4.3 Article

Bayesian and frequentist approaches to resonance searches

期刊

JOURNAL OF INSTRUMENTATION
卷 14, 期 -, 页码 -

出版社

IOP Publishing Ltd
DOI: 10.1088/1748-0221/14/10/P10031

关键词

Analysis and statistical methods; Data analysis

向作者/读者索取更多资源

We investigate Bayesian and frequentist approaches to resonance searches using a toy model based on an ATLAS search for the Higgs boson in the diphoton channel. We draw pseudo-data from the background only model and background plus signal model at multiple luminosities, from 10(-3)/fb to 10(7)/fb. We chart the change in the Bayesian posterior of the background only model and the global p-value. We find that, as anticipated, the posterior converges to certainty about the model as luminosity increases. The p-value, on the other hand, randomly walks between 0 and 1 if the background only model is true, and otherwise converges to 0. After briefly commenting on the frequentist properties of the posterior, we make a direct comparison of the significances obtained in Bayesian and frequentist frameworks. We find that the well-known look-elsewhere effect reduces local significances by about 1 sigma. We furthermore find that significances from our Bayesian framework are typically about 1 - 2 sigma smaller than the global significances, though the reduction depends on the prior, global significance and integrated luminosity. This suggests that even global significances could significantly overstate the evidence against the background only model. We checked that this effect - the Bayes effect - was robust with respect to fourteen choices of prior and investigated the Jeffreys-Lindley paradox for three of them.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.3
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

Article Physics, Particles & Fields

PhaseTracer: tracing cosmological phases and calculating transition properties

Peter Athron, Csaba Balazs, Andrew Fowlie, Yang Zhang

EUROPEAN PHYSICAL JOURNAL C (2020)

Article Astronomy & Astrophysics

Nested sampling cross-checks using order statistics

Andrew Fowlie, Will Handley, Liangliang Su

MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY (2020)

Article Statistics & Probability

Objective Bayesian approach to the Jeffreys-Lindley paradox

Andrew Fowlie

Summary: In an attempt to address the Jeffreys-Lindley paradox, we explore finding priors that are completely indifferent to sample size. We show that while an improper scale-invariant prior is obtained, a truncated scale-invariant prior can delay the dependence on sample size. Additionally, we shed light on the paradox by relating it to the improper nature of the scale-invariant prior.

COMMUNICATIONS IN STATISTICS-THEORY AND METHODS (2022)

Article Astronomy & Astrophysics

Nested sampling with plateaus

Andrew Fowlie, Will Handley, Liangliang Su

Summary: The article introduces a modified version of nested sampling (NS) algorithm that can handle plateaus in the likelihood function and can be retrospectively applied to runs from popular NS software. The algorithm evicts live points in plateaus one by one without replacement, addressing the faulty estimates produced by traditional NS algorithm when dealing with plateaus. The simplicity of the modification suggests that it should become the canonical version of Skilling's NS algorithm.

MONTHLY NOTICES OF THE ROYAL ASTRONOMICAL SOCIETY (2021)

Article Physics, Particles & Fields

A comparison of optimisation algorithms for high-dimensional particle and astrophysics applications

Csaba Balazs, Melissa van Beekveld, Sascha Caron, Barry M. Dillon, Ben Farmer, Andrew Fowlie, Eduardo C. Garrido-Merchan, Will Handley, Luc Hendriks, Gudlaugur Johannesson, Adam Leinweber, Judita Mamuzic, Gregory D. Martinez, Sydney Otten, Roberto Ruiz de Austri, Pat Scott, Zachary Searle, Bob Stienen, Joaquin Vanschoren, Martin White

Summary: The research focused on optimization problems prevalent in particle and astrophysics, evaluating various global optimization algorithms and drawing general conclusions on their relative merits for different functions.

JOURNAL OF HIGH ENERGY PHYSICS (2021)

Article Statistics & Probability

Neyman-Pearson lemma for Bayes factors

Andrew Fowlie

Summary: The Neyman-Pearson lemma applies to Bayes factors when considering expected type-1 and type-2 error rates, with Bayes factor as the test statistic maximizing expected power for fixed expected type-1 error rate. For Bayes factors involving a simple null hypothesis, the expected type-1 error rate is equivalent to the frequentist type-1 error rate. The connection between the Karlin-Rubin theorem, uniformly most powerful tests, and Bayes factors can provide frequentist motivations for computing Bayes factors and potentially help reconcile differences between Bayesians and frequentists.

COMMUNICATIONS IN STATISTICS-THEORY AND METHODS (2021)

Article Physics, Particles & Fields

Thermal WIMPs and the scale of new physics: global fits of Dirac dark matter effective field theories

Peter Athron, Neal Avis Kozar, Csaba Balazs, Ankit Beniwal, Sanjay Bloor, Torsten Bringmann, Joachim Brod, Christopher Chang, Jonathan M. Cornell, Ben Farmer, Andrew Fowlie, Tomas E. Gonzalo, Will Handley, Felix Kahlhoefer, Anders Kvellestad, Farvah Mahmoudi, Markus T. Prim, Are Raklev, Janina J. Renk, Andre Scaffidi, Pat Scott, Patrick Stoecker, Aaron C. Vincent, Martin White, Sebastian Wild, Jure Zupan

Summary: In this study, a wide class of WIMP dark matter models were assessed in light of the latest experimental results using the global fitting framework GAMBIT. The global analysis of effective field theory operators revealed challenges in satisfying all constraints simultaneously while maintaining EFT validity at LHC energies. However, large regions of viable parameter space were found where the EFT is valid and the relic density can be reproduced, suggesting that WIMPs can still account for the DM of the universe while being consistent with the latest data.

EUROPEAN PHYSICAL JOURNAL C (2021)

Article Physics, Multidisciplinary

Nested Sampling for Frequentist Computation: Fast Estimation of Small p-Values

Andrew Fowlie, Sebastian Hoof, Will Handley

Summary: This paper proposes a novel method for computing p-values based on nested sampling applied to the sampling space of the problem. Through experiments, it is shown that this method has lower computational cost and requires fewer simulations compared to traditional Monte Carlo estimates, particularly for significances greater than about 4 sigma in high-energy physics.

PHYSICAL REVIEW LETTERS (2022)

Article Physics, Multidisciplinary

Publishing statistical models: Getting the most out of particle physics experiments

Kyle Cranmer, Sabine Kraml, Harrison B. Prosper, Philip Bechtle, Florian U. Bernlochner, Itay M. Bloch, Enzo Canonero, Marcin Chrzaszcz, Andrea Coccaro, Jan Conrad, Glen Cowan, Matthew Feickert, Nahuel F. Ferreiro, Andrew Fowlie, Lukas Heinrich, Alexander Held, Thomas Kuhr, Anders Kvellestad, Maeve Madigan, Farvah Mahmoudi, Knut D. Mora, Mark S. Neubauer, Maurizio Pierini, Juan Rojo, Sezen Sekmen, Luca Silvestrini, Veronica Sanz, Giordon Stark, Riccardo Torre, Robert Thorne, Wolfgang Waltenberger, Nicholas Wardle, Jonas Wittbrodt

Summary: The researchers present a scientific case for systematically publishing the full statistical models to enhance the impact of experimental results. They discuss the technical developments that make this practical and provide examples to illustrate the effectiveness of detailed statistical modeling in various physics cases.

SCIPOST PHYSICS (2022)

Editorial Material Astronomy & Astrophysics

Comment on Accumulating evidence for the associate production of a neutral scalar with mass around 151 GeV

Andrew Fowlie

Summary: This paper combines multiple experimental results to provide evidence for a new fundamental particle. However, the researchers cast doubt on the strength of the evidence and obtained lower significances through reanalysis.

PHYSICS LETTERS B (2022)

Review Physics, Multidisciplinary

Simple and statistically sound recommendations for analysing physical theories

Shehu S. AbdusSalam, Fruzsina J. Agocs, Benjamin C. Allanach, Peter Athron, Csaba Balazs, Emanuele Bagnaschi, Philip Bechtle, Oliver Buchmueller, Ankit Beniwal, Jihyun Bhom, Sanjay Bloor, Torsten Bringmann, Andy Buckley, Anja Butter, Jose Eliel Camargo-Molina, Marcin Chrzaszcz, Jan Conrad, Jonathan M. Cornell, Matthias Danninger, Jorge de Blas, Albert De Roeck, Klaus Desch, Matthew Dolan, Herbert Dreiner, Otto Eberhardt, John Ellis, Ben Farmer, Marco Fedele, Henning Flaecher, Andrew Fowlie, Tomas E. Gonzalo, Philip Grace, Matthias Hamer, Will Handley, Julia Harz, Sven Heinemeyer, Sebastian Hoof, Selim Hotinli, Paul Jackson, Felix Kahlhoefer, Kamila Kowalska, Michael Kraemer, Anders Kvellestad, Miriam Lucio Martinez, Farvah Mahmoudi, Diego Martinez Santos, Gregory D. Martinez, Satoshi Mishima, Keith Olive, Ayan Paul, Markus Tobias Prim, Werner Porod, Are Raklev, Janina J. Renk, Christopher Rogan, Leszek Roszkowski, Roberto Ruiz de Austri, Kazuki Sakurai, Andre Scaffidi, Pat Scott, Enrico Maria Sessolo, Tim Stefaniak, Patrick Stoecker, Wei Su, Sebastian Trojanowski, Roberto Trotta, Yue-Lin Sming Tsai, Jeriek Van den Abeele, Mauro Valli, Aaron C. Vincent, Georg Weiglein, Martin White, Peter Wienemann, Lei Wu, Yang Zhang

Summary: This article discusses the unique challenges that physical theories dependent on many parameters or tested against data from multiple experiments pose to statistical inference. The authors provide clear guidance and recommendations for statistically sound inference methods, as well as readily-available software tools and standards. Examples provided in the article can be reproduced using publicly available code on Zenodo.

REPORTS ON PROGRESS IN PHYSICS (2022)

Article Multidisciplinary Sciences

Hadronic uncertainties versus new physics for the W boson mass and Muon g-2 anomalies

Peter Athron, Andrew Fowlie, Chih-Ting Lu, Lei Wu, Yongcheng Wu, Bin Zhu

Summary: In this article, the authors demonstrate that the anomalies in the measurements of the W mass and the muon g - 2 are unlikely to have a common origin, but a model involving leptoquarks might explain both anomalies. The electroweak fits show that including the g - 2 measurement worsens the tension with the CDF measurement, and conversely, adjustments that alleviate the CDF tension worsen the g - 2 tension beyond 5 sigma. Therefore, regardless of the size of the hadronic contributions in the Standard Model, new physics explanations are inevitable if we adopt the CDF W mass measurement.

NATURE COMMUNICATIONS (2023)

Article Multidisciplinary Sciences

Nested sampling for physical scientists

Greg Ashton, Noam Bernstein, Johannes Buchner, Xi Chen, Gabor Csanyi, Andrew Fowlie, Farhan Feroz, Matthew Griffiths, Will Handley, Michael Habeck, Edward Higson, Michael Hobson, Anthony Lasenby, David Parkinson, Livia B. Partay, Matthew Pitkin, Doris Schneider, Joshua S. Speagle, Leah South, John Veitch, Philipp Wacker, David J. Wales, David Yallup

Summary: This Primer examines Skilling's nested sampling algorithm and its application in Bayesian inference and multidimensional integration. The principles of nested sampling are summarized and recent developments using efficient nested sampling algorithms in high dimensions are surveyed. Detailed examples from cosmology, gravitational-wave astronomy, and materials science are provided. Finally, the Primer includes recommendations for best practices and a discussion of potential limitations and optimizations of nested sampling.

NATURE REVIEWS METHODS PRIMERS (2022)

Article Astronomy & Astrophysics

Global fits of SUSY at future Higgs factories.

Peter Athron, Csaba Balazs, Andrew Fowlie, Huifang Lv, Wei Su, Lei Wu, Jin Min Yang, Yang Zhang

Summary: In this work, the impact of electroweak and Higgs precision measurements at future electron positron colliders on various supersymmetric models is studied. The results show that future precision measurements can further test the currently allowed parameter space of these models and potentially distinguish different dark matter annihilation mechanisms.

PHYSICAL REVIEW D (2022)

Article Physics, Particles & Fields

Global fits of axion-like particles to XENON1T and astrophysical data

Peter Athron, Csaba Balazs, Ankit Beniwal, J. Eliel Camargo-Molina, Andrew Fowlie, Tomas E. Gonzalo, Sebastian Hoof, Felix Kahlhoefer, David J. E. Marsh, Markus Tobias Prim, Andre Scaffidi, Pat Scott, Wei Su, Martin White, Lei Wu, Yang Zhang

Summary: The excess of electron recoil events observed in the XENON1T experiment has been proposed to potentially indicate axion-like particles (ALPs) originated from the Sun or dark matter halo, or due to trace amounts of tritium in the experiment. Combining XENON1T data with astrophysical probes supports the dark matter ALP hypothesis, despite the need for tuning unknown parameters. Bayesian analysis does not show strong preference for the ALP interpretation of the XENON1T excess over the background hypothesis, despite the tensions in the case of solar ALPs.

JOURNAL OF HIGH ENERGY PHYSICS (2021)

暂无数据