Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Bayesian Robustness”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 325 records · Page 18

Opportunities in AI/ML for the Rubin LSST Dark Energy Science Collaboration

The Vera C. Rubin Observatory's Legacy Survey of Space and Time (LSST) will produce unprecedented volumes of heterogeneous astronomical data (images, catalogs, and alerts) that challenge traditional analysis pipelines. The LSST Dark Energy Science Collaboration (DESC) aims to derive robust constraints on dark energy and dark matter from these data, requiring methods that are statistically powerful, scalable, and operationally reliable. Artificial intelligence and machine learning (AI/ML) are already embedded across DESC science workflows, from photometric redshifts and transient classification to weak lensing inference and cosmological simulations. Yet their utility for precision cosmology hinges on trustworthy uncertainty quantification, robustness to covariate shift and model misspecification, and reproducible integration within scientific pipelines. This white paper surveys the current landscape of AI/ML across DESC's primary cosmological probes and cross-cutting analyses, revealing that the same core methodologies and fundamental challenges recur across disparate science cases. Since progress on these cross-cutting challenges would benefit multiple probes simultaneously, we identify key methodological research priorities, including Bayesian inference at scale, physics-informed methods, validation frameworks, and active learning for discovery. With an eye on emerging techniques, we also explore the potential of the latest foundation model methodologies and LLM-driven agentic AI systems to reshape DESC workflows, provided their deployment is coupled with rigorous evaluation and governance. Finally, we discuss critical software, computing, data infrastructure, and human capital requirements for the successful deployment of these new methodologies, and consider associated risks and opportunities for broader coordination with external actors.

Aubourg, Eric [APC, Paris] (ORCID:000000025592023X↗

Process–structure–property relationships in bimodal machined microstructures using robust structure descriptors

Machining is a severe plastic deformation process that subjects materials to high rates of deformation and elevated temperatures. Dynamic recrystallization during severe plastic deformation drives grain refinement into the sub-micron range but the ductility and thermal stability of these structures is poor. In contrast, bimodal microstructures consisting of both fine and coarse grains have been shown to exhibit attractive strength and ductility properties at these temperatures. In this paper, we study the process–structure–property relationships for pure copper subject to a machining process where the cutting speeds are varied. Further, we investigate the role of thermal effects by varying the post-deformation cooling rates. Microstructures generated under these deformation conditions are quantified using angularly resolved chord length statistics. The derived metrics are found to be robust descriptors of the studied microstructures as they automatically capture complex features such as unimodal and bimodal distributions of grain structure, grain constituent length scales, and morphological anisotropy induced by shear deformation. The uniaxial equivalent yield strength of the generated structures are estimated using spherical nanoindentation and inverse modeling techniques. Lastly, we present a methodology for identifying the constrained property-process inverse mapping for machining using a Bayesian framework and the established forward process–structure–property mapping.

36 MATERIALS SCIENCE↗

MOA-2011-BLG-293Lb: A TEST OF PURE SURVEY MICROLENSING PLANET DETECTIONS

Because of the development of large-format, wide-field cameras, microlensing surveys are now able to monitor millions of stars with sufficient cadence to detect planets. These new discoveries will span the full range of significance levels including planetary signals too small to be distinguished from the noise. At present, we do not understand where the threshold is for detecting planets. MOA-2011-BLG-293Lb is the first planet to be published from the new surveys, and it also has substantial follow-up observations. This planet is robustly detected in survey+follow-up data ({Delta}{chi}{sup 2} {approx} 5400). The planet/host mass ratio is q = (5.3 {+-} 0.2) Multiplication-Sign 10{sup -3}. The best-fit projected separation is s = 0.548 {+-} 0.005 Einstein radii. However, due to the s{r_reversible}s{sup -1} degeneracy, projected separations of s{sup -1} are only marginally disfavored at {Delta}{chi}{sup 2} = 3. A Bayesian estimate of the host mass gives M{sub L} = 0.43{sup +0.27}{sub -0.17} M{sub Sun }, with a sharp upper limit of M{sub L} < 1.2 M{sub Sun} from upper limits on the lens flux. Hence, the planet mass is m{sub p} = 2.4{sup +1.5}{sub -0.9} M{sub Jup}, and the physical projected separation is either r {approx_equal} 1.0 AU or r {approx_equal} 3.4 AU. We show that survey data alone predict this solution and are able to characterize the planet, but the {Delta}{chi}{sup 2} is much smaller ({Delta}{chi}{sup 2} {approx} 500) than with the follow-up data. The {Delta}{chi}{sup 2} for the survey data alone is smaller than for any other securely detected planet. This event suggests a means to probe the detection threshold, by analyzing a large sample of events like MOA-2011-BLG-293, which have both follow-up data and high-cadence survey data, to provide a guide for the interpretation of pure survey microlensing data.

79 ASTRONOMY AND ASTROPHYSICS↗

Bayesian optimization of the beam injection process into a storage ring

We have evaluated the data-efficient Bayesian optimization method for the specific task of injection tuning in a circular accelerator. In this paper, we describe the implementation of this method at the Karlsruhe Research Accelerator with up to nine tuning parameters, including the determination of the associated hyperparameters. We show that the Bayesian optimization method outperforms manual tuning and the commonly used Nelder-Mead optimization algorithm in both simulation and experiment. The algorithm was also successfully used to ease the commissioning phase after the installation of new injection magnets and is regularly used during accelerator operations. We demonstrate that the introduction of context variables that include intrabunch scattering effects, such as the Touschek effect, further improves the control and robustness of the injection process.

43 PARTICLE ACCELERATORS↗

Characterizing and Mitigating Intraday Variability: Reconstructing Source Structure in Accreting Black Holes with mm-VLBI

The extraordinary physical resolution afforded by the Event Horizon Telescope has opened a window onto the astrophysical phenomena unfolding on horizon scales in two known black holes, M87* and Sgr A*. However, with this leap in resolution has come a new set of practical complications. Sgr A* exhibits intraday variability that violates the assumptions underlying Earth aperture synthesis, limiting traditional image reconstruction methods to short timescales and data sets with very sparse (u, v) coverage. We present a new set of tools to detect and mitigate this variability. We develop a data-driven, model-agnostic procedure to detect and characterize the spatial structure of intraday variability. This method is calibrated against a large set of mock data sets, producing an empirical estimator of the spatial power spectrum of the brightness fluctuations. We present a novel Bayesian noise modeling algorithm that simultaneously reconstructs an average image and statistical measure of the fluctuations about it using a parameterized form for the excess variance in the complex visibilities not otherwise explained by the statistical errors. These methods are validated using a variety of simulated data, including general relativistic magnetohydrodynamic simulations appropriate for Sgr A* and M87*. We find that the reconstructed source structure and variability are robust to changes in the underlying image model. We apply these methods to the 2017 EHT observations of M87*, finding evidence for variability across the EHT observing campaign. The variability mitigation strategies presented are widely applicable to very long baseline interferometry observations of variable sources generally, for which they provide a data-informed averaging procedure and natural characterization of inter-epoch image consistency.

79 ASTRONOMY AND ASTROPHYSICS↗

Non-Intrusive Appliance Identification with Appliance-Specific Networks

The problem of noninstrusive load monitoring (NILM) is usually formulated as a single-channel blind source separation task, whose successful solution enable fast and convenient load identification and energy disaggregation. When applied at test time, NILM algorithms aim to identify the operating characteristics of individual appliances from an aggregate power measurement of the entire house. Recent advances in deep learning gave rise to many methods that mostly focus on learning a direct mapping from aggregate measurement to individual appliance power. However, these methods are not only computationally expensive, but they often suffer from overfitting and do not generalize very well. In this article, we propose a novel NILM method that leverages advances in statistical learning that have not been properly applied in this domain before. The proposed method consists of three stages: first, a Bayesian nonparametric learning-based approach for appliance state extraction; second, synthetic minority oversampling technique for data augmentation and mitigating the heavy imbalance in switching events; and third, appliance-specific lightweight long short-term memory networks for status classification for each appliance. Here, we adopt a “differential” input (the difference before and after the switching event) to reduce the complexity of network training and make the proposed method robust to multiappliance switching events. Experiments are conducted to demonstrate the effectiveness of the proposed method, achieving superior performance when compared to recent methods. An ablation study is conducted to demonstrate the effectiveness of each module of our method. Finally, we investigate the quality of generated synthetic samples.

42 ENGINEERING↗

A framework for strategic discovery of credible neural network surrogate models under uncertainty

The widespread integration of deep neural networks in developing data-driven surrogate models for high-fidelity simulations of complex physical systems highlights the critical necessity for robust uncertainty quantification techniques and credibility assessment methodologies, ensuring the reliable deployment of surrogate models in consequential decision-making. Here, this study presents the Occam Plausibility Algorithm for surrogate models (OPAL-surrogate), providing a systematic framework to uncover predictive neural network-based surrogate models within the large space of potential models, including various neural network classes and choices of architecture and hyperparameters. The framework is grounded in hierarchical Bayesian inferences and employs model validation tests to evaluate the credibility and prediction reliability of the surrogate models under uncertainty. Leveraging these principles, OPAL-surrogate introduces a systematic and efficient strategy for balancing the trade-off between model complexity, accuracy, and prediction uncertainty. The effectiveness of OPAL-surrogate is demonstrated through two modeling problems, including the deformation of porous materials for building insulation and turbulent combustion flow for ablation of solid fuels within hybrid rocket motors.

42 ENGINEERING↗

How Well Do We Know the Neutron-Matter Equation of State at the Densities Inside Neutron Stars? A Bayesian Approach with Correlated Uncertainties

Here, we introduce a new framework for quantifying correlated uncertainties of the infinite-matter equation of state derived from chiral effective field theory (𝜒⁢EFT ). Bayesian machine learning via Gaussian processes with physics-based hyperparameters allows us to efficiently quantify and propagate theoretical uncertainties of the equation of state, such as 𝜒⁢EFT truncation errors, to derived quantities. We apply this framework to state-of-the-art many-body perturbation theory calculations with nucleon-nucleon and three-nucleon interactions up to fourth order in the 𝜒⁢EFT expansion. This produces the first statistically robust uncertainty estimates for key quantities of neutron stars. We give results up to twice nuclear saturation density for the energy per particle, pressure, and speed of sound of neutron matter, as well as for the nuclear symmetry energy and its derivative. At nuclear saturation density, the predicted symmetry energy and its slope are consistent with experimental constraints.

79 ASTRONOMY AND ASTROPHYSICS↗

Dynamical Sketching for Enhanced Communication Efficiency in Federated Learning

Federated learning (FL) has revolutionized distributed machine learning by enabling collaborative model training without sharing local data. However, communication efficiency and privacy guarantees remain significant challenges. This paper introduces a dynamic sketching mechanism in FL, optimizing the trade-off between communication efficiency and model accuracy. By dynamically selecting the sketch matrix size, our approach adapts to the evolving characteristics of the data and the model, ensuring optimal performance across diverse scenarios. We leverage Bayesian optimization to systematically tune the sketch parameters, achieving an effective balance between resource efficiency and model performance. Experimental results on the MNIST dataset using a convolutional neural network (CNN) architecture validate the proposed method's efficiency and scalability. Our dynamic sketching approach significantly outperforms fixed-size sketching techniques, achieving higher compression ratios (up to 62x) and providing better privacy guarantees while maintaining high model accuracy. These findings highlight the robustness and versatility of our approach and make it a valuable solution for privacy-preserving, communication-efficient federated learning.

Afrose, Sharmin [ORNL]↗

Data Assimilation for Robust UQ Within Agent-Based Simulation on HPC Systems

Agent-based simulation provides a powerful tool for in silico system modeling. However, these simulations do not provide built-in methods for uncertainty quantification (UQ). Within these types of models a typical approach to UQ is to run multiple realizations of the model then compute aggregate statistics. This approach is limited due to the compute time required for a solution. When faced with an emerging biothreat, public health decisions need to be made quickly and solutions for integrating near real-time data with analytic tools are needed. We propose an integrated Bayesian UQ framework for agent-based models based on sequential Monte Carlo sampling. Given streaming or static data about the evolution of an emerging pathogen this Bayesian framework provides a distribution over the parameters governing the spread of a disease through a population. These estimates of the spread of a disease may be provided to public health agencies seeking to abate the spread. By coupling agent-based simulations with Bayesian modeling in a data assimilation, our proposed framework provides a powerful tool for modeling dynamical systems in silico. We propose a method which reduces model error and provides a range of realistic possible outcomes. Moreover, our method addresses two primary limitations of ABMs: the lack of UQ and an inability to assimilate data. Our proposed framework combines the flexibility of an agent-based model with UQ provided by the Bayesian paradigm in a workflow which scales well to HPC systems. We provide algorithmic details and results on a simulated outbreak with both static and streaming data.

Spannaus, Adam [ORNL] (ORCID:0000000225213657)↗

Deconvolving distribution of relaxation times, resistances and inductance from electrochemical impedance spectroscopy via statistical model selection: Exploiting structural-sparsity regularization and data-driven parameter tuning

The distribution of relaxation times (DRT) has drawn increasing attention for interpreting electrochemical impedance spectroscopy (EIS). Deconvolution of DRT from EIS is a challenging ill-posed problem that requires regularization methods. In this work, we formulate DRT reconstruction task as a statistical model selection problem with structural-sparsity penalties. We utilize the Elastic net regularization that simultaneously benefits from Ridge and Lasso regularizations with optimal tuning parameter automatically determined by the information criteria. We benchmark our approach on four synthetic experiments (a ZARC element, ZARC mixtures, a RC circuit and a Fractal element) and two real EIS datasets of a Lithium ion battery and an organic-inorganic halide class of perovskites in oxygen environment at different gas pressures. We demonstrate the superiority of proposed model selection procedure, that is capable of eliminating pseudo peaks and representing asymmetries in DRT as well as precisely estimating resistances. Here, we highlight our approach is robust to reducing and subsampling EIS frequency range, making it a promising tool for timing-resolved, localized and large scale EIS data analysis. For the Lithium ion battery data analysis, we extend the classical DRT model to incorporate the inductive effect and illustrate DRT as a guidance for equivalent circuit modeling to refine impedance reconstruction at low risks of overfitting. Furthermore, the structural-sparsity regularization could be extended for multidimensional and Bayesian EIS data analysis.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Multipoint-BAX: a new approach for efficiently tuning particle accelerator emittance via virtual objectives

Abstract Although beam emittance is critical for the performance of high-brightness accelerators, optimization is often time limited as emittance calculations, commonly done via quadrupole scans, are typically slow. Such calculations are a type of multipoint query , i.e. each query requires multiple secondary measurements. Traditional black-box optimizers such as Bayesian optimization are slow and inefficient when dealing with such objectives as they must acquire the full series of measurements, but return only the emittance, with each query. We propose a new information-theoretic algorithm, Multipoint-BAX , for black-box optimization on multipoint queries, which queries and models individual beam-size measurements using techniques from Bayesian Algorithm Execution (BAX). Our method avoids the slow multipoint query on the accelerator by acquiring points through a virtual objective , i.e. calculating the emittance objective from a fast learned model rather than directly from the accelerator. We use Multipoint-BAX to minimize emittance at the Linac Coherent Light Source (LCLS) and the Facility for Advanced Accelerator Experimental Tests II (FACET-II). In simulation, our method is 20× faster and more robust to noise compared to existing methods. In live tests, it matched the hand-tuned emittance at FACET-II and achieved a 24% lower emittance than hand-tuning at LCLS. Our method represents a conceptual shift for optimizing multipoint queries, and we anticipate that it can be readily adapted to similar problems in particle accelerators and other scientific instruments.

43 PARTICLE ACCELERATORS↗

Probabilistic Photometric Redshifts in the Era of Petascale Astronomy

With the growth of large photometric surveys, accurately estimating photometric redshifts, preferably as a probability density function (PDF), and fully understanding the implicit systematic uncertainties in this process has become increasingly important. These surveys are expected to obtain images of billions of distinct galaxies. As a result, storing and analyzing all of these photometric redshift PDFs will be non-trivial, and this challenge becomes even more severe if a survey plans to compute and store multiple different PDFs. In this thesis, we have developed an end-to-end framework that will compute accurate and robust photometric redshift PDFs for massive data sets by using two new, state-of-the-art machine learning techniques that are based on a random forest and a random atlas, respectively. By using data from several photometric surveys, we demonstrate the applicability of these new techniques, and we demonstrate that our new approach is among the best techniques currently available. We also show how different techniques can be combined by using novel Bayesian techniques to improve the photometric redshift precision to unprecedented levels while also presenting new approaches to better identify outliers. In addition, our framework provides supplementary information regarding the data being analyzed, including unbiased estimates of the accuracy of the technique without resorting to a validation data set, identification of poor photometric redshift areas within the parameter space occupied by the spectroscopic training data, and a quantification of the relative importance of the variables used during the estimation process. Furthermore, we present a new approach to represent and store photometric redshift PDFs by using a sparse representation with outstanding compression and reconstruction capabilities. We also demonstrate how this framework can also be directly incorporated into cosmological analyses. The new techniques presented in this thesis are crucial to enable the development of precision cosmology in the era of petascale astronomical surveys.

79 ASTRONOMY AND ASTROPHYSICS↗

A neural networks framework for real-time unfolding of neutron spectroscopic data at JET

A determination of fast ion population parameters such as intensity and kinetic temperature is important for fusion reactors. This becomes more challenging with finer time resolution of the measurements, since the limited data in each time slice cause increasing statistical variations in the data. This paper describes a framework using Bayesian-regularized neural networks (NNs) designed for such a task. The method is applied to the TOFOR 2.5 MeV fusion neutron spectrometer at JET. NN training data are generated by random sampling of variables in neutron spectroscopy models. Ranges and probability distributions of the parameters are chosen to match the experimental data. Results have shown good performance both on synthetic and experimental data. The latter was assessed by statistical considerations and by examining the robustness and time consistency of the results. The regularization of the training algorithm allowed for higher time resolutions than simple forward methods. The fast execution time makes this approach suitable for real-time analysis with a time resolution limit in the microsecond time scale.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

MC 2 : A Deeper Look at ZwCl 2341.1+0000 with Bayesian Galaxy Clustering and Weak Lensing Analyses

ZwCl 2341.1+0000, a merging galaxy cluster with disturbed X-ray morphology and widely separated (~3 Mpc) double radio relics, was thought to be an extremely massive (10 - 30 X 10 14 M⊙) and complex system with little known about its merger history. We present JVLA 2-4 GHz observations of the cluster, along with new spectroscopy from our Keck/DEIMOS survey, and apply Gaussian Mixture Modeling to the three-dimensional distribution of 227 con rmed cluster galaxies. After adopting the Bayesian Information Criterion to avoid over tting, which we discover can bias total dynamical mass estimates high, we nd that a three-substructure model with a total dynamical mass estimate of 9:39 ± 0:81 X 10 14 M⊙ is favored. We also present deep Subaru imaging and perform the rst weak lensing analysis on this system, obtaining a weak lensing mass estimate of 5:57±2:47X10 14 M⊙. This is a more robust estimate because it does not depend on the dynamical state of the system, which is disturbed due to the merger. Our results indicate that ZwCl 2341.1+0000 is a multiple merger system comprised of at least three substructures, with the main merger that produced the radio relics occurring near to the plane of the sky, and a younger merger in the North occurring closer to the line of sight. Dynamical modeling of the main merger reproduces observed quantities (relic positions and polarizations, subcluster separation and radial velocity difference), if the merger axis angle of ~10 +34 -6 degrees and the collision speed at pericenter is ~1900 +300 -200 km/s.

79 ASTRONOMY AND ASTROPHYSICS↗

Efficient Bayesian inference with latent Hamiltonian neural networks in No-U-Turn Sampling

When sampling for Bayesian inference, one popular approach in the computational field is to use Hamiltonian Monte Carlo (HMC) and specifically the No-U-Turn Sampler (NUTS), which automatically decides the end time of the Hamiltonian trajectory. However, HMC and NUTS can require numerous numerical gradients of the target density and can prove slow in practice when relying on computationally expensive forward models. We propose Latent Hamiltonian neural networks (L-HNNs) with HMC and NUTS for solving Bayesian inference problems. Once trained, L-HNNs do not require numerical gradients of the target density during sampling, and hence numerous evaluations of the forward computational model. Moreover, L-HNNs satisfy important properties such as perfect time reversibility and Hamiltonian conservation, making them well-suited for use within HMC and NUTS because stationarity can be shown. We also propose the integration of L-HNNs in an online error monitoring scheme, in which numerical gradients of the target density are used for a few samples whenever the L-HNNs prediction errors are large. This online error monitor scheme prevents sample degeneracy in regions of low probability density and ensures robust uncertainty quantification. We demonstrate L-HNNs in NUTS with online error monitoring on several analytical examples involving complex, heavy-tailed, and high-local-curvature probability densities. We then demonstrate the applicability of L-HNNs in NUTS to two computational case studies, namely the Allen-Cahn stochastic partial differential equation and an elliptic partial differential equation with 25 and 50 inference parameters, respectively. Overall, the L-HNNs in NUTS with online error monitoring satisfactorily inferred these probability densities. In conclusion, compared to traditional NUTS, L-HNNs in NUTS with online error monitoring required 1–2 orders of magnitude fewer numerical gradients of the target density and improved the effective sample size (ESS) per gradient (which is a measure of both the sampling quality and the computational expense) by an order of magnitude.

97 MATHEMATICS AND COMPUTING↗

Parallelizing autotuning for HPC applications: Unveiling the potential of the speculation strategy in Bayesian optimization

In the exascale computing era, tuning High-Performance Computing (HPC) applications has become a significant computational challenge. Although Bayesian optimization (BO) has emerged as a promising tool for HPC performance tuning, the BO workflow is inherently sequential (i.e., one function evaluation at a time) and cannot leverage the huge amount of parallel resources present in modern supercomputers, resulting in a considerable underutilization of their computational capabilities. This paper explores the trade-off between search quality and parallelism in BO, investigating a diverse set of methods. Building upon both previous approaches from the literature and novel methodologies introduced in this work, our study provides a deep analysis to accelerate BO performance tuning. By examining a set of synthetic functions and practical HPC applications, our exploration analyzes the interaction among various BO methods for parallelization, the quantity of parallel resources, the runtime distribution of target HPC applications, and the costs associated with different search orchestration mechanisms that have been overlooked in previous studies. Compared to sequential BO, our novel methodology achieves comparable quality while demonstrating robust scalability in search time as the amount of parallel resources increases; it also outperforms a state-of-the-art tuner, which supports parallelization, achieving up to 3.67x faster search time. We provide high-value insights for practitioners seeking to leverage the power of parallel computing for efficient HPC application tuning. Additionally, to further assist researchers in accelerating the performance tuning of their HPC applications, we provide an extension of an existing open-source tuning framework that incorporates our methods.

Bayesian optimization↗

Bayesian optimization to design a novel x-ray shaping device

In radiation therapy, x-ray dose must be precisely sculpted to the tumor, while simultaneously avoiding surrounding organs at risk. This requires modulation of x-ray intensity in space and/or time. Typically, this is achieved using a multi leaf collimator (MLC) - a complex mechatronic device comprising over one hundred individually powered tungsten ‘leaves’ that move in or out of the radiation field as required. Here, an all-electronic x-ray collimation concept with no moving parts is presented, termed “SPHINX”: Scanning Pencil-beam High-speed Intensity-modulated X-ray source. SPHINX utilizes a spatially distributed bremsstrahlung target and collimator array in conjunction with magnetic scanning of a high energy electron beam to generate a plurality of small x-ray “beamlets.” A simulation framework was developed in Topas Monte Carlo incorporating a phase space electron source, transport through user defined magnetic fields, bremsstrahlung x-ray production, transport through a SPHINX collimator, and dose in water. This framework was completely parametric, meaning a simulation could be built and run for any supplied geometric parameters. This functionality was coupled with Bayesian optimization to find the best parameter set based on an objective function which included terms to maximize dose rate for a user defined beamlet width while constraining inter-channel cross talk and electron contamination. Designs for beamlet widths of 5, 7, and 10 mm 2 were generated. Each optimization was run for 300 iterations and took approximately 40 h on a 24-core computer. For the optimized 7-mm model, a simulation of all beamlets in water was carried out including a linear scanning magnet calibration simulation. Finally, a back-of-envelope dose rate formalism was developed and used to estimate dose rate under various conditions. The optimized 5–, 7–, and 10-mm models had beamlet widths of 5.1 , 7.2 , and 10.1 mm 2 and dose rates of 3574, 6351, and 10 015 Gy/C, respectively. The reduction in dose rate for smaller beamlet widths is a result of both increased collimation and source occlusion. For the simulation of all beamlets in water, the scanning magnet calibration reduced the offset between the collimator channels and beam centroids from 2.9 ±1.9 mm to 0.01 ±0.03 mm. A slight reduction in dose rate of approximately 2% per degree of scanning angle was observed. Based on a back-of-envelope dose rate formalism, SPHINX in conjunction with next-generation linear accelerators has the potential to achieve substantially higher dose rates than conventional MLC-based delivery, with delivery of an intensity modulated 100 x 100 mm 2 field achievable in 0.9 to 10.6 s depending on the beamlet widths used. Bayesian optimization was coupled with Monte Carlo modeling to generate SPHINX geometries for various beamlet widths. A complete Monte Carlo simulation for one of these designs was developed, including electron beam transport of all beamlets through scanning magnets, x-ray production and collimation, and dose in water. These results demonstrate that SPHINX is a promising candidate for sculpting radiation dose with no moving parts, and has the potential to vastly improve both the speed and robustness of radiotherapy delivery. A multi-beam SPHINX system may be a candidate for delivering magavoltage FLASH RT in humans.

62 RADIOLOGY AND NUCLEAR MEDICINE↗