Advanced statistical methods to fit nuclear models

2018/11/22 by Matthew Shelley, Shelley, M., P. Becker +5
Computer Science · Physics and Astronomy · Mathematics · #Gaussian Processes and Bayesian Inference #Scientific Research and Discoveries #Markov Chains and Monte Carlo Methods

paper · pdf · doi:10.48550/arxiv.1811.09130

Abstract

We discuss advanced statistical methods to improve parameter estimation of nuclear models. In particular, using the Liquid Drop Model for nuclear binding energies, we show that the area around the global χ² minimum can be efficiently identified using Gaussian Process Emulation. We also demonstrate how Markov-chain Monte-Carlo sampling is a valuable tool for visualising and analysing the associated multidimensional likelihood surface.

Related

• An introduction to computational complexity and statistical learning theory applied to nuclear models
• The DNA of nuclear models: How AI predicts nuclear masses
• The statistical model and nuclear level densities
• Direct microscopic calculations of nuclear level densities in the shell model Monte Carlo approach
• Trees and Islands -- Machine learning approach to nuclear physics