High-fidelity physics simulations are powerful tools in the design and optimization of charged particle accelerators. However, the computational burden of these simulations often limits their use in practice for design optimization and experiment planning. It also precludes their use as on-line models tied directly to accelerator operation. We introduce an approach based on machine learning to create nonlinear, fastexecuting surrogate models that are informed by a sparse sampling of the physics simulation. The models are Oð10 6 Þ-Oð10 7 Þ times more computationally efficient to execute. We also demonstrate that these models can be reliably used with multiobjective optimization to obtain orders-of-magnitude speedup in initial design studies and experiment planning. For example, we required 132 times fewer simulation evaluations to obtain an equivalent solution for our main test case, and initial studies suggest that between 330-550 times fewer simulation evaluations are needed when using an iterative retraining process. Our approach enables new ways for high-fidelity particle accelerator simulations to be used, at comparatively little computational cost.

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 to transport optics tuning to optimize groups of quadrupole magnets. We use a Gaussian process to provide a probabilistic model of the machine response with respect to control parameters from a modest number of samples. Subsequent samples are selected during optimization using a statistical test combining the model prediction and uncertainty. The model parameters are fit from archived scans, and correlations between devices are added from a simple beam transport model. The result is a sample-efficient optimization routine, which we show significantly outperforms existing optimizers.

We report on the application of machine learning (ML) methods for predicting the longitudinal phase space (LPS) distribution of particle accelerators. Our approach consists of training a ML-based virtual diagnostic to predict the LPS using only nondestructive linac and e-beam measurements as inputs. We validate this approach with a simulation study for the FACET-II linac and with an experimental demonstration conducted at LCLS. At LCLS, the e-beam LPS images are obtained with a transverse deflecting cavity and used as training data for our ML model. In both the FACET-II and LCLS cases we find good agreement between the predicted and simulated/measured LPS profiles, an important step towards showing the feasibility of implementing such a virtual diagnostic on particle accelerators in the future.

The dynamics of intense electron bunches in free electron lasers and plasma wakefield accelerators are dominated by complex collective effects such as wakefields, space charge, coherent synchrotron radiation, and drift unpredictably with time, making it difficult to control and tune beam properties using model-based approaches. We report on a first of its kind combination of automatic, model-independent feedback with a neural network for control of the longitudinal phase space of relativistic electron beams with femtosecond resolution based only on transverse deflecting cavity measurements.