skip to main content

Title: Automation and control of laser wakefield accelerators using Bayesian optimization
Abstract Laser wakefield accelerators promise to revolutionize many areas of accelerator science. However, one of the greatest challenges to their widespread adoption is the difficulty in control and optimization of the accelerator outputs due to coupling between input parameters and the dynamic evolution of the accelerating structure. Here, we use machine learning techniques to automate a 100 MeV-scale accelerator, which optimized its outputs by simultaneously varying up to six parameters including the spectral and spatial phase of the laser and the plasma density and length. Most notably, the model built by the algorithm enabled optimization of the laser evolution that might otherwise have been missed in single-variable scans. Subtle tuning of the laser pulse shape caused an 80% increase in electron beam charge, despite the pulse length changing by just 1%.
Authors:
; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; ; more » ; ; ; ; ; ; ; ; ; « less
Award ID(s):
1804463
Publication Date:
NSF-PAR ID:
10273602
Journal Name:
Nature Communications
Volume:
11
Issue:
1
ISSN:
2041-1723
Sponsoring Org:
National Science Foundation
More Like this
  1. Abstract Temperature control is essential for regulating material properties in laser-based manufacturing. Motion and power of the scanning laser affect local temperature evolution, which in turn determines the a posteriori microstructure. This paper addresses the problem of adjusting the laser speed and power to achieve the desired values of key process parameters: cooling rate and melt pool size. The dynamics of a scanning laser system is modeled by a one-dimensional (1D) heat conduction equation, with laser power as the heat input and heat dissipation to the ambient. Since the model is 1D, length and size are essentially the same. Wemore »pose the problem as a regulation problem in the (moving) laser frame. The first step is to obtain the steady-state temperature distribution and the corresponding input based on the desired cooling rate and melt pool size. The controller adjusts the input around the steady-state feedforward based on the deviation of the measured temperature field from the steady-state distribution. We show that with suitably defined outputs, the system is strictly passive from the laser motion and power. To avoid over-reliance on the model, the steady-state laser speed and power are adaptively updated, resulting in an integral-like update law for the feedforward. Moreover, the heat transfer coefficient to the ambient may be uncertain, and can also be adaptively updated. The final form of the control law combines passive error temperature field feedback with adaptive feedforward and parameter estimation. The closed-loop asymptotical stability is shown using the Lyapunov arguments, and the controller performance is demonstrated in a simulation.« less
  2. Abstract Due to the highly nonlinear nature of the beam-loading, it is currently not possible to analytically determine the beam parameters needed in a two-bunch plasma wakefield accelerator for maintaining a low energy spread. Therefore in this paper, by using the Broyden–Fletcher–Goldfarb–Shanno algorithm for the parameter scanning with the code QuickPIC and the polynomial regression together with k -fold cross-validation method, we obtain two fitting formulas for calculating the parameters of tri-Gaussian electron beams when minimizing the energy spread based on the beam-loading effect in a nonlinear plasma wakefield accelerator. One formula allows the optimization of the normalized charge permore »unit length of a trailing beam to achieve the minimal energy spread, i.e. the optimal beam-loading. The other one directly gives the transformer ratio when the trailing beam achieves the optimal beam-loading. A simple scaling law for charges of drive beams and trailing beams is obtained from the fitting formula, which indicates that the optimal beam-loading is always achieved for a given charge ratio of the two beams when the length and separation of two beams and the plasma density are fixed. The formulas can also help obtain the optimal plasma densities for the maximum accelerated charge and the maximum acceleration efficiency under the optimal beam-loading respectively. These two fitting formulas will significantly enhance the efficiency for designing and optimizing a two-bunch plasma wakefield acceleration stage.« less
  3. Continuous advancements in LiDAR technology have enabled compelling wind turbulence measurements within the atmospheric boundary layer with range gates shorter than 20 m and sampling frequency of the order of 10 Hz. However, estimates of the radial velocity from the back-scattered laser beam are inevitably affected by an averaging process within each range gate, generally modeled as a convolution between the actual velocity projected along the LiDAR line-of-sight and a weighting function representing the energy distribution of the laser pulse along the range gate. As a result, the spectral energy of the turbulent velocity fluctuations is damped within the inertialmore »sub-range with respective reduction of the velocity variance, and, thus, not allowing to take advantage of the achieved spatio-temporal resolution of the LiDAR technology. In this article, we propose to correct this turbulent energy damping on the LiDAR measurements by reversing the effect of a low-pass filter, which can be estimated directly from the LiDAR measurements. LiDAR data acquired from three different field campaigns are analyzed to describe the proposed technique, investigate the variability of the filter parameters and, for one dataset, assess the procedure for spectral LiDAR correction against sonic anemometer data. It is found that the order of the low-pass filter used for modeling the energy damping on the LiDAR velocity measurements has negligible effects on the correction of the second-order statistics of the wind velocity. In contrast, its cutoff frequency plays a significant role in the spectral correction encompassing the smoothing effects connected with the LiDAR gate length.« less
  4. The generation of colloidal solutions of chemically clean nanoparticles through pulsed laser ablation in liquids (PLAL) has evolved into a thriving research field that impacts industrial applications. The complexity and multiscale nature of PLAL make it difficult to untangle the various processes involved in the generation of nanoparticles and establish the dependence of nanoparticle yield and size distribution on the irradiation parameters. Large-scale atomistic simulations have yielded important insights into the fundamental mechanisms of ultrashort (femtoseconds to tens of picoseconds) PLAL and provided a plausible explanation of the origin of the experimentally observed bimodal nanoparticle size distributions. In this paper,more »we extend the atomistic simulations to short (hundreds of picoseconds to nanoseconds) laser pulses and focus our attention on the effect of the pulse duration on the mechanisms responsible for the generation of nanoparticles at the initial dynamic stage of laser ablation. Three distinct nanoparticle generation mechanisms operating at different stages of the ablation process and in different parts of the emerging cavitation bubble are identified in the simulations. These mechanisms are (1) the formation of a thin transient metal layer at the interface between the ablation plume and water environment followed by its decomposition into large molten nanoparticles, (2) the nucleation, growth, and rapid cooling/solidification of small nanoparticles at the very front of the emerging cavitation bubble, above the transient interfacial metal layer, and (3) the spinodal decomposition of a part of the ablation plume located below the transient interfacial layer, leading to the formation of a large population of nanoparticles growing in a high-temperature environment through inter-particle collisions and coalescence. The coexistence of the three distinct mechanisms of the nanoparticle formation at the initial stage of the ablation process can be related to the broad nanoparticle size distributions commonly observed in nanosecond PLAL experiments. The strong dependence of the nanoparticle cooling and solidification rates on the location within the low-density metal–water mixing region has important implications for the long-term evolution of the nanoparticle size distribution, as well as for the ability to quench the nanoparticle growth or dope them by adding surface-active agents or doping elements to the liquid environment.« less
  5. The efficiency of an accelerator depends on three factors—mapping, deep neural network (DNN) layers, and hardware—constructing extremely complicated design space of DNN accelerators. To demystify such complicated design space and guide the DNN accelerator design for better efficiency, we propose an analytical cost model, MAESTRO. MAESTRO receives DNN model description and hardware resources information as a list, and mapping described in a data-centric representation we propose as inputs. The data centric representation consists of three directives that enable concise description of mappings in a compiler-friendly form. MAESTRO analyzes various forms of data reuse in an accelerator based on inputs quicklymore »and generates more than 20 statistics including total latency, energy, throughput, etc., as outputs. MAESTRO’s fast analysis enables various optimization tools for DNN accelerators such as hardware design exploration tool we present as an example.« less