arXiv subjects

Chad E. Mitchell

Publications and source records attributed to Chad E. Mitchell.

8 recordsLinked to original sources

arXiv2026

The Memory Scaling of Reverse-Mode Differentiation in Particle Accelerator Simulations with Space Charge

The recent development of differentiable simulation codes for particle accelerators has enabled gradient-based workflows that promise finer control and more realistic modeling of accelerator facilities. However, when using reverse-mode automatic differentiation, the memory usage continuously increases during the simulation, and can potentially exceed the available hardware memory -- especially when costly space charge computation is included. To study the memory requirements for differentiable simulations, we have implemented space charge in Cheetah, a PyTorch-based beam tracking code that supports reverse-mode differentiation. We find that the memory usage for reverse-mode differentiation grows linearly with the number of macroparticles and cells, and that it is proportional to the number of space charge kicks involved in the simulation. This general scaling can be used to evaluate whether a given differentiable simulation is feasible given hardware memory constraints.

physics.acc-ph↗

arXiv2024

Laser-Plasma Ion Beam Booster Based on Hollow-Channel Magnetic Vortex Acceleration

Laser-driven ion acceleration provides ultra-short, high-charge, low-emittance beams, which are desirable for a wide range of high-impact applications. Yet after decades of research, a significant increase in maximum ion energy is still needed. This work introduces a quality-preserving staging concept for ultra-intense ion bunches that is seamlessly applicable from the non-relativistic plasma source to the relativistic regime. Full 3D particle-in-cell simulations prove robustness and capture of a high-charge proton bunch, suitable for readily available and near-term laser facilities.

physics.acc-ph↗

arXiv2024

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

Particle accelerator modeling is an important field of research and development, essential to investigating, designing and operating some of the most complex scientific devices ever built. Kinetic simulations of relativistic, charged particle beams and advanced plasma accelerator elements are often performed with high-fidelity particle-in-cell simulations, some of which fill the largest GPU supercomputers. Start-to-end modeling of a particle accelerator includes many elements and it is desirable to integrate and model advanced accelerator elements fast, in effective models. Traditionally, analytical and reduced-physics models fill this role. The vast data from high-fidelity simulations and power of GPU-accelerated computation open a new opportunity to complement traditional modeling without approximations: surrogate modeling through machine learning. In this paper, we implement, present and benchmark such a data-driven workflow, synthesising a fully GPU-accelerated, conventional-surrogate simulation for hybrid particle accelerator beamlines.

physics.acc-ph↗

arXiv2023

From Compact Plasma Particle Sources to Advanced Accelerators with Modeling at Exascale

Developing complex, reliable advanced accelerators requires a coordinated, extensible, and comprehensive approach in modeling, from source to the end of beam lifetime. We present highlights in Exascale Computing to scale accelerator modeling software to the requirements set for contemporary science drivers. In particular, we present the first laser-plasma modeling on an exaflop supercomputer using the US DOE Exascale Computing Project WarpX. Leveraging developments for Exascale, the new DOE SCIDAC-5 Consortium for Advanced Modeling of Particle Accelerators (CAMPA) will advance numerical algorithms and accelerate community modeling codes in a cohesive manner: from beam source, over energy boost, transport, injection, storage, to application or interaction. Such start-to-end modeling will enable the exploration of hybrid accelerators, with conventional and advanced elements, as the next step for advanced accelerator modeling. Following open community standards, we seed an open ecosystem of codes that can be readily combined with each other and machine learning frameworks. These will cover ultrafast to ultraprecise modeling for future hybrid accelerator design, even enabling virtual test stands and twins of accelerators that can be used in operations.

physics.acc-ph↗

arXiv2022

Next Generation Computational Tools for the Modeling and Design of Particle Accelerators at Exascale

Particle accelerators are among the largest, most complex devices. To meet the challenges of increasing energy, intensity, accuracy, compactness, complexity and efficiency, increasingly sophisticated computational tools are required for their design and optimization. It is key that contemporary software take advantage of the latest advances in computer hardware and scientific software engineering practices, delivering speed, reproducibility and feature composability for the aforementioned challenges. A new open source software stack is being developed at the heart of the Beam pLasma Accelerator Simulation Toolkit (BLAST) by LBNL and collaborators, providing new particle-in-cell modeling codes capable of exploiting the power of GPUs on Exascale supercomputers. Combined with advanced numerical techniques, such as mesh-refinement, and intrinsic support for machine learning, these codes are primed to provide ultrafast to ultraprecise modeling for future accelerator design and operations.

physics.acc-ph↗

arXiv2022

Using Kernel-Based Statistical Distance to Study the Dynamics of Charged Particle Beams in Particle-Based Simulation Codes

Measures of discrepancy between probability distributions (statistical distance) are widely used in the fields of artificial intelligence and machine learning. We describe how certain measures of statistical distance can be implemented as numerical diagnostics for simulations involving charged-particle beams. Related measures of statistical dependence are also described. The resulting diagnostics provide sensitive measures of dynamical processes important for beams in nonlinear or high-intensity systems, which are otherwise difficult to characterize. The focus is on kernel-based methods such as Maximum Mean Discrepancy, which have a well-developed mathematical foundation and reasonable computational complexity. Several benchmark problems and examples involving intense beams are discussed. While the focus is on charged-particle beams, these methods may also be applied to other many-body systems such as plasmas or gravitational systems.

physics.acc-ph↗

arXiv2021

Extracting Dynamical Frequencies from Invariants of Motion in Finite-Dimensional Nonlinear Integrable Systems

Integrable dynamical systems play an important role in many areas of science, including accelerator and plasma physics. An integrable dynamical system with $n$ degrees of freedom (DOF) possesses $n$ nontrivial integrals of motion, and can be solved, in principle, by covering the phase space with one or more charts in which the dynamics can be described using action-angle coordinates. To obtain the frequencies of motion, both the transformation to action-angle coordinates and its inverse must be known in explicit form. However, no general algorithm exists for constructing this transformation explicitly from a set of $n$ known (and generally coupled) integrals of motion. In this paper we describe how one can determine the dynamical frequencies of the motion as functions of these $n$ integrals in the absence of explicitly-known action-angle variables, and we provide several examples.

physics.acc-ph↗

arXiv2010

Accurate Transfer Maps for Realistic Beamline Elements: Part I, Straight Elements

The behavior of orbits in charged-particle beam transport systems, including both linear and circular accelerators as well as final focus sections and spectrometers, can depend sensitively on nonlinear fringe-field and high-order-multipole effects in the various beam-line elements. The inclusion of these effects requires a detailed and realistic model of the interior and fringe fields, including their high spatial derivatives. A collection of surface fitting methods has been developed for extracting this information accurately from 3-dimensional field data on a grid, as provided by various 3-dimensional finite-element field codes. Based on these realistic field models, Lie or other methods may be used to compute accurate design orbits and accurate transfer maps about these orbits. Part I of this work presents a treatment of straight-axis magnetic elements, while Part II will treat bending dipoles with large sagitta. An exactly-soluble but numerically challenging model field is used to provide a rigorous collection of performance benchmarks.

physics.acc-ph↗