Search NASA⌕ Search

SEARCH · Search NASA

Results for “Robust optimization”

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 1,405 records · Page 78

An optimization-based approach to parameter learning for fractional type nonlocal models

Nonlocal operators of fractional type are a popular modeling choice for applications that do not adhere to classical diffusive behavior; however, one major challenge in nonlocal simulations is the selection of model parameters. In this work we propose an optimization-based approach to parameter identification for fractional models with an optional truncation radius. We formulate the inference problem as an optimal control problem where the objective is to minimize the discrepancy between observed data and an approximate solution of the model, and the control variables are the fractional order and the truncation length. For the numerical solution of the minimization problem we propose a gradient-based approach, where we enhance the numerical performance by an approximation of the bilinear form of the state equation and its derivative with respect to the fractional order. Several numerical tests in one and two dimensions illustrate the theoretical results and show the robustness and applicability of our method.

97 MATHEMATICS AND COMPUTING↗

Design Status Update of the Electron-Ion Collider

The design of the electron-ion collider EIC to be constructed at Brookhaven National Laboratory has been continuously evolving towards a realistic and robust design that meets all the requirements set forth by the nuclear physics community in the White Paper. Over the past year activities have been focused on maturing the design, and on developing alternatives to mitigate risk. These include improvements of the interaction region design as well as modifications of the hadron ring vacuum system to accommodate the high average and peak beam currents. Beam dynamics studies have been performed to determine and optimize the dynamic aperture in the two collider rings and the beam-beam performance. We will present the EIC design with a focus on recent developments.

43 PARTICLE ACCELERATORS↗

A hybrid differential evolution/Levenberg-Marquardt method for solving inverse transport problems

Recently, the Differential Evolution (DE) optimization method was applied to solve inverse transport problems in finite cylindrical geometries and was shown to be far superior to the Levenberg-Marquardt optimization method at finding a global optimum for problems with several unknowns. However, while extremely adept at finding a global optimum solution, the DE method often requires a large number (hundreds or thousands) of transport calculations, making it much slower than the Levenberg-Marquardt method. In this paper, a hybridization of the Differential Evolution and Levenberg-Marquardt approaches is presented. This hybrid method takes advantage of the robust search capability of the Differential Evolution method and the speed of the Levenberg-Marquardt technique.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

Chemical solution deposition derived (001)-oriented epitaxial BiFeO{sub 3} thin films with robust ferroelectric properties using stoichiometric precursors (invited)

Phase pure bismuth ferrite (BiFeO{sub 3}) thin films with (001)-oriented epitaxial structure are realized on lanthanum strontium manganite (La{sub 0.67}Sr{sub 0.33}MnO{sub 3}) buffered (001)-SrTiO{sub 3} substrates by chemical solution deposition. The annealing process is optimized such that a stoichiometric precursor can be used to accurately control the Bi:Fe ratio. Ferroelectric, dielectric, and resistive switching behaviours are investigated for 40 nm, 70 nm, and 150 nm BFO thin films. While the thinnest film (40 nm) shows very leaky loops, square and fully saturated polarization hysteresis loops are shown for the thicker films. The highest remanent polarization (2P{sub r} = 100 μC/cm{sup 2}) and relative dielectric constant (ε{sub r} = 613) are obtained in the 150 nm BFO thin film. High cycle fatigue tests show that the thick films are resistant to polarization fatigue. Piezoresponse force microscopy results show that the domain structure varies with thickness. Resistive switching and polarization mediated diode effects are also observed. These robust properties suggest that chemical solution deposition derived BiFeO{sub 3} thin films can offer a viable low cost alternative.

36 MATERIALS SCIENCE↗

Robust direct laser acceleration of electrons with flying-focus laser pulses

Direct laser acceleration (DLA) offers a compact source of high-charge, energetic electrons for generating secondary radiation or neutrons. While DLA in high-density plasma optimizes the energy transfer from a laser pulse to electrons, it exacerbates nonlinear propagation effects, such as filamentation, that can disrupt the acceleration process. Here, we show that superluminal flying-focus pulses (FFPs) mitigate nonlinear propagation, thereby enhancing the number of high-energy electrons and resulting x-ray yield. Three-dimensional particle-in-cell simulations show that, compared to a Gaussian pulse of equal energy (1 J) and intensity (2 × 10 20 W/cm 2 ), an FFP produces 80 × more electrons above 100 MeV, increases the electron cutoff energy by 20%, triples the high-energy x-ray yield, and improves x-ray collimation. These results illustrate the ability of spatiotemporally structured laser pulses to provide additional control in the highly nonlinear, relativistic regime of laser-plasma interactions.

Laser-produced plasmas↗

Robust Algorithm for Computing Statistical Stark Broadening of Spectral Lines

A method previously developed to solve large-scale linear systems is applied to statistical Stark broadened line shape calculations. The method is formally exact, numerically stable, and allows optimization of the integration over the quasi-static field to assure numerical accuracy. Furthermore, the method does not increase the computational effort and often can decrease it compared to the conventional approach.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Generation of Intense Low-Divergence Isolated Soft-X-Ray Attosecond Pulses in a Gas-Filled Waveguide Using Three-Color Synthesized Laser Pulses

Isolated attosecond pulses (IAPs) in the soft-x-ray (SXR) region have been successfully produced via high-order harmonic generation in a gas medium at several laboratories, but pulse energies are much lower compared to the IAPs in the extreme ultraviolet. Here, we show that in a gas-filled hollow waveguide, an efficient gating method using a few-cycle three-color synthesizer is able to generate stronger SXR IAPs. We show that an 104-as IAP in the SXR region (central energy of 210 eV) with a higher pulse energy (0.9 nJ) and a lower divergence (within 1.5 mrad) can be generated with a three-color laser beam for input pulse energy at 0.62 mJ, under the condition of optimal gas pressure and waveguide length. The enhancement of the IAP is attributed to favorable phase matching of SXR high harmonics caused by the evolution of the three-color waveform during its propagation in the waveguide, which can be understood by analyzing the time shift of the waveform due to the interplay of the waveguide mode, atomic dispersion, and plasma dispersion. We check that the gating method is quite robust with respect to the phase jitter in the waveform. The advantage of this method is further demonstrated by comparing the generation of SXR high harmonics and attosecond pulses with two other schemes: the three-color waveform in a gas cell and a previously optimized two-color waveform in a gas-filled waveguide.

74 ATOMIC AND MOLECULAR PHYSICS↗

IMPROVEMENT OF THE RICHNESS ESTIMATES OF maxBCG CLUSTERS

Minimizing the scatter between cluster mass and accessible observables is an important goal for cluster cosmology. In this work, we introduce a new matched filter richness estimator, and test its performance using the maxBCG cluster catalog. Our new estimator significantly reduces the variance in the L{sub X} -richness relation, from sigma{sub lnLx}{sup 2} = (0.86+-0.02){sup 2} to sigma{sub lnLx}{sup 2} = (0.69+-0.02){sup 2}. Relative to the maxBCG richness estimate, it also removes the strong redshift dependence of the L{sub X} -richness scaling relations, and is significantly more robust to photometric and redshift errors. These improvements are largely due to the better treatment of galaxy color data. We also demonstrate the scatter in the L{sub X} -richness relation depends on the aperture used to estimate cluster richness, and introduce a novel approach for optimizing said aperture which can easily be generalized to other mass tracers.

79 ASTRONOMY AND ASTROPHYSICS↗

Robust, multidimensional mesh motion based on Monge-Kantorovich equidistribution

Mesh-motion (r-refinement) grid adaptivity schemes are attractive due to their potential to minimize the numerical error for a prescribed number of degrees of freedom. However, a key roadblock to a widespread deployment of this class of techniques has been the formulation of robust, reliable mesh-motion governing principles, which (1) guarantee a solution in multiple dimensions (2D and 3D), (2) avoid grid tangling (or folding of the mesh, whereby edges of a grid cell cross somewhere in the domain), and (3) can be solved effectively and efficiently. In this study, we formulate such a mesh-motion governing principle, based on volume equidistribution via Monge-Kantorovich optimization (MK). In earlier publications [1] and [2], the advantages of this approach with regard to these points have been demonstrated for the time-independent case. In this study, we demonstrate that Monge-Kantorovich equidistribution can in fact be used effectively in a time-stepping context, and delivers an elegant solution to the otherwise pervasive problem of grid tangling in mesh-motion approaches, without resorting to ad hoc time-dependent terms (as in moving-mesh PDEs, or MMPDEs [3] and [4]). We explore two distinct r-refinement implementations of MK: the direct method, where the current mesh relates to an initial, unchanging mesh, and the sequential method, where the current mesh is related to the previous one in time. We demonstrate that the direct approach is superior with regard to mesh distortion and robustness. The properties of the approach are illustrated with a hyperbolic PDE, the advection of a passive scalar, in 2D and 3D. Velocity flow fields with and without flow shear are considered. Three-dimensional grid, time-step, and nonlinear tolerance convergence studies are presented which demonstrate the optimality of the approach.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

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↗

Limits of multifunctionality in tunable networks

Nature is rife with networks that are functionally optimized to propagate inputs to perform specific tasks. Whether via genetic evolution or dynamic adaptation, many networks create functionality by locally tuning interactions between nodes. Here we explore this behavior in two contexts: strain propagation in mechanical networks and pressure redistribution in flow networks. By adding and removing links, we are able to optimize both types of networks to perform specific functions. We define a single function as a tuned response of a single “target” link when another, predetermined part of the network is activated. Using network structures generated via such optimization, we investigate how many simultaneous functions such networks can be programed to fulfill. We find that both flow and mechanical networks display qualitatively similar phase transitions in the number of targets that can be tuned, along with the same robust finite-size scaling behavior. We discuss how these properties can be understood in the context of constraint–satisfaction problems.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗

Effect of Dosimetric Outliers on the Performance of a Commercial Knowledge-Based Planning Solution

Purpose: RapidPlan, a commercial knowledge-based planning solution, uses a model library containing the geometry and associated dosimetry of existing plans. This model predicts achievable dosimetry for prospective patients that can be used to guide plan optimization. However, it is unknown how suboptimal model plans (outliers) influence the predictions or resulting plans. We investigated the effect of, first, removing outliers from the model (cleaning it) and subsequently adding deliberate dosimetric outliers. Methods and Materials: Clinical plans from 70 head and neck cancer patients comprised the uncleaned (UC) Model{sub UC}, from which outliers were cleaned (C) to create Model{sub C}. The last 5 to 40 patients of Model{sub C} were replanned with no attempt to spare the salivary glands. These substantial dosimetric outliers were reintroduced to the model in increments of 5, creating Model{sub 5} to Model{sub 40} (Model{sub 5-40}). These models were used to create plans for a 10-patient evaluation group. Plans from Model{sub UC} and Model{sub C}, and Model{sub C} and Model{sub 5-40} were compared on the basis of boost (B) and elective (E) target volume homogeneity indexes (HI{sub B}/HI{sub E}) and mean doses to oral cavity, composite salivary glands (comp{sub sal}) and swallowing (comp{sub swal}) structures. Results: On average, outlier removal (Model{sub C} vs Model{sub UC}) had minimal effects on HI{sub B}/HI{sub E} (0%-0.4%) and sparing of organs at risk (mean dose difference to oral cavity and comp{sub sal}/comp{sub swal} were ≤0.4 Gy). Model{sub 5-10} marginally improved comp{sub sal} sparing, whereas adding a larger number of outliers (Model{sub 20-40}) led to deteriorations in comp{sub sal} up to 3.9 Gy, on average. These increases are modest compared to the 14.9 Gy dose increases in the added outlier plans, due to the placement of optimization objectives below the inferior boundary of the dose-volume histogram-predicted range. Conclusions: Overall, dosimetric outlier removal from or addition of 5 to 10 outliers to a 70-patient model had marginal effects on resulting plan quality. Although the addition of >20 outliers deteriorated plan quality, the effect was modest. In this study, RapidPlan demonstrated robustness for moderate proportions of salivary gland dosimetric outliers.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Attribute-Guided Adversarial Training for Robustness to Natural Perturbations

We report while existing work in robust deep learning has focused on small pixel-level norm-based perturbations, this may not account for perturbations encountered in several real world settings. In many such cases although test data might not be available, broad specifications about the types of perturbations (such as an unknown degree of rotation) may be known. We consider a setup where robustness is expected over an unseen test domain that is not i.i.d. but deviates from the training domain. While this deviation may not be exactly known, its broad characterization is specified a priori, in terms of attributes. We propose an adversarial training approach which learns to generate new samples so as to maximize exposure of the classifier to the attributes-space, without having access to the data from the test domain. Our adversarial training solves a min-max optimization problem, with the inner maximization generating adversarial perturbations, and the outer minimization finding model parameters by optimizing the loss on adversarial perturbations generated from the inner maximization. We demonstrate the applicability of our approach on three types of naturally occurring perturbations --- object-related shifts, geometric transformations, and common image corruptions. Our approach enables deep neural networks to be robust against a wide range of naturally occurring perturbations. We demonstrate the usefulness of the proposed approach by showing the robustness gains of deep neural networks trained using our adversarial training on MNIST, CIFAR-10, and a new variant of the CLEVR dataset.

97 MATHEMATICS AND COMPUTING↗

Gradient algorithm applied to laboratory quantum control

The exploration of a quantum control landscape, which is the physical observable as a function of the control variables, is fundamental for understanding the ability to perform observable optimization in the laboratory. For high control variable dimensions, trajectory-based methods provide a means for performing such systematic explorations by exploiting the measured gradient of the observable with respect to the control variables. This paper presents a practical, robust, easily implemented statistical method for obtaining the gradient on a general quantum control landscape in the presence of noise. In order to demonstrate the method's utility, the experimentally measured gradient is utilized as input in steepest-ascent trajectories on the landscapes of three model quantum control problems: spectrally filtered and integrated second harmonic generation as well as excitation of atomic rubidium. The gradient algorithm achieves efficiency gains of up to approximately three times that of the standard genetic algorithm and, as such, is a promising tool for meeting quantum control optimization goals as well as landscape analyses. The landscape trajectories directed by the gradient should aid in the continued investigation and understanding of controlled quantum phenomena.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Parametrization of textural patterns in {sup 123}I-ioflupane imaging for the automatic detection of Parkinsonism

Purpose: A novel approach to a computer aided diagnosis system for the Parkinson's disease is proposed. This tool is intended as a supporting tool for physicians, based on fully automated methods that lead to the classification of{sup 123}I-ioflupane SPECT images. Methods: {sup 123}I-ioflupane images from three different databases are used to train the system. The images are intensity and spatially normalized, then subimages are extracted and a 3D gray-level co-occurrence matrix is computed over these subimages, allowing the characterization of the texture using Haralick texture features. Finally, different discrimination estimation methods are used to select a feature vector that can be used to train and test the classifier. Results: Using the leave-one-out cross-validation technique over these three databases, the system achieves results up to a 97.4% of accuracy, and 99.1% of sensitivity, with positive likelihood ratios over 27. Conclusions: The system presents a robust feature extraction method that helps physicians in the diagnosis task by providing objective, operator-independent textural information about{sup 123}I-ioflupane images, commonly used in the diagnosis of the Parkinson's disease. Textural features computation has been optimized by using a subimage selection algorithm, and the discrimination estimation methods used here makes the system feature-independent, allowing us to extend it to other databases and diseases.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Identifying attributes of CO 2 leakage zones in shallow aquifers using a parametric level set method: Original Research Article: Identifying attributes of CO 2 leakage zones in shallow aquifers using a parametric level set method

Leakage through abandoned wells and geologic faults poses the greatest risk to CO 2 storage permanence. For shallow aquifers, secondary CO 2 plumes emanating from the leak zones may go undetected for a sustained period of time and has the greatest potential to cause large-scale and long-term environmental impacts. Identification of the attributes of leak zones, including their shape, location, and strength, is required for environmental risk assessment. This study applies a parametric level set (PaLS) method to characterize leak zones. Level set methods are appealing for tracking topological changes and recovering unknown shapes of objects. However, level set evolution using the conventional level set methods is computationally challenging. In PaLS, the level set function is approximated by using a weighted sum of basis functions and the level set evolution problem is replaced by an optimization problem. The efficacy of PaLS is demonstrated by reconstructing attributes of a CO 2 leak zone in a carbonate aquifer. Our results show that PaLS is a robust source identification method that can recover the approximate leak zone locations in the presence of measurement errors, model parameter uncertainty, and inaccurate guesses of source flux strengths. Accurate delineation of source geometry is achievable when a relatively dense observation network is used. The PaLS-based, source recovery framework introduced in this work is generic and can be adapted for any reactive transport model by simply switching the pre- and post-processing routines.

54 ENVIRONMENTAL SCIENCES↗

Quantification and Visualization of Charged Particle Range Variations

Purpose: Range variations during respiration affect the penetration of charged particle beams and can result in beam overshoot or undershoot to the target. We have developed analysis tools to quantify the water equivalent pathlength (WEL) variations resulting from respiration (WEL analyzer, Aqualyzer), as well as a data explorer to view WEL variations interactively. Methods and Materials: The metrics to characterize and quantify penetration of a charged particle beam during respiration were calculated semiautomatically. The analysis involved the generation of images that (1) encode the radiologic pathlength across a beam's eye view image during the respiratory phase, (2) display the variation of the radiologic pathlength relative to a reference respiratory phase, (3) display isopenetration as a function of breathing, and (4) show range fluctuation images for a compensating bolus when applied to four-dimensional computed tomography. Additional quantities relevant to the analysis of charged particle beams in a breathing patient that are calculated include the beam overshoot volume and beam overshoot distance. These quantities are calculated as a function of time, gantry angle, and position. Results: The software was applied to test cases to illustrate its utility in the analysis of range variations of charged particle beams in the treatment of lung tumors. Conclusion: WEL analysis is useful in the rapid assessment of range variations in the treatment of lung tumors and in determining the optimal gantry angle and respiratory gating window. The extension of encoding range fluctuations to a beam's eye view display is helpful in designing plans that are more robust in the presence of motion.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

Optimization-based mesh correction with volume and convexity constraints

Here, we consider the problem of finding a mesh such that 1) it is the closest, with respect to a suitable metric, to a given source mesh having the same connectivity, and 2) the volumes of its cells match a set of prescribed positive values that are not necessarily equal to the cell volumes in the source mesh. Also, this volume correction problem arises in important simulation contexts, such as satisfying a discrete geometric conservation law and solving transport equations by incremental remapping or similar semi-Lagrangian transport schemes. In this paper we formulate volume correction as a constrained optimization problem in which the distance to the source mesh defines an optimization objective, while the prescribed cell volumes, mesh validity and/or cell convexity specify the constraints. We solve this problem numerically using a sequential quadratic programming (SQP) method whose performance scales with the mesh size. To achieve scalable performance we develop a specialized multigrid-based preconditioner for optimality systems that arise in the application of the SQP method to the volume correction problem. Numerical examples illustrate the importance of volume correction, and showcase the accuracy, robustness and scalability of our approach.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗