A5068226355- Particle accelerators and beam dynamics
- Particle Accelerators and Free-Electron Lasers
- Particle Detector Development and Performance
- Superconducting Materials and Applications
- Magnetic confinement fusion research
- Advanced X-ray Imaging Techniques
- Laser-Plasma Interactions and Diagnostics
- Advanced Multi-Objective Optimization Algorithms
- Electron and X-Ray Spectroscopy Techniques
- Advanced Electron Microscopy Techniques and Applications
- Gyrotron and Vacuum Electronics Research
- Advancements in Photolithography Techniques
- Scientific Computing and Data Management
- Microfluidic and Capillary Electrophoresis Applications
- Plasma Diagnostics and Applications
- Laser Design and Applications
- Machine Learning and Data Classification
- Gaussian Processes and Bayesian Inference
- Magnetic Properties and Applications
- Spectroscopy and Laser Applications
- Radiation Therapy and Dosimetry
- Nuclear reactor physics and engineering
- Reservoir Engineering and Simulation Methods
- Advanced Bandit Algorithms Research
- Muon and positron interactions and applications
SLAC National Accelerator Laboratory
2018-2024
Stanford University
2020-2023
Fermi National Accelerator Laboratory
2016-2021
Fermi Research Alliance
2016-2021
University of Chicago
2021
Los Alamos Medical Center
2021
Menlo School
2018
Colorado State University
2015-2017
The Linac Coherent Light Source changes configurations multiple times per day, necessitating fast tuning strategies to reduce setup time for successive experiments. To this end, we employ a Bayesian approach transport optics optimize groups of quadrupole magnets. We use Gaussian process provide probabilistic model the machine response with respect control parameters from modest number samples. Subsequent samples are selected during optimization using statistical test combining prediction and...
We report on the application of machine learning (ML) methods for predicting longitudinal phase space (LPS) distribution particle accelerators. Our approach consists training a ML-based virtual diagnostic to predict LPS using only nondestructive linac and e-beam measurements as inputs. validate this with simulation study FACET-II an experimental demonstration conducted at LCLS. At LCLS, images are obtained transverse deflecting cavity used data our ML model. In both LCLS cases we find good...
High-fidelity physics simulations are powerful tools in the design and optimization of charged particle accelerators. However, computational burden these often limits their use practice for experiment planning. It also precludes as on-line models tied directly to accelerator operation. We introduce an approach based on machine learning create nonlinear, fast-executing surrogate that informed by a sparse sampling simulation. The $\mathcal{O}({10}^{6})--\mathcal{O}({10}^{7})$ times more...
The full optimization of the design and operation instruments whose functioning relies on interaction radiation with matter is a super-human task, due to large dimensionality space possible choices for geometry, detection technology, materials, data-acquisition, information-extraction techniques, interdependence related parameters. On other hand, massive potential gains in performance over standard, "experience-driven" layouts are principle within our reach if an objective function fully...
Characterizing the phase space distribution of particle beams in accelerators is a central part understanding beam dynamics and improving accelerator performance. However, conventional analysis methods either use simplifying assumptions or require specialized diagnostics to infer high-dimensional (>2D) properties. In this Letter, we introduce general-purpose algorithm that combines neural networks with differentiable tracking efficiently reconstruct distributions without using manipulations....
Accelerator physics relies on numerical algorithms to solve optimization problems in online accelerator control and tasks such as experimental design model calibration simulations. The effectiveness of discovering ideal solutions for complex challenges with limited resources often determines the problem complexity these methods can address. community has recognized advantages Bayesian algorithms, which leverage statistical surrogate models objective functions effectively address challenges,...
Particle accelerators are host to myriad nonlinear and complex physical phenomena. They often involve a multitude of interacting systems, subject tight performance demands, should be able run for extended periods time with minimal interruptions. Often times, traditional control techniques cannot fully meet these requirements. One promising avenue is introduce machine learning sophisticated inspired by artificial intelligence, particularly in light recent theoretical practical advances...
The dynamics of intense electron bunches in free lasers and plasma wakefield accelerators are dominated by complex collective effects such as wakefields, space charge, coherent synchrotron radiation, drift unpredictably with time, making it difficult to control tune beam properties using model-based approaches. We report on a first its kind combination automatic, model-independent feedback neural network for the longitudinal phase relativistic beams femtosecond resolution based only...
Particle accelerators require constant tuning during operation to meet beam quality, total charge and particle energy requirements for use in a wide variety of physics, chemistry biology experiments. Maximizing the performance an accelerator facility often necessitates multiobjective optimization, where operators must balance trade-offs between multiple competing objectives simultaneously, using limited, temporally expensive observations. Usually, optimization problems are solved off-line,...
Ultrafast electron diffraction using MeV energy beams(MeV-UED) has enabled unprecedented scientific opportunities in the study of ultrafast structural dynamics a variety gas, liquid and solid state systems. Broad applications usually pose different requirements for probe properties. Due to complex, nonlinear correlated nature accelerator systems, beam property optimization is time-taking process often relies on extensive hand-tuning by experienced human operators. Algorithm based efficient...
Beams with cross-plane coupling or extreme asymmetries between the two transverse phase spaces are often encountered in particle accelerators. Flat beams large transverse-emittance ratios critical for future linear colliders. Similarly, magnetized significant expected to enhance performance of electron cooling hadron beams. Preparing these requires precise control and characterization four-dimensional space. In this study, we employ generative phase-space reconstruction techniques rapidly...
Next-generation accelerator concepts, which hinge on the precise shaping of beam distributions, demand equally diagnostic methods capable reconstructing distributions within six-dimensional position-momentum spaces. However, characterization intricate features using current techniques necessitates a substantial number measurements, many hours valuable time. Novel phase space reconstruction are needed to reduce measurements required reconstruct detailed, high-dimensional in order resolve...
Abstract Particle accelerator operation requires simultaneous optimization of multiple objectives. Multi-Objective Optimization (MOO) is particularly challenging due to trade-offs between the Evolutionary algorithms, such as genetic algorithm (GA), have been leveraged for many problems, however, they do not apply complex control problems by design. This paper demonstrates power differentiability solving MOO using a Deep Differentiable Reinforcement Learning (DDRL) in particle accelerators....
High-dimensional optimization is a critical challenge for operating large-scale scientific facilities. We apply physics-informed Gaussian process (GP) optimizer to tune complex system by conducting efficient global search. Typical GP models learn from past observations make predictions, but this reduces their applicability new systems where archive data not available. Instead, here we use fast approximate model physics simulations design the model. The then employed inferences sequential...
With the advent of increased computational resources and improved algorithms, machine learning-based models are being increasingly applied to complex problems in particle accelerators. However, such data-driven may provide overly confident predictions with unknown errors uncertainties. For reliable deployment learning high-regret safety-critical systems as accelerators, estimates prediction uncertainty needed along accurate point predictions. In this investigation, we evaluate Bayesian...
Particle accelerators are invaluable discovery engines in the chemical, biological and physical sciences. Characterization of accelerated beam response to accelerator input parameters is of-ten first step when conducting accelerator-based experiments. Currently used techniques for characterization, such as grid-like parameter sampling scans, become impractical extended higher dimensional spaces, complicated measurement constraints present, or prior information known about scarce. In this...
Future improvements in particle accelerator performance are predicated on increasingly accurate online modeling of accelerators. Hysteresis effects magnetic, mechanical, and material components accelerators often neglected models used to inform control algorithms, even though reproducibility errors from systems exhibiting hysteresis not negligible high precision In this Letter, we combine the classical Preisach model with machine learning techniques efficiently create nonparametric,...
Abstract C 3 is an opportunity to realize e + - collider for the study of Higgs boson at √ s = 250 GeV, with a well defined upgrade path 550 GeV while staying on same short facility footprint [2,3]. based fundamentally new approach normal conducting linear accelerators that achieves both high gradient and efficiency relatively low cost. Given advanced state designs, key system requires technical maturation main linac. This paper presents staged towards demonstrate technology Direct (source...
C$^3$ is an opportunity to realize e$^+$e$^-$ collider for the study of Higgs boson at $\sqrt{s} = 250$ GeV, with a well defined upgrade path 550 GeV while staying on same short facility footprint. based fundamentally new approach normal conducting linear accelerators that achieves both high gradient and efficiency relatively low cost. Given advanced state designs, key system requires technical maturation main linac. This white paper presents staged towards demonstrate technology Direct...
Abstract Ultrafast electron diffraction using MeV energy beams(MeV-UED) has enabled unprecedented scientific opportunities in the study of ultrafast structural dynamics a variety gas, liquid and solid state systems. Broad applications usually pose different requirements for probe properties. Due to complex, nonlinear correlated nature accelerator systems, beam property optimization is time-taking process often relies on extensive hand-tuning by experienced human operators. Algorithm based...
Particle accelerators are extremely complex machines that challenging to simulate, design, and control. Over the past decade, artificial intelligence (AI) machine learning (ML) techniques have made dramatic advancements across various scientific industrial domains, rapid improvements been in availability power of computing resources. These developments begun revolutionize way particle designed controlled, AI/ML beginning be incorporated into regular operations for accelerators. This article...
Machine learning (ML) is a subfield of artificial intelligence. The term applies broadly to collection computational algorithms and techniques that train systems from raw data rather than priori models. ML are now technologically mature enough be applied particle accelerators, we expect will become an increasingly valuable tool meet new demands for beam energy, brightness, stability. intent this white paper provide high-level introduction problems in accelerator science operation where...