Abstract:Machine learning has emerged as a powerful solution to the modern challenges in accelerator physics. However, the limited availability of beam time, the computational cost of simulations, and the high dimensionality of optimization problems pose significant challenges in generating the required data for training state-of-the-art machine learning models. In this work, we introduce c heetah, a p y t orch-based high-speed differentiable linear beam dynamics code. c heetah enables the fast collection of large datasets by reducing computation times by multiple orders of magnitude and facilitates efficient gradient-based optimization for accelerator tuning and system identification. This positions c heetah as a user-friendly, readily extensible tool that integrates seamlessly with widely adopted machine learning tools. We showcase the utility of c heetah through five examples, including reinforcement learning training, gradient-based beamline tuning, gradient-based system identification, physics-informed Bayesian optimization priors, and modular neural network surrogate modeling of space charge effects. The use of such a high-speed differentiable simulation code will simplify the development of machine learning-based methods for particle accelerators and fast-track their integration into everyday operations of accelerator facilities. https://doi.org/10.1103/PhysRevAccelBeams.27.054601 Published by the American Physical Society under the terms of the Creative Commons Attribution 4.0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article's title, journal citation, and DOI. Published by the American Physical Society

Parallelisation of PyHEADTAIL, a Collective Beam Dynamics Code for Particle Accelerator Physics

Parallelized Implementation of the Finite Particle Method for Explicit Dynamics in GPU

The Implementation of the Three-Dimensional Unified Gas-Kinetic Wave-Particle Method on Multiple Graphics Processing Units

Heterogeneous Programming and Optimization of Gyrokinetic Toroidal Code and Large-Scale Performance Test on TH-1A.

Accelerating Dissipative Particle Dynamics Simulations on GPUs: Algorithms, Numerics and Applications

GPU Parallel Acceleration of Beam Tracing for Propagation Prediction in Urban Environments

Study of A Gpu-Based Parallel Computing Method for the Monte Carlo Program

An Algorithm to Parallelise Parton Showers on a GPU

Bridging the gap between machine learning and particle accelerator physics with high-speed, differentiable simulations

On the parallelization of multipacting simulation codes for the design of particle accelerator components

GPU acceleration and performance of the particle-beam-dynamics code Elegant

Accelerating Astrophysical Particle Simulations with Programmable Hardware (FPGA and GPU)

Accelerating Pythonic coupled cluster implementations: a comparison between CPUs and GPUs

Parallelization of Kinetic Theory Simulations

Synthesizing Particle-in-Cell Simulations Through Learning and GPU Computing for Hybrid Particle Accelerator Beamlines

A GPGPU based program to solve the TDSE in intense laser fields through the finite difference approach

Parallelization and Optimization of Laser-Plasma-Interaction Simulation

Efficient molecular dynamics simulations with many-body potentials on graphics processing units

Parallelization and Optimization of Application for Phonon BTE

EVALUATION OF SPEEDUP OF MONTE CARLO CALCULATIONS OF TWO SIMPLE REACTOR PHYSICS PROBLEMS CODED FOR THE GPU/CUDA ENVIRONMENT

Cheetah: Bridging the Gap Between Machine Learning and Particle Accelerator Physics with High-Speed, Differentiable Simulations