PhyStat-v Workshop on Statistical Issues in Experimental Neutrino Physics

Asia/Tokyo
Yoshi Uchida (Imperial College London)
Description

Overview

Neutrino experiments are reaching exciting new levels of precision: percent-level systematic uncertainties in oscillation experiments, the eV scale in direct mass measurements, 1025 year half-lives in double beta decay experiments, and neutrino-nucleus cross section measurements at the 5% level. With this new precision comes an important focus on the statistical understanding of data from neutrino experiments, and techniques to ensure the quality of the results produced.
 
The PhyStat series of workshops has been a successful institution in the world of collider physics over the past 15 years, providing a forum for discussing topics such as setting upper limits on or discovering rare processes, and combining results from different experiments. We now aim to turn this success to the field of neutrino physics.

PhyStat-ν will address statistical issues in the broad range of modern neutrino physics, including making measurements of model parameters, setting limits, discovery criteria, discrete choice of models, Bayesian vs. frequentist inference, and combining experiments. Two workshops are planned for 2016: this workshop at the Kavli Institute for the Physics and Mathematics of the Universe in Kashiwa, Japan in May, and a second workshop at Fermilab in the autumn. These workshops will consist of invited and contributed talks, and poster and discussion sessions, and will bring together experts in the analysis of neutrino data with experts in statistics to explore the statistical issues in neutrino physics. 
 

We are looking forward to seeing you in Kashiwa.

Local Organizing Committee Members:
Chair: Mark HARTZ (Kavli IPMU/TRIUMF) 
Christophe BRONNER (Kavli IPMU)
Richard CALLAND (Kavli IPMU)
Yoshinari HAYATO (ICRR)
Yasuhiro NISHIMURA (ICRR)
Kimihiro OKUMURA (ICRR)

Scientific Organizing Committee Members:
Chair: Yoshi UCHIDA (Imperial College London)
Jun CAO (IHEP, CAS)
Daniel CHERDACK (Colorado State University)
Robert COUSINS (UCLA)
David VAN DYK (Imperial College London)
Joe FORMAGGIO (MIT)
Mark HARTZ (Kavli IPMU (WPI), The University of Tokyo)
Pilar HERNANDEZ (University of Valencia)
Thomas JUNK (Fermilab)
Asher KABOTH (Royal Holloway, University of London)
Louis LYONS (Imperial College London)
Shun SAITO (Kavli IPMU (WPI), The University of Tokyo)
Subir SARKAR (University of Oxford, Neils Bohr Institute)
Elizabeth WORCESTER (BNL)
Kai ZUBER (TU Dresden)
 

 
Poster
Programme
Slides
Participants
  • AMIT KUMAR SRIVASTAVA
  • Andrew Cudd
  • Arturo Fiorentini
  • Asher Kaboth
  • Benda Xu
  • Callum Wilkinson
  • ChangDong Shin
  • Christine Nielsen
  • Christophe Bronner
  • Christopher Barry
  • Ciro Riccio
  • Clarence Wret
  • Cristovao Vilela
  • Daniel Cherdack
  • David van Dyk
  • Davide Sgalaberna
  • Dean Karlen
  • Elder Pinzon
  • Elizabeth Worcester
  • Emilio Ciuffoli
  • Eunhyang Kwon
  • Ewen Gillies
  • Francesco Gizzarelli
  • Francisco Villaescusa-Navarro
  • Gabriel Collin
  • Gabriel Santucci
  • Georgios Christodoulou
  • Hiroyuki Sekiya
  • Hyunkwan Seo
  • InSung Yeo
  • Itaru Shimizu
  • Jacob Morrison
  • Joao Pedro Athayde Marcondes de Andre
  • Johannes Bergström
  • Jose Palomino
  • Josh Amey
  • Katsuki Hiraide
  • Keigo Nakamura
  • Kimihiro Okumura
  • Kyungmin Seo
  • Leïla Haegel
  • Livia Ludhova
  • Louis Lyons
  • Lu Lu
  • Luke Pickering
  • Marat Khabibullin
  • Marco Kleesiek
  • Mark Hartz
  • Mark McCarthy
  • Mathieu Lamoureux
  • Megan Friend
  • Michael Betancourt
  • Michael Smy
  • Michiru Kaneda
  • Mikio Morii
  • Mitchell Yu
  • MOHAMMAD SAJJAD ATHAR
  • Monojit Ghosh
  • Morgan Wascko
  • Patrick Dunne
  • Patrick Stowell
  • Paul Martins
  • Phillip Litchfield
  • Pierre Lasorak
  • Raj Shah
  • Ralitsa Sharankova
  • Richard Calland
  • Robert Cousins
  • Ronald Bruijn
  • Sara Algeri
  • Seokhee Park
  • Shigetaka Moriyama
  • Shiro Ikeda
  • Son Cao
  • Stephen Dolan
  • Sunny Seo
  • Teppei Katori
  • Tianmeng Lou
  • Tomislav Vladisavljevic
  • Tomoyo Yoshida
  • Wing Yan Ma
  • Wonqook Choi
  • Yasuhiro NISHIMURA
  • Yoomin Oh
  • Yoshi Uchida
  • Yoshinari Hayato
  • Youngju Ko
  • YUE WANG
  • Yufeng Li
  • Zoya Vallari
    • 08:30 09:00
      Registration
    • 09:00 10:30
      Introductory Lecture
      • 09:00
        Introduction to Statistics for PhyStat-𝜈 1h 20m
        Speaker: Louis Lyons (Imperial College London)
        Slides
    • 10:30 11:00
      Registration and Tea and Coffee 30m
    • 11:00 12:10
      Opening Session

      Introductory Overview, with talks on general neutrino physics and the topics to be discussed at the workshop

      • 11:00
        Welcome to PhyStat-nu and Logistics 15m
        Speaker: Prof. Mark Hartz (Kavli IPMU (WPI), The University of Tokyo/TRIUMF)
        Slides
      • 11:15
        Introduction to the Workshop 10m
        Introduction of the format of the workshop and the invited experts and scientific organisers.
        Speaker: Dr Yoshi Uchida (Imperial College London)
        Slides
      • 11:25
        The Current Status of Neutrino Physics 45m
        Speaker: Sergeuy Petcov
        Slides
    • 12:10 13:30
      Lunch 1h 20m
    • 13:30 16:40
      Statistical Methods at Current Experiments

      Overview talks on how current and recent experiments are treating their data

      • 13:30
        Long-Baseline Accelerator Oscillation Experiments 40m
        Speaker: Dr Asher Kaboth (RHUL)
        Slides
      • 14:10
        Short- and Medium-Baseline Reactor Oscillation Experiments 40m
        Speaker: Sunny Seo (Seoul National University)
        Slides
      • 14:50
        Tea and Coffee 30m
      • 15:20
        Direct Neutrino Mass Measurements Using Weak Decays 40m
        Speaker: Dr Marco Kleesiek (KIT)
        Slides
      • 16:00
        Solar Neutrinos 40m
        Speaker: Dr Michael Smy (University of California, Irvine)
        Slides
    • 16:40 17:40
      Short Poster Talks

      Short introductory talks from poster presenters

      • 16:40
        Estimation of neutrino oscillation parameters with Markov Chain Monte Carlo 6m
        The T2K experiment is a long-baseline accelerator neutrino experiment with the ability to measure Δm32, θ23, θ13 and δCP from (anti-)νμ disappearance as well as (anti-)νe appearance. In this poster we show a Bayesian framework using the Metropolis-Hastings algorithm to estimate the oscillation parameters. The advantage of Markov Chain Monte Carlo is its ability to scan a multidimensional parameter space, allowing to evaluate at the same time the oscillation parameters and the systematic uncertainties on the neutrino interaction model, accelerator flux model and detector reconstruction. We present the results obtained for a simultaneous fit of νμ disappearance and νe appearance, and a probability of anti-νe appearance.
        Speaker: Ms Leïla HAEGEL (Université de Genève)
        Slides
      • 16:46
        Statistical techniques used for neutrino oscillation analyses with the VALOR framework 6m
        VALOR is a well established neutrino fitting group that is leading several neutrino oscillation analyses in T2K, producing world leading results. The analysis framework has been validated and optimised for both a hybrid Bayesian-Frequentist approach to fitting parameters, where the systematic parameters are marginalised, as well as a frequentist analysis, that consists of the likelihood ratio maximisation as a function of all the fit parameters. The VALOR group is involved in many other neutrino oscillation experiments such as DUNE and Hyper-K, which goal is the measurement of δCP , and the Fermilab Short Baseline Neutrino program, a novel experiment aiming to search for sterile neutrinos. In this talk the statistical techniques used in the neutrino oscillation analyses performed by the VALOR group are shown.
        Speaker: Davide Sgalaberna (University of Geneva)
        Slides
      • 16:52
        Anti-𝜈e Appearance at T2K using VALOR 6m
        VALOR is a well established neutrino fitting group that is leading several neutrino oscillation analyses in T2K, producing world leading results. The analysis framework has been validated and optimised for both a hybrid Bayesian-Frequentist approach to fitting parameters, where the systematic parameters are marginalised, as well as a frequentist analysis, that consists of the likelihood ratio maximisation as a function of all the fit parameters. The VALOR group is involved in many other neutrino oscillation experiments such as DUNE and Hyper-K, which goal is the measurement of δCP , and the Fermilab Short Baseline Neutrino program, a novel experiment aiming to search for sterile neutrinos. In this talk the statistical techniques used in the neutrino oscillation analyses performed by the VALOR group are shown.
        Speaker: Mr raj shah (STFC, UK)
        Slides
      • 16:58
        Tuning generators to the global neutrino cross-section dataset. 6m
        The T2K Neutrino Oscillation Experiment uses a prior estimation of a scattering cross-section model when generating nominal Monte-Carlo (MC) neutrino interaction events. A selection of the available models in the NEUT MC generator are fit to a range of historical neutrino interaction cross-section measurements made external to T2K. A nominal model is then chosen which is capable of finding the best agreement with the fitted data. One difficulty that must be overcome when performing these fits is the need to include datasets where only limited information on the observed event kinematics and associated systematic uncertainties are provided. In this poster we review the current efforts to produce tighter constraints on the neutrino interaction cross-section model. We also highlight recent attempts to use multiple kinematic distributions from a single experiment when performing likelihood fits, whilst using the MC model to approximate the true correlations between the different kinematic variables.
        Speaker: Mr John Stowell (University of Sheffield)
        Slides
      • 17:04
        Empirical Parametrisations of Neutrino Oscillations 6m
        The discovery of neutrino oscillations driven by the U[e3] element of the PMNS mixing matrix has prompted a step change in the way neutrino oscillation experiments are analysed, commonly described by the shorthand of "switching from a 2-neutrino to 3-neutrino picture". The difference is less clear cut than this. On the one hand, some older measurements used parametrisations that implicitly assumed at least three neutrino flavours (and less commonly required 2 independent mass splittings). The more interesting change relates to the way oscillations are parametrised for fitting and inference. "Two flavour" analyses used parameters that -- regardless of how they were named -- directly described the shape of the observed data (the location and depth of a 'dip'), whereas newer analyses are explicitly concerned with mixing angles that are not direct observables and only related to the data through mathematical transformations. This has non-trivial consequences for inferences from the data. This poster highlights some of these unintended consequences and considers whether alternative formulations for analysing and presenting data are possible.
        Speakers: Dr Phillip Litchfield (Imperial College London), Dr Phillip Litchfield (Imperial College London)
        Slides
      • 17:10
        Neutrino Oscillation on Epicycles 6m
        Global fit of neutrino oscillations is shown to be equivalent to Fourier transformation in a complex plane. Adding new sterile neutrinos or new oscillation modes is equivalent to adding Fourier components. Therefore new sterile neutrinos should be introduced with physics motivations beyond phenomenology.
        Speaker: Benda Xu (RCNS, Tohoku Univ., Japan)
        Slides
      • 17:16
        Study of the Neutrino Oscillation at Very Short Baseline 6m
        Verification of neutrino oscillations at a very short baseline of a nuclear reactor is one way to check the possible existence of the sterile neutrinos. Challenging features of this short baseline experiment come from: the high muon rates at shallow depth of overburden, the escaping gamma energy due to the small size of the detector in a limited space, and the inaccurate predictions of the reactor anti-neutrino flux. The oscillation parameters of the sterile neutrino have been studied using the energy spectrum from the NEOS (Neutrino Experiment for Oscillation at Short baseline) data collected for 6 months at 24 m baseline of a 3 GW_th commercial reactor.
        Speaker: Mr Youngju Ko (Chung-Ang University)
        Slides
      • 17:22
        Optimization of detector design of NuPRISM, a new water-Cherenkov neutrino near-detector 6m
        NuPRISM is a proposed novel water Cherenkov detector in the J-PARC neutrino beam line that will make neutrino cross section measurements critical to both T2K and T2HK as well as searches for short baseline neutrino oscillations. By observing neutrino interactions over off-axis angles ranging from 1 to 4 degrees with NuPRISM detector, systematic uncertainties for T2K and T2HK oscillation measurements can be significantly reduced. Evaluation of the sensitivity to neutrino interaction and oscillation physics with the NuPRISM detector is ongoing. In this poster, the event reconstruction performance for different configurations of photosensor sizes and photocathode coverages are shown.
        Speaker: Tomoyo Yoshida (Tokyo Institute of Technology)
        Slides
      • 17:28
        Boosted Decision Tree Approach to Track Finding in the COMET Experiment 6m
        The Coherent Muon to Electron Transition (COMET) experiment is designed to search for muon to electron conversion, a process which has very good sensitivity to Beyond the Standard Model physics. The first phase of the experiment is currently under construction at J-PARC. This phase is designed to probe muon to electron conversion 100 times better than the current limit. The experiment will achieve this sensitivity by directing a high intensity muon beam at a stopping target. The detectors probe the resulting events for the signal 105 MeV electron from muon to electron conversion. A boosted decision tree (BDT) algorithm has been developed to find this signal track. This BDT is used to combine energy deposition and timing information with a reweighted inverse hough transform to filter out background hits. The resulting hits are fit using a RANdom SAmple Consensus (RANSAC) fit, which chooses the best fit parameters for an optimized selection of the filtered hits. These hits are then passed to the track fitting algorithm.
        Speaker: Mr Ewen Gillies (Imperial College London)
        Slides
    • 17:40 18:00
      Pre-Reception Talk
      • 17:40
        Welcome from the Director 20m
        Speaker: Hitoshi Murayama (Kavli IPMU)
    • 18:00 20:00
      Poster Session and Drinks Reception 2h
    • 09:00 11:30
      Statistical Methods at Current Experiments

      Overview talks on how current and recent experiments are treating their data

      • 09:00
        Neutrino Nucleus Cross Section Experiments 40m
        Speaker: Dr Morgan Wascko (Imperial College London)
        Slides
      • 09:40
        Atmospheric Neutrinos 40m
        Speaker: Joao Pedro de Andre (Michigan State University)
        Slides
      • 10:20
        Tea and Coffee 30m
      • 10:50
        Double Beta Decay 40m
        Speaker: Dr Itaru Shimizu (Tohoku University)
        Slides
    • 11:30 12:40
      Contributed and invited talks and discussions on statistical methods and issues
      • 11:30
        Combination of Current and Historical Cross Section Data 35m
        Speaker: Dr Callum Wilkinson (Bern)
        Slides
      • 12:05
        Statistical Issues in Neutrino Event Reconstruction 35m
        Speaker: Dr Richard Calland (IPMU)
        Slides
    • 12:40 14:00
      Lunch 1h 20m
    • 14:00 18:00
      Contributed and invited talks and discussions on statistical methods and issues
      • 14:00
        Searching for PeV neutrinos with IceCube 35m
        The IceCube Neutrino Observatory at the geographic South Pole has been taking high-quality data in the deep-ice since 2010. Extremely high energy neutrinos that carry energies above 1 PeV have been observed. The most constraining limits on the flux of ultra-high energy neutrinos, which could come from cosmogenic origins have been obtained based on a binned Poisson-likelihood-ratio method. In this talk, we will discuss the implementation of this statistical tool in light of an extremely rare event scenario.
        Speaker: Dr Lu Lu (Chiba University)
        Slides
      • 14:35
        Event Classification 35m
        Speaker: Shiro Ikeda (Institute of Statistical Mathematics)
        Slides
      • 15:10
        Computational Aspects of Statistical Methods 35m
        Speaker: Michael Betancourt (University of Warwick)
        Slides
      • 15:45
        Tea and Coffee 30m
      • 16:15
        Cosmological Constraints on Neutrino Properties 35m
        Speaker: Francisco Villaescusa-Navarro (INAF-Trieste)
        Slides
      • 16:50
        Three-Neutrino Mixing Fits 35m
        Speaker: Bergstrom Johannes (Universitat de Barcelona)
        Slides
      • 17:25
        Frequentist & Bayesian Bounds and Data Presentation 35m
        Speaker: Steve Biller (University of Oxford)
        Slides
    • 18:00 18:30
      Travel to Dinner Venue 30m
    • 18:30 21:00
      Workshop Dinner 2h 30m
    • 09:00 13:15
      Contributed and invited talks and discussions on statistical methods and issues
      • 09:00
        Status of Global Sterile Neutrino Fits 35m
        Speaker: Gabriel Collin (Massachusetts Institute of Technology)
        Slides
      • 09:35
        Neutrino Mass Hierarchy at JUNO and Related Statistical Issues 35m
        Speaker: Yufeng Li (Institute of High Energy Physics)
        Slides
      • 10:10
        Tea and Coffee 30m
      • 10:40
        Sensitivity to the Neutrino Mass Hierarchy 35m
        Precisely one neutrino mass hierarchy is realized in nature. How can this crucial fact be incorporated into a statistical analysis of the mass hierarchy? It implies that the problem is not hypothesis testing, but rather model selection. The corresponding tools are standard, but different. We will discuss some statistical issues that will be relevant in this kind of experiment, present a simple formula for the corresponding sensitivity and confront it with simulations of the experiment JUNO.
        Speaker: Emilio Ciuffoli (IMP, CAS)
        Slides
      • 11:15
        The KM3NeT/ORCA detector 35m
        The ORCA detector is the low-energy branch of the new generation neutrino telescope KM3NeT which is currently being constructed in the Mediterranean Sea. Interactions of neutrinos that are created in cosmic ray interactions travel and through the Earth can be observed in the detector. By measuring the matter-induced oscillation effects on this flux, KM3NeT/ORCA aims at settling the question on the neutrino mass ordering and to constrain other oscillation parameters. In this contribution an introduction to the ORCA detector will be given, together with the Monte-Carlo simulation efforts and statistical methods used to estimate the sensitivity to determine whether the neutrino mass hierarchy is normal or inverted.
        Speaker: Dr Ronald Bruijn (University of Amsterdam/Nikhef)
        Slides
      • 11:50
        Generative Modelling as a Basis for Statistics 40m
        Speaker: Michael Betancourt (University of Warwick)
        Slides
      • 12:30
        Multiple Hypothesis Testing and Testing One Hypothesis Multiple Times: a unified (re)view 45m
        In statistics, the problem of testing one hypothesis multiple times can be formulated in terms of hypothesis testing when a nuisance parameter is present only under the alternative, also known as “look-elsewhere effect”. Each possible value of the nuisance parameter specifies a different alternative hypothesis and a unique global p-value is provided to summarize the statistical evidence in support (or against) the null hypothesis. From a physics perspective, this scenario occurs quite often in the searches for new signals over an energy or mass spectrum, and in both nested and non-nested frameworks. An alternative way to search for new emissions is to refer to the classical and widely known multiple hypothesis testing approach. Separate tests of hypothesis are conducted at different locations producing an ensemble of local p-values, the smallest is reported as evidence for the new resonance, once adequately adjusted to control the false detection rate (type I error rate). The aim of this talk is to review both approaches, and propose a graphical tool to identify recurrent scenarios where a simple multiple hypothesis testing procedure can be used to provide valid inference with respect to stringent significance requirements, without encountering the usual problem of over-conservativeness.
        Speaker: Ms Sara Algeri (Imperial College London)
        Slides
    • 13:15 14:15
      Lunch 1h
    • 14:15 16:00
      Panel Discussion

      Discussion on matters arising during the workshop

      • 14:15
        Panel Discussion 1h 15m
        Speakers: Dean Karlen (University of Victoria), Dr Elizabeth Worcester (BNL), Michael Betancourt (University of Warwick), Ms Sara Algeri (Imperial College London), Dr Yoshi Uchida (Imperial College London)
      • 15:30
        Tea and Coffee 30m
    • 16:00 18:30
      Closing Session

      Experimental Summary talkby Bob Cousins, Theoretical Summary

      • 16:00
        Statistician's Summary 45m
        Speaker: David Van Dyk (Imperial College London)
        Slides
      • 16:45
        Physicist's Summary 45m
        Speaker: Robert Cousins (UCLA)
        Slides
        Note: new file uploaded 6 June 2016 with minor corrections, additional slides based on the talk as presented, and further references.
      • 17:30
        Closing Statements 1h
Your browser is out of date!

Update your browser to view this website correctly. Update my browser now

×