Fast and Precise Learned Charged-Particle Trajectory Regression at the Large Hadron Collider

2026-09-26

AI Summary

This study addresses the prohibitive computational cost of traditional algorithms for charged particle track fitting at the Large Hadron Collider (LHC), which hinders large-scale data processing. We formulate track parameter regression as a sequential task and propose an end-to-end learning approach based on a Bidirectional Gated Linear Recurrent encoder (BiGLR) to replace conventional Kalman filtering and least-squares fitting. This work is the first to achieve end-to-end track fitting with full classical precision, further accelerated by custom GPU kernels during inference. Experiments demonstrate that the proposed method reproduces the $10^{-5}$-level accuracy of traditional algorithms while substantially reducing computational costs under equivalent hardware conditions. Consequently, this approach establishes an efficient new paradigm for pattern recognition in LHC experiments.

Abstract

We propose a training recipe that treats charged-particle trajectory parameter regression on high-energy physics detector data as a sequence-modeling task. Kalman filters and linearized least-squares fits have been the classical standard approach for this task: they are optimal estimators for sparsely sampled linear-Gaussian data and are commonly used for trajectory parameter regression (fitting). The classical fitting techniques implemented for this domain reach a final precision of one part in $10^5$ through detailed modeling of detector geometry, material, detection effects and precise numerical integration of the equations of motion through the detector's inhomogeneous magnetic field. With this study, we demonstrate that using a bidirectional gated linear recurrent encoder, one is able to reproduce the full precision of classical track fitting techniques. Using a custom kernel, we also achieve significantly higher throughput during GPU inference, compared to classical fitting software running on similarly priced multi-core CPU servers representing typically employed hardware. Such a speedup would lead to considerable cost savings for the pattern recognition at the Large Hadron Collider. To our knowledge, this is the first end-to-end learned track fit to reach the full precision and, at the same time, offer the opportunity to reduce the computing costs.

Problem

• charged-particle trajectory regression
• track fitting
• Large Hadron Collider
• high-energy physics
• computing cost reduction

Innovation

• sequence modeling
• trajectory regression
• gated linear recurrent encoder
• custom GPU kernel
• end-to-end learning