Probing intractable beyond-standard-model parameter spaces armed with Machine Learning

TL;DR

This paper reviews how machine learning and advanced sampling techniques are used to efficiently explore high-dimensional parameter spaces in beyond-the-standard-model physics, overcoming computational challenges.

Contribution

It summarizes recent efforts and highlights key machine learning and sampling methods that improve the exploration of complex BSM parameter spaces.

Findings

ML algorithms enhance sampling efficiency in high-dimensional spaces

Advanced sampling methods mitigate the curse of dimensionality

Significant progress in constraining BSM models with ML techniques

Abstract

This article attempts to summarize the effort by the particle physics community in addressing the tedious work of determining the parameter spaces of beyond-the-standard-model (BSM) scenarios, allowed by data. These spaces, typically associated with a large number of dimensions, especially in the presence of nuisance parameters, suffer from the curse of dimensionality and thus render naive sampling of any kind -- even the computationally inexpensive ones -- ineffective. Over the years, various new sampling (from variations of Markov Chain Monte Carlo (MCMC) to dynamic nested sampling) and machine learning (ML) algorithms have been adopted by the community to alleviate this issue. If not all, we discuss potentially the most important among them and the significance of their results, in detail.