Bulletin of the American Physical Society
2019 Fall Meeting of the APS Division of Nuclear Physics
Volume 64, Number 12
Monday–Thursday, October 14–17, 2019; Crystal City, Virginia
Session 1WB: New Data Analysis Methods I |
Hide Abstracts |
Chair: Michael Doring, GWU Room: Salon 2 |
Monday, October 14, 2019 9:00AM - 9:30AM |
1WB.00001: Introduction to Bayesian methods and uncertainty quantification Invited Speaker: Sarah Wesolowski The nuclear physics community has largely embraced Bayesian statistical methods in the past decade, due to an increasing need for uncertainty quantification (UQ) on predictions and measurements and the availability of computational power. The Bayesian interpretation of probability allows for an expanded view of what can be treated as a random variable, and also facilitates the inclusion of physical knowledge quantitatively in the form of prior probabilities. Priors make the assumptions in a statistical analysis explicit, which enhances both reproducibility and clarity. I will discuss how Bayesian statistics can be used in nuclear physics for parameter estimation (fitting), quantitative model comparison, and for UQ. I will also discuss the advantages of being able to perform validation of statistical analyses, e.g., uncertainties should neither be over- or under-estimated and how to test that statistically. [Preview Abstract] |
Monday, October 14, 2019 9:30AM - 10:00AM |
1WB.00002: Application of Bayesian methods in effective field theory Invited Speaker: Richard Furnstahl The use of effective field theory (EFT) methods to describe nuclear systems holds the promise of model independence and order-by-order convergence for the calculation of observables. We will describe why Bayesian statistics is an ideal framework to quantify uncertainties from estimating EFT low-energy constants, guided by theoretical expectations about EFT truncation that are explicitly specified through prior probability densities. These uncertainties can then be combined with other sources of error and propagated to observables. Bayesian model selection identifies how many EFT orders can be extracted from given data and offers the possibility of distinguishing between alternative EFT formulations. Bayesian model-checking diagnostics are powerful tools for EFT validation. These statistical methods provide more than just theoretical error bars but can serve as catalysts for physics discovery. [Preview Abstract] |
Monday, October 14, 2019 10:00AM - 10:30AM |
1WB.00003: Uncertainty quantification in Nuclear Reactions Invited Speaker: F. M. Nunes Nuclear reactions are one of the most versatile probes to study nuclei from which an array of properties can be extracted, as well as astrophysical information. Reaction models needed in the interpretation of the data often rely on nucleon-nucleus effective interactions that depend on a number of parameters, typically constrained by elastic scattering data. It is important to quantify the uncertainties associated with the observables produced by these models. In this presentation, we discuss a systematic comparative study made between the standard frequentist approach and the Bayesian approach when applied to nuclear reactions [1]. We find that the Bayesian approach produces larger confidence intervals than the standard chi2 minimization approach and that the standard method underestimates the error, particularly for the higher confidence levels. We also explore different avenues with the aim of reducing the uncertainties.\\ \\$[1]$ G. King, A. Lovell, L. Neufcourt, F.M. Nunes, Phys. Rev. Lett 122, 232502 (2019). [Preview Abstract] |
Follow Us |
Engage
Become an APS Member |
My APS
Renew Membership |
Information for |
About APSThe American Physical Society (APS) is a non-profit membership organization working to advance the knowledge of physics. |
© 2024 American Physical Society
| All rights reserved | Terms of Use
| Contact Us
Headquarters
1 Physics Ellipse, College Park, MD 20740-3844
(301) 209-3200
Editorial Office
100 Motor Pkwy, Suite 110, Hauppauge, NY 11788
(631) 591-4000
Office of Public Affairs
529 14th St NW, Suite 1050, Washington, D.C. 20045-2001
(202) 662-8700