] R. P. Feynman, M. Kislinger, and F. Ravndal. Current matrix elements from a

relativistic quark model. Phys. Rev. D, 3:2706–2732, Jun 1971.

[177] T. Yang, C. Andreopoulos, H. Gallagher, K. Hofmann, and P. Kehayias. A

hadronization model for few-GeV neutrino interactions. Eur. Phys. J. C,
63(1):1–10, 2009.

[178] Torbjörn Sjöstrand, Stephen Mrenna, and Peter Skands. PYTHIA 6.4 physics
and manual. J. High Energy Phys. 06 (2006) 026, 2006.

[179] J.Tena-Vidal et al. Neutrino-nucleon cross-section model tuning in genie v3.

213
[180] S.A. Dytman and A.S. Meyer. Final state interactions in genie. AIP Conf.
Proc., 1405:213, 2011.

[181] S. G. Mashnik, A. J. Sierk, K. K. Gudima, and M. I. Baznat. Cem03 and

laqgsm03: New modelling tools for nuclear applications. J. Phys. Conf. Ser.,
41:340, 2006.

[182] R. M. Sealock, K. L. Giovanetti, S. T. Thornton, Z. E. Meziani, O. A. Rondon-

Aramayo, S. Auffret, J.-P. Chen, D. G. Christian, D. B. Day, J. S. McCarthy,
R. C. Minehart, L. C. Dennis, K. W. Kemper, B. A. Mecking, and J. Morgen-
stern. Electroexcitation of the ∆(1232) in nuclei. Phys. Rev. Lett., 62:1350–
1353, Mar 1989.

[183] H. Dai et al. First Measurement of the Ti(𝑒, 𝑒′ )X Cross Section at Jefferson
Lab. Phys. Rev. C, 98(1):014617, 2018.

[184] D.B. Day et al. Inclusive electron nucleus scattering at high momentum transfer.
Phys. Rev. C, 48:1849–1863, 1993.
′
[185] H. Dai et al. First measurement of the Ar(𝑒, 𝑒 )𝑥 cross section at jefferson
laboratory. Phys. Rev. C, 99:054608, May 2019.

[186] M. I. Niculescu. Inclusive resonance electroproduction data from hydrogen and

deuterium and studies of quark-hadron duality, Ph.D. thesis, Hampton Univer-
sity, 2009.

[187] I. Niculescu et al. Experimental verification of quark-hadron duality. Phys.

Rev. Lett., 85:1186–1189, Aug 2000.

[188] S. P. Malace. Measurements of inclusive resonance cross sections for quark-

hadron duality studies, Ph.D. thesis, Hampton University, 2006.

[189] S. Stein, W. B. Atwood, E. D. Bloom, R. L. A. Cottrell, H. DeStaebler, C. L.

Jordan, H. G. Piel, C. Y. Prescott, R. Siemann, and R. E. Taylor. Electron
scattering at 4° with energies of 4.5-20 gev. Phys. Rev. D, 12:1884–1919, Oct
1975.

[190] M. E. Christy and Peter E. Bosted. Empirical fit to precision inclusive electron-
proton cross- sections in the resonance region. Phys. Rev. C, 81:055213, 2010.

[191] P. E. Bosted and M. E. Christy. Empirical fit to inelastic electron-deuteron

and electron-neutron resonance region transverse cross-sections. Phys. Rev. C,
77:065206, 2008.

[192] D. Drechsel, O. Hanstein, S.S. Kamalov, and L. Tiator. A unitary isobar model
for pion photo- and electroproduction on the proton up to 1 gev. Nucl. Phys.,
A645:145, 1999.

214
[193] H. Kamano, S. X. Nakamura, T. S. H. Lee, and T. Sato. Isospin decomposition
of 𝛾𝑁 → 𝑁 * transitions within a dynamical coupled-channels model. Phys.
Rev. C, 94(1):015201, 2016.

[194] S. X. Nakamura, H. Kamano, T. S. H. Lee, and T. Sato. Neutrino-induced

meson productions off nucleon at forward limit in nucleon resonance region.
AIP Conf. Proc., 1663(1):070005, 2015.

[195] Accelerator Science. https://www.jlab.org/accelerator.

[196] Christoph W. Leemann, David R. Douglas, and Geoffrey A. Krafft. The

continuous electron beam accelerator facility: Cebaf at the jefferson laboratory.
https://www.jlab.org/div_dept/physics_division/talks/Background/Accelerator/CEBAF_Ann_

[197] A J Street, J S.H. Ross, and S M Harrison. Final site assembly and testing of the
superconducting toroidal magnet for the cebaf large acceptance spectrometer
(clas). IEEE Transactions on Magnetics, 32, 7 1996.

[198] M.D Mestayer, D.S Carman, Burin Asavapibhop, F.J. Barbosa, P Bonneau,
S.B Christo, G.E Dodge, T Dooling, W.S Duncan, S.A Dytman, R Feuerbach,
Gerard Gilfoyle, Vardan Gyurjyan, K.H Hicks, R.S Hicks, C.E Hyde-Wright,
G Jacobs, Andi Klein, F.J Klein, and Jaycee Yun. The clas drift chamber
system. Nuclear Instruments and Methods in Physics Research Section A: Ac-
celerators, Spectrometers, Detectors and Associated Equipment, 449:81–111, 07
2000.

[199] M. Amarian et al. The clas forward electromagnetic calorimeter. Nuclear Instru-
ments and Methods in Physics Research Section A: Accelerators, Spectrometers,
Detectors and Associated Equipment, 460:239–265, 2001.

[200] G. Adams et al. The CLAS Cherenkov detector. Nucl. Instrum. Meth. A,
465:414–427, 2001.

[201] E. Smith et al. The time-of-flight system for CLAS. Nucl. Instrum. Meth. A,
432:265–298, 1999.

[202] M. Osipenko et al. Measurement of the Nucleon Structure Function F2 in the

Nuclear Medium and Evaluation of its Moments. Nucl. Phys. A, 845:1–32, 2010.

[203] K. S. Egiyan et al. Measurement of 2- and 3-nucleon short range correlation

probabilities in nuclei. Phys. Rev. Lett., 96:082501, 2006.

[204] D. Protopopescu et al. Survey of A(LT-prime) asymmetries in semi-exclusive

electron scattering on He-4 and C-12. Nucl. Phys. A, 748:357–373, 2005.

[205] A. V. Stavinsky et al. Proton source size measurements in the eA —> e-prime
ppX reaction. Phys. Rev. Lett., 93:192301, 2004.

215
[206] R. A. Niyazov et al. Two nucleon momentum distributions measured in He-
3(e,e-prime pp)n. Phys. Rev. Lett., 92:052303, 2004. [Erratum: Phys.Rev.Lett.
92, 099903 (2004)].

[207] K. S. Egiyan et al. Observation of nuclear scaling in the A(e, e-prime) reaction
at x(B) greater than 1. Phys. Rev. C, 68:014313, 2003.

[208] M. Khachatryan, A. Papadopoulou, et al. Electron-beam energy reconstruction

for neutrino oscillation measurements. Nature, 599:565, 2021.

[209] B. A. Mecking et al. The CEBAF Large Acceptance Spectrometer (CLAS).

Nucl. Instrum. Meth., A503:513–553, 2003.

[210] K. Abe et al. Improved constraints on neutrino mixing from the t2k experiment
with 3.13 × 1021 protons on target, 2021.

[211] L. Aliaga et al. Design, calibration, and performance of the minerva detector.
Nucl. Instrum. Methods, A743:130 – 159, 2014.

[212] K. Abe et al. Hyper-Kamiokande Design Report. arXiv:1805.04163, 2018.

[213] R. Acciarri et al. Long-Baseline Neutrino Facility (LBNF) and Deep Under-
ground Neutrino Experiment (DUNE). arXiv:1512.06148, 2015.

[214] O. Hen et al. Momentum sharing in imbalanced Fermi systems. Science,

346:614–617, 2014.

[215] P. Abratenko et al. First Measurement of Inclusive Muon Neutrino Charged

Current Differential Cross Sections on Argon at 𝐸𝜈 ∼0.8 GeV with the Micro-
BooNE Detector. Phys. Rev. Lett., 123(13):131801, 2019.

[216] Simc Monte Carlo. Simc Monte Carlo, https://hallcweb.jlab.org/wiki/ in-

dex.php/SIMC_Monte_Carlo, 2020.

[217] Luke W. Mo and Yung-Su Tsai. Radiative Corrections to Elastic and Inelastic
e p and mu p Scattering. Rev. Mod. Phys., 41:205–235, 1969.

[218] R. Ent, B. W. Filippone, N. C. R. Makins, R. G. Milner, T. G. O’Neill, and

′
D. A. Wasson. Radiative corrections for (𝑒, 𝑒 𝑝) reactions at gev energies. Phys.
Rev. C, 64:054610, Oct 2001.

[219] R. Cruz-Torres et al. Comparing proton momentum distributions in 𝐴 = 2 and

3 nuclei via 2 H 3 H and 3 He (𝑒, 𝑒′ 𝑝) measurements. Phys. Lett. B, 797:134890,
2019.

[220] Teppei Katori and Marco Martini. Neutrino–nucleus cross sections for oscilla-
tion experiments. J. Phys. G, 45(1):013001, 2018.

216
[221] N. Markov et al. Exclusive 𝜋 0 𝑝 electroproduction off protons in the resonance
region at photon virtualities 0.4 gev2 ≤ 𝑄2 ≤ 1 gev2 . Phys. Rev. C, 101:015208,
Jan 2020.

[222] M. Osipenko. A Kinematically Complete Measurement of the Proton Structure

Function 𝐹2 in the Resonance Region and Evaluation of its Moments. PhD
thesis, Moscow State University, 2002.

[223] X.-G. Lu, L. Pickering, S. Dolan, G. Barr, D. Coplowe, Y. Uchida, D. Wark,

M. O. Wascko, A. Weber, and T. Yuan. Measurement of nuclear effects in
neutrino interactions with minimal dependence on neutrino energy. Phys. Rev.
C, 94:015503, Jul 2016.

[224] K. Abe et al. Characterization of nuclear effects in muon-neutrino scattering

on hydrocarbon with a measurement of final-state kinematics and correlations
in charged-current pionless interactions at t2k. Phys. Rev. D, 98:032003, Aug
2018.

[225] A. Bodek and T. Cai. Removal energies and final state interaction in lepton
nucleus scattering. Eur. Phys. J. C 79, 293, 2019.

[226] K. Abe et al. Hyper-Kamiokande Design Report. arXiv:1805.04163, 2018.

[227] e4𝜈 collaboration. Electrons for neutri-

nos: Addressing critical neutrino-nucleus issues.
https://misportal.jlab.org/pacProposals/proposals/1377/attachments/104415/Proposal.pdf.

[228] Andrew P. Furmanski and Jan T. Sobczyk. Neutrino energy reconstruction

from one-muon and one-proton events. Phys. Rev. C, 95:065501, Jun 2017.

[229] Golub and Van Loan (1996, p. 143), Horn and Johnson (1985, p. 407), Trefethen
and Bau (1997, p. 174).

[230] Banerjee, Sudipto; Roy, Anindya (2014), Linear Algebra and Matrix Analysis
for Statistics, Texts in Statistical Science (1st ed.), Chapman and Hall/CRC,
ISBN 978-1420095388.

[231] A. Hocker and V. Kartvelishvili, SVD approach to data unfolding, Nucl. In-
strum. Meth. A 372 (1996) 469 [hep-ph/9509307].

[232] Rustam Niyazov. Measurement of Correlated Pair Momentum Distributions on

3
𝐻𝑒(e,e’pp)n with CLAS. PhD thesis, Old Dominion University, 2003.

[233] Dan Protopopescu. Fiducial cuts for electrons in the CLAS/E2 data at 4.4 GeV.
CLAS-NOTE 2000-007 JLAB, 2000.

[234] Bin Zhang. Electron fiducial cuts. https://www.jlab.org/Hall-B/secure/

e2/bzh/efiducialcut.html, 2003.

217
[235] M. Khachatryan and L. Weinstein. Validation of neutrino energy estimation
using electron scattering data. Technical report, Old Dominion University,
November 2019. CLAS-Note.

[236] Steven McLauchlan. ∆ electroproduction in 12

𝐶. PhD thesis, University of
Glasgow, January 2003. Appendix C.

[237] Lawrence Weinstein Rustam Niyazov. Fiducial cut for positive hadrons in
CLAS/E2 data at 4.4 GeV. CLAS-NOTE 2001-013 JLAB, 2001.

218