NuXTract 2023 - Towards a consensus in neutrino cross sections

2 Oct 2023, 09:00 → 6 Oct 2023, 17:00 Europe/Zurich

222/R-001 (CERN)

Workshop photo:

Thanks everyone for a great workshop, see you again soon!

---

We are thrilled to invite you to NuXTract, a new in-person NuSTEC workshop scheduled for the week of October 2nd 2023 at CERN from Monday 2nd 2pm until Friday 6th 1pm. 

This special workshop will be dedicated to open issues in neutrino cross-section extraction, and data presentation and preservation efforts. It is an excellent opportunity for experiments to discuss how they present data, for theorists to specify their needs from a data release, and for all of us to dive into the statistical challenges in cross-section extraction (e.g. unfolding, regularisation methods, efficiency corrections, background constraints). 

The agenda will be composed of invited talks, selected abstracts, and lots of discussions. 

A google doc is available in this link to all participants to share their thoughts, questions, discussion points 

We have high hopes and wish to come out of this workshop with a set of NuSTEC best-practice guidelines for how to extract a cross-section measurement, and start a neutrino cross-section database with comparable results we know how to interpret.

The workshop will bring together both experts and early career scientists to share experiences with cross-section extraction challenges and use of experimental data releases. We will also hear about data preservation efforts and novel approaches to cross-section extractions (e.g. forward folding or ML methods). 

If you wish to take part, please mark your schedules, be sure to register and to submit an abstract using the links on the left of this page if you wish to attend and/or present anything.

We hope to see many of you there!

Fees

We are planning on hosting a social buffet on Thursday evening as well as providing coffee and snacks. To cover this, we will ask for a small attendance fee of 100 CHF for those attending in person. Please pay the using this form. 

Accommodation

CERN has a hotel on site close to our workshop rooms. We have booked 50 of these rooms for the workshop that will be available until 30 days before the workshop. If you want to book one of these, please use this booking form. Leave the team account / budget code part empty. If you have any trouble accessing the form or booking a room please contact the organisers.

Abstract submission and registration deadline: August 31st.

---

The organising committee:   
Artur Ankowski   
Adi Ashkenazi   
Pablo Barham (LOC)   
Stephen Dolan (LOC)   
Laura Munteanu (LOC)   
Vishvas Pandey   
Joanna Sobczyk   
Clarence Wret   
CERN Neutrino Secretariat: Antonella and Elena

Monday 2 October

13:30 → 14:00 Welcome Coffee Break

14:00 → 15:30 Welcome and introduction

Convener: Dr Stephen Dolan (CERN)

14:00 Welcome

NuXTract intro.pdf

slides (2).pdf

14:15 Introduction: T2K

Speaker: Sam Jenkins (University of Liverpool)

T2Kintro@NuXtract.pdf

14:20 Introduction: MINERvA

Speaker: Noah Vaughan (Oregon State University)

NHVaughan_NuXTract_MnvIntroTalk_final.pdf

14:25 Introduction: NOvA

Speaker: Gregory Pawloski

NuXtract-NOvA-QuickFire.pdf

14:30 Introduction: MicroBooNE

Speaker: Afroditi Papadopoulou

NuXTract23_Papadopoulou_Oct2_2023_MicroBooNE_Intro.pdf

14:35 Introduction: ProtoDUNE-SP

Speaker: Richard Diurba (University of Bern)

ProtoDUNESPNuSTEC2023Intro.pdf

14:40 Introduction: NINJA

Speaker: Hitoshi Oshima (ICRR, the University of Tokyo)

NuXTract23_QuickFire_NINJA_oshima_v2.pdf

14:45 Introduction: ICARUS

Speaker: Dr Stephen Dolan (CERN)

NUXTractIntro.pdf

14:50 Introduction: SBND

Speaker: Dr Vishvas Pandey (Fermi National Accelerator Laboratory)

NuXTract2023_SBND_VPandey.pdf

14:55 Discussion

15:30 → 16:00 Coffee break

16:00 → 18:00 General overview

Convener: Dr Vishvas Pandey (Fermi National Accelerator Laboratory)

16:00 What do next-generation experiments need

Speaker: Callum David Wilkinson (Lawrence Berkeley National Lab. (US))

NuXTract_Oct23.pdf

16:40 Statistical methods: past, present and future

Speaker: Lukas Koch (Johannes Gutenberg Universitaet Mainz (DE))

Stat_Methods.pdf

17:20 Discussion

Tuesday 3 October

09:00 → 10:00 General overview

Conveners: Dr Clarence Wret (University of Rochester (US)), Clarence Wret

09:00 Weiner SVD, conditional covariance formalism

Speaker: Xin Qian (Brookhaven National Laboratory (US))

Wiener-SVD Unfolding.pdf

Wiener-SVD Unfolding.pptx

10:00 → 10:30 Experiment Overview

Conveners: Clarence Wret, Dr Clarence Wret (University of Rochester (US))

10:00 NOvA’s approach to cross-section analyses and publishing results

Speakers: Gregory Pawloski, Gregory Pawloski (Fermi National Accelerator Laboratory)

NuXtract-NOvA.pdf

10:30 → 11:00 Coffee break

11:00 → 12:10 Experiment Overview

Conveners: Dr Clarence Wret (University of Rochester (US)), Clarence Wret, Laura Munteanu (CERN), Laura-Iuliana Munteanu (CERN)

11:00 MINERvA's experience of cross-section extraction methods

MINERvA is a dedicated neutrino-nucleus cross section experiment at Fermilab in the NuMI (anti)neutrino beamline, where it was exposed to both low and medium energy configurations with a neutrino energy spectrum ranging from a few GeV to tens of GeV. The large high statistics in the data are complemented by a large simulated sample, and their comparison allows MINERvA to probe nuclear structure and test interaction models important for oscillation experiments. Analyzers in MINERvA extract inclusive and exclusive cross sections with a comprehensive accounting of systematic uncertainty, making use of various methods of data driven background constraint, and reporting cross section data at the truth level by utilizing the D’Agostini method of iterative Bayesian unfolding. Other methods of unfolding have been proposed and explored (e.g., singular value decomposition) in MINERvA, however the D’Agostini method remains standard in the collaboration’s published results. An overview of the methods used in MINERvA cross section extraction and results will be presented.

Speaker: Mx Noah Vaughan (Oregon State University)

NHVaughan_NuXTract_MnvXSec_final.pdf

11:30 Cross-section extraction using template fitting in T2K cross-section measurements

Speakers: Margherita Buizza Avanzini (Centre National de la Recherche Scientifique (FR)), Margherita Buizza Avanzini

XsecFitter@NuXtract_021023.pdf

12:10 → 13:40 Lunch break

13:40 → 15:30 Methods

Convener: Adi Ashkenazi (Tel Aviv University (IL))

13:40 Binned log-likelihood template fitting with T2K

When performing neutrino cross section extractions, it is desirable to avoid dependence on the interaction model used in Monte-Carlo generation. Log-likelihood template fitting is a method which can be used to explore the input parameter space and find the best ways of describing data. This method has several advantages, including straightforward background constraining and informative diagnostics. This talk will explore the techniques used in a cross section extraction performed with a binned maximum log-likelihood method in the context of a new ongoing $\nu_{e}$CC single-pion production cross section analysis at T2K.

Speaker: Nick Latham

NuXTract23_fitting.pdf

14:00 Forward folding

Statistical methods

Speaker: Lukas Koch (Johannes Gutenberg Universitaet Mainz (DE))

ForwardFolding.pdf

14:30 Unbiased reconstruction of calorimetric variables in cross-section analyses

Neutrino cross-sections are often extracted purely in terms of lepton kinematics. In recent years more detailed analyses have been developed that additionally make use of kinematics in the hadronic system, which has proven very successful. However, even with new detector technologies of unparalleled precision, pattern recognition and reconstruction algorithms still require particle momenta above a given detection threshold. Calorimetric variables such as the hadronic energy provide alternative handles on the kinematics in the hadronic system and do not rely on a successful reconstruction of the hadron track. This talk will detail different methods to measure cross sections as a function of calorimetric variables in plastic scintillator based detectors and the problems and potential biases accompanying them.

Speaker: Katharina Lachner (University of Warwick)

unbiased_calorimetry_recon_for_xsec_analyses.pdf

14:50 Unfolding: Machine learning approaches

OmniFold: A Method to Simultaneously Unfold All Observables

Speaker: Ben Nachman (Lawrence Berkeley National Lab. (US))

NuXTract2023.pdf

15:10 Machine Learning-Assisted Unfolding for Neutrino Cross-section Measurements

The choice of unfolding method for a cross-section measurement is tightly coupled to the model dependence of the efficiency correction and the overall impact of cross-section modeling uncertainties in the analysis. A key issue is the dimensionality used, as the kinematics of all outgoing particles in an event typically affects the reconstruction performance in a neutrino detector. OmniFold is an unfolding method that iteratively reweights a simulated dataset, using machine learning to utilize arbitrarily high-dimensional information, that has previously been applied to collider and cosmology datasets. Here, we explore its use for neutrino physics using a public T2K near detector simulated dataset, and compare its performance with more traditional approaches, under a series of mock data studies.

Speaker: Andrew Cudd (University of Colorado Boulder)

nuxtract_omnifold_20231003.pdf

15:30 → 16:00 Coffee break

16:00 → 17:00 Methods

Convener: Adi Ashkenazi (Tel Aviv University (IL))

16:00 Unfolding: a statistician's perspective

Unfolding: a statistician's perspective

Speaker: Mikael Kuusela (Carnegie Mellon University (US))

NuXTract_Oct_2023.pdf

16:25 Treating Detector Systematics via a Likelihood Free Inference Method

Estimating the impact of systematic uncertainties in particle physics experiments is challenging, especially since the detector response is unknown analytically in most situations and needs to be estimated through Monte Carlo (MC) simulations. Typically, detector property variations are parameterized in ways that implicitly assume a specific physics model, which can introduce biases on quantities measured by an analysis. This talk presents a method to recover a model-independent, event-wise estimate of the detector response variation by applying a likelihood-free inference method to a set of MC simulations representing discrete detector realizations. The method provides a re-weighting scheme for every event, which can be used to apply the effects of detector property variations decoupled from the assumed physics model. We demonstrate the performance of the method on a simplified MC model of a neutrino oscillation experiment and show that it fully decouples the modeling of the detector response from the physics parameters to be measured in an MC forward-folding analysis.

Speaker: Ms Alexandra Trettin (University of Manchester)

NuXtract LLH free inference.key

NuXtract LLH free inference.pdf

Wednesday 4 October

09:00 → 10:30 Experiment Overview

Convener: Pablo Samuel Barham Alzas (CERN)

09:00 MicroBooNE's approach to exclusive cross-section measurements

Speaker: Afroditi Papadopoulou

NuXTract23_Papadopoulou_Oct_2023_XSecTechniques.pdf

09:30 MicroBooNE's approach to inclusive cross-section measurements

MicroBooNE is a liquid argon time projection chamber (LArTPC) located along the Fermilab Booster Neutrino Beam with a mean neutrino energy of approximately 0.8 GeV. The analysis of neutrino interactions leverage the mm level spatial resolutions of LArTPCs to provide a detailed description of the interaction in the detector. As a result, MicroBooNE has collected the largest neutrino-argon scattering dataset to date and can probe many questions related to the nuclear state and the interplay between different interaction pathways. A particular focus is the $\nu_{\mu}$-CC inclusive channel and it's energy dependence, especially in the GeV and sub-GeV regime. Understanding this is crucial for the next generation of neutrino oscillation experiments such as DUNE. At MicroBooNE, we introduce a novel conditional constraint approach that validates the hadronic model, in particular the mapping between reconstructed visible hadronic energy to true hadronic energy. This enables an extraction of neutrino energy-dependent cross-sections with suitable model discrimination power and a comprehensive measurement of the inclusive scattering channel. This provides a foundation from which one can begin to probe modelling strengths and weaknesses in relevant regions of phase space. This talk will outline such a complementary approach to extracting cross-sections and invite feedback on how such an analysis fits into the larger goal of understanding neutrino scattering in this energy regime.

Speaker: Bannanje Nitish Nayak (University of California Irvine (US))

xsec_nuxtract_final.pdf

10:00 Preview of new techniques for future MicroBooNE cross-section data releases

The MicroBooNE liquid argon time projection chamber experiment has a vibrant neutrino interaction physics program, with over thirty active analyses and a growing library of recent publications. As the precision and detail of MicroBooNE's cross-section measurements continue to improve, the collaboration is exploring new methods for maximizing the usefulness of the results for the nuclear modeling community. This talk will motivate and describe the technical implementation of these new methods, including a technique for reporting accurate covariances between measurements of different observables and potentially between measurements obtained from distinct analyses. We anticipate that these methods may be incorporated into the data release strategy for new cross-section measurements by other neutrino experiments in a relatively straightforward way.

Speaker: Steven Gardiner

NuXTract_talk.pdf

10:30 → 11:00 Coffee break

11:00 → 12:30 Data release

Convener: Dr Stephen Dolan (CERN)

11:00 Experimental data from a theoretical needs POV I

Speaker: Kajetan Niewczas

niewczas_231004_v2.pdf

11:30 Experimental data from a theoretical needs POV II

Speaker: Alexis Nikolakopoulos (Ghent University)

NuXTract_4_oct.pdf

12:30 → 14:00 Lunch break

14:00 → 15:15 Data release

Convener: Dr Vishvas Pandey (Fermi National Accelerator Laboratory)

14:00 Enabling BSM Studies With Neutrino Data

Speaker: Joachim Kopp (CERN)

nuxtract.pdf

14:30 The theory experience using experimental data

Speaker: Dr Guillermo D. Megias (University of Seville (Spain))

Link to slides

15:15 → 15:30 Experiment Overview

15:15 A modified slicing method with multi-dimensional unfolding to measure hadron-argon cross-section on ProtoDUNE-SP

Final-state interaction (FSI) is one of the most relevant nuclear effects involved in the neutrino-nucleus interactions, where the produced hadrons re-scatter with the nucleons within the nucleus. FSI accounts for a large source of uncertainties to neutrino detections, and hadron-nucleus scattering data can provide experimental constraints. Historically, experiments use hadron beam impinge on a thin target of material and measure its survival rate to extract the cross-section. However, in the case of liquid argon (LAr), which is used in many modern neutrino experiments, the large-size tank is not a thin-target in terms of hadrons. Therefore, the slicing method is proposed by the LArIAT collaboration, which hypothetically divide the LAr detector into several thin slices, and perform measurement in each slice. We further develop the method, deriving three energy-related variables in each event, and enable multi-dimensional unfolding of these variables, in order to fully consider the correlations among different slices. In this talk, the method as well as the procedures are applied to ProtoDUNE-SP Monte-Carlo (MC) sample. The consistency between the measured cross-section and the theoretical curve used for MC generation serves as a validation.

Speaker: Yinrui Liu (University of Chicago (US))

hadronXS_NuXTract_231004.pdf

15:30 → 16:00 Coffee break

16:00 → 18:00 Data release

Convener: Adi Ashkenazi (Tel Aviv University (IL))

16:00 Standardizing data releases

Speaker: Dr Luke Pickering (Royal Holloway, University of London)

NuXTract-data-releases.pdf

16:30 MINERvA's approach to data preservation

Since ending data collection in 2019 and the subsequent decommissioning of its detector, the MINERvA collaboration has turned its focus towards the long term preservation of its neutrino scattering data and tools for its analysis in parallel with the active ongoing analysis of that data. This preservation project includes the development of the MINERvA Analysis Toolkit (MAT) which centralizes how systematic uncertainties are handled and contains comprehensive cross section analysis tools open to the HEP community for use and development. Additionally, the data and simulated samples are processed into a standardized Master Ana Dev (MAD) tuple for use by analyzers. These standard tuples allow the total MINERvA data and simulation to be centralized and compact for long term use and storage. The MAT and MADTuples together represent the strategy for preserving MINERvA’s legacy, one that is open for the public to develop and use. An overview of this ongoing data preservation effort will be presented.

Speaker: Mx Noah Vaughan (Oregon State University)

NHVaughan_NuXTract_MnvDataPres_Final.pdf

17:00 Data release discussion

Thursday 5 October

09:00 → 10:00 Experiment Overview

Conveners: Laura Munteanu (CERN), Laura-Iuliana Munteanu (CERN)

09:00 Dealing with high dimensional efficiency corrections in T2K's cross-section measurements

Speakers: Sam Jenkins (University of Sheffield), Sam Jenkins (University of Liverpool)

HighDimensionalEffCorrection.pdf

09:30 Neutrino cross-section measurements in the NINJA experiment

The NINJA collaboration aims to study neutrino-nucleus interactions in the energy range of hundreds of MeV to a few GeV using an emulsion-based detector. A series of neutrino-nucleus interaction measurements was conducted using the emulsion detector with water and iron targets in the near detector hall of the T2K experiment at J-PARC. The emulsion detector is suitable for precision measurements of charged particles produced in neutrino interactions with a low momentum threshold, especially low momentum protons as low as 200$\,$MeV/$c$, thanks to its thin-layered structure and sub-$\mu$m spatial resolution. In the neutrino cross-section measurement, we have measured flux-averaged charged-current inclusive cross sections on iron both neutrino and anti-neutrino. In this talk, we will present the extraction methods and the results of cross section using the emulsion detector.

Speaker: Hitoshi Oshima (ICRR, the University of Tokyo)

NuXTract23_NINJA_oshima.pdf

10:00 → 10:30 Data release

10:00 Common tools (NuSystematics, GUNDAM, NUISANCE etc)

Speaker: Laura Munteanu (CERN)

Munteanu_NuXTract_CommonTools.pdf

10:30 → 11:00 Coffee break

11:00 → 12:00 Experiment Overview

Convener: Pablo Samuel Barham Alzas (CERN)

11:00 Measuring an Argon-Kaon Plus Cross Section by Unfolding Thin-Slice Targets

ProtoDUNE-SP is a single-phase liquid argon time projection chamber that took hadron test beam data in 2018. The test beam included positively charged kaons with test beam momenta of 6 GeV/c and 7 GeV/c, providing a sample to study kaons to benefit future DUNE proton decay and neutrino interaction studies with kaons in the final state. The total inelastic cross section of a positively charged kaon was measured at these test beam settings using the LArIAT thin-slice method of dividing the wires of the time projection chamber into target slices for calculating the cross section, which leverages the monolithic quality of liquid argon detectors. A Bayesian-like unfolding method using RooUnfold was applied to both the incident and interacting slice distributions to measure the cross sections at both test beam momenta. The talk will discuss the method of unfolding, optimization studies for unfolding, and applying systematic uncertainties using a LArIAT-style hadronic cross section using unfolding to extract a kaon total inelastic cross section.

Speaker: Richard Diurba (University of Bern)

ProtoDUNESPNuSTEC2023Thurs.pdf

11:25 Likelihood-fit-based Pion-Ar Cross Section Measurement Using ProtoDUNE-SP

ProtoDUNE-SP was a large-scale prototype of the single phase DUNE far detector which took test beam data in Fall 2018. The beam consisted of positive pions, kaons, muons, and protons, and this data is being used to measure the various hadron-Ar interaction cross sections. These measurements will provide important constraints for the nuclear ground state, final state interaction, and secondary interaction models of argon-based neutrino experiments. This talk will focus on the measurement of the pion-argon inelastic cross section broken down into three channels: absorption (no pions in the final state), charge exchange (a neutral pion in the final state), and other interactions. This measurement uses data collected with an central incident beam momentum of 1 GeV/c, and employs a likelihood fit as opposed to Bayesian unfolding to extract the cross sections.

Speaker: Jacob Michael Calcutt (Oregon State University (US))

NuXTract Talk-4.pdf

12:00 → 12:30 Fitting and tuning to data

Conveners: Dr Clarence Wret (University of Rochester (US)), Clarence Wret

12:00 GENIE’s fits to cross section data

Speaker: Julia Tena Vidal (Tel Aviv University (IL))

NuXTRACTTalk_JTenaVidal.pdf

12:30 → 12:50 Workshop photo, by the LHC beam pipe outside R1

Google maps: https://www.google.co.uk/maps/place/46%C2%B013'51.2%22N+6%C2%B003'18.8%22E/@46.2308899,6.0545893,19z/data=!3m1!4b1!4m4!3m3!8m2!3d46.230889!4d6.055233?entry=ttu

12:50 → 14:00 Lunch

14:00 → 15:30 Fitting and tuning to data

Conveners: Clarence Wret, Dr Clarence Wret (University of Rochester (US))

14:00 MicroBooNE Tuning

Speaker: Michael Kirby (Fermi National Accelerator Laboratory)

MicroBooNE_Tuning_NuXTract2023.pdf

14:30 Challenges in fitting to xsec data

Speaker: Jaafar Chakrani (Centre National de la Recherche Scientifique (FR))

20231005_NuXTract_v2.pdf

15:00 Discussion

15:30 → 16:00 Coffee break

16:00 → 18:00 Breakout sessions

Breakout_Sessions_Info.pdf

19:00 → 21:00 Party

Friday 6 October

09:30 → 10:30 Summary and closing discussions

09:30 Data release, data presentation & documentation, and data preservation

Data release, data presentation & documentation, and data preservation

10:30 → 11:00 Coffee break

11:00 → 12:30 Summary and closing discussions

Convener: Dr Stephen Dolan (CERN)

11:00 Unfolding, regularisation and efficiency corrections

Unfolding, regularisation and efficiency corrections

11:20 Fake data studies and validations

11:40 Fitting cross-section measurements

Fitting cross-section measurements

12:30 → 14:00 Lunch