Flow Annealed Importance Sampling Bootstrap meets Differentiable Particle Physics

2024-11-25

arXiv.org

AI Summary

High-energy physics simulations require efficient sampling from complex matrix-element distributions, yet existing surrogate models—such as normalizing flows—rely on costly pre-generated training datasets. To address this, we propose FAB-Physics, the first framework to integrate Flow Annealed Importance Sampling Bootstrap (FAB) into particle physics generation tasks. It synergistically combines differentiable particle simulators with normalizing flow–based density modeling, enabling adaptive importance sampling driven by differentiable evaluation of the target density—eliminating the need for pretraining data. By incorporating annealed importance sampling, bootstrap resampling, and physics-informed constraints, FAB-Physics significantly improves sampling efficiency in high-dimensional spaces. Experiments on representative matrix-element distributions demonstrate a >30% reduction in target-density evaluations and a 2.1× increase in effective sample size compared to state-of-the-art baselines.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Calibration & Uncertainty Quantification
• Reasoning under Uncertainty: Probabilistic Programming

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Web Mining and Content Analysis: Web data generation and simulation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

High-energy physics requires the generation of large numbers of simulated data samples from complex but analytically tractable distributions called matrix elements. Surrogate models, such as normalizing flows, are gaining popularity for this task due to their computational efficiency. We adopt an approach based on Flow Annealed importance sampling Bootstrap (FAB) that evaluates the differentiable target density during training and helps avoid the costly generation of training data in advance. We show that FAB reaches higher sampling efficiency with fewer target evaluations in high dimensions in comparison to other methods.

Problem

• Generating simulated data from complex high-energy physics distributions
• Improving sampling efficiency with fewer target evaluations
• Avoiding costly pre-generation of training data

Innovation

• Flow Annealed Importance Sampling Bootstrap
• Differentiable target density evaluation
• Higher sampling efficiency