HyperTrack: Neural Combinatorics for High Energy Physics

TL;DR

HyperTrack introduces a novel deep learning-based clustering algorithm employing graph neural networks and set transformers to address complex combinatorial inverse problems in high energy physics, demonstrating effectiveness in particle tracking simulations.

Contribution

The paper presents a new AI-driven clustering method combining graph neural networks and set transformers for high energy physics inverse problems, with innovative loss functions and training strategies.

Findings

Effective in particle tracking simulations

Applicable to calorimetry and pile-up discrimination

Outperforms traditional methods in accuracy

Abstract

Combinatorial inverse problems in high energy physics span enormous algorithmic challenges. This work presents a new deep learning driven clustering algorithm that utilizes a space-time non-local trainable graph constructor, a graph neural network, and a set transformer. The model is trained with loss functions at the graph node, edge and object level, including contrastive learning and meta-supervision. The algorithm can be applied to problems such as charged particle tracking, calorimetry, pile-up discrimination, jet physics, and beyond. We showcase the effectiveness of this cutting-edge AI approach through particle tracking simulations. The code is available online.