Ellipsoidal nested sampling, expression of the model uncertainty and measurement

7th International Workshop DICE2014 Spacetime – Matter – Quantum Mechanics

物理学;力学

Palmisano, C.^1 ; Mana, G.^2 ; Gervino, G.^1,3

Dipartimento di Fisica, Università di Torino, Via P. Giuria, 1, Torino

10125, Italy^1

INRIM - Istituto Nazionale di Ricerca Metrologica, Str. delle Cacce 91, Torino

10135, Italy^2

Istituto Nazionale di Fisica Nucleare (INFN), Sezione di Torino, Via P. Giuria, 1, Torino

10125, Italy^3

关键词: Appropriate models;    Bayesian model selection;    Computational costs;    Data distribution;    Model uncertainties;    Nuisance parameter;    Numerical strategies;    Two dimensional distribution;   

Others  :  https://iopscience.iop.org/article/10.1088/1742-6596/626/1/012070/pdf
DOI  :  10.1088/1742-6596/626/1/012070

学科分类：力学，机械学

来源: IOP

【 摘 要 】

The measurand value, the conclusions, and the decisions inferred from measurements may depend on the models used to explain and to analyze the results. In this paper, the problems of identifying the most appropriate model and of assessing the model contribution to the uncertainty are formulated and solved in terms of Bayesian model selection and model averaging. As computational cost of this approach increases with the dimensionality of the problem, a numerical strategy, based on multimodal ellipsoidal nested sampling, to integrate over the nuisance parameters and to compute the measurand post-data distribution is outlined. In order to illustrate the numerical strategy, by use of MATHEMATICA an elementary example concerning a bimodal, two-dimensional distribution has also been studied.