Efficient Monte Carlo Integration Using Boosted Decision Trees and Generative Deep Neural Networks

TL;DR

This paper introduces machine learning algorithms using boosted decision trees and deep neural networks to significantly improve the efficiency and precision of Monte Carlo integration for complex, non-factorizable integrands in collider physics.

Contribution

The paper presents novel ML-based algorithms that outperform existing methods in Monte Carlo integration accuracy and efficiency for complex physics simulations.

Findings

Algorithms show substantial improvements in integration precision.

Enhanced efficiency in generating complex collider physics processes.

Potential for integration into standard Monte Carlo generators.

Abstract

New machine learning based algorithms have been developed and tested for Monte Carlo integration based on generative Boosted Decision Trees and Deep Neural Networks. Both of these algorithms exhibit substantial improvements compared to existing algorithms for non-factorizable integrands in terms of the achievable integration precision for a given number of target function evaluations. Large scale Monte Carlo generation of complex collider physics processes with improved efficiency can be achieved by implementing these algorithms into commonly used matrix element Monte Carlo generators once their robustness is demonstrated and performance validated for the relevant classes of matrix elements.