Search NASA⌕ Search

SEARCH · Search NASA

Results for “randomization”

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 487 records · Page 27

Building Ultra-Low False Alarm Rate Support Vector Classifier Ensembles Using Random Subspaces

This paper presents the Cost-Sensitive Random Subspace Support Vector Classifier (CS-RS-SVC), a new learning algorithm that combines random subspace sampling and bagging with Cost-Sensitive Support Vector Classifiers to more effectively address detection applications burdened by unequal misclassification requirements. When compared to its conventional, non-cost-sensitive counterpart on a two-class signal detection application, random subspace sampling is shown to very effectively leverage the additional flexibility offered by the Cost-Sensitive Support Vector Classifier, yielding a more than four-fold increase in the detection rate at a false alarm rate (FAR) of zero. Moreover, the CS-RS-SVC is shown to be fairly robust to constraints on the feature subspace dimensionality, enabling reductions in computation time of up to 82% with minimal performance degradation.

42 ENGINEERING↗

Fabrication of Quasi-Random Photonic Crystals for Ultrathin Solar Cells

Quasi-random structures for light management present an intermediate ordering between a completely random rough surface and a perfect ordered photonic crystal. They are attractive for broadband applications due to their power spectral density. This makes them excellent candidates for photovoltaic applications. Here we present preliminary fabrication results creating quasi-random structures using a self-assembly approach based on polymer blends and wet etching. We are able to create patterns with average lattice constant of five to two microns. This structure is proposed to be used as a rear patterning layer for ultrathin GaAs cells for better sustainability and radiation- hard devices.

41 EE - Solar Energy Technologies Office (EE-4S)↗

Quickest Detection for Changes in Maximal kNN Coherence of Random Matrices

This paper addresses the problem of quickest detection of a change in the maximal coherence between columns of a n p random matrix based on a sequence of matrix observations having a single unknown change point. The random matrix is assumed to have identically distributed rows and the maximal coherence is defined as the largest of the ( 2 p) correlation coefficients associated with any row. Likewise the κ nearest neighbor (kNN) coherence is defined as the κ-th largest of these correlation coefficients. The forms of the preand postchange distributions of the observed matrices are assumed to belong to the family of elliptically contoured densities with sparse dispersion matrices but are otherwise unknown. A nonparametric stopping rule is proposed that is based on the maximal kNN sample coherence between columns of each observed random matrix. This is a summary statistic that is related to a test of existence of a hub vertex in a sample correlation graph having degree at least κ. Performance bounds on the delay and false alarm performance of the proposed stopping rule are obtained in the purely high-dimensional regime where p → ∞ and n is fixed. When the prechange dispersion matrix is diagonal it is shown that, among all functions of the proposed summary statistic, the proposed stopping rule is asymptotically optimal under a minimax quickest change detection model as the stopping threshold approaches infinity. Furthermore, the theory developed also applies to sequential hypothesis testing and fixed sample size tests.

96 KNOWLEDGE MANAGEMENT AND PRESERVATION↗

Monte Carlo investigation of the increased radiation deposition due to gold nanoparticles using kilovoltage and megavoltage photons in a 3D randomized cell model

Purpose: Investigation of increased radiation dose deposition due to gold nanoparticles (GNPs) using a 3D computational cell model during x-ray radiotherapy.Methods: Two GNP simulation scenarios were set up in Geant4; a single 400 nm diameter gold cluster randomly positioned in the cytoplasm and a 300 nm gold layer around the nucleus of the cell. Using an 80 kVp photon beam, the effect of GNP on the dose deposition in five modeled regions of the cell including cytoplasm, membrane, and nucleus was simulated. Two Geant4 physics lists were tested: the default Livermore and custom built Livermore/DNA hybrid physics list. 10{sup 6} particles were simulated at 840 cells in the simulation. Each cell was randomly placed with random orientation and a diameter varying between 9 and 13 {mu}m. A mathematical algorithm was used to ensure that none of the 840 cells overlapped. The energy dependence of the GNP physical dose enhancement effect was calculated by simulating the dose deposition in the cells with two energy spectra of 80 kVp and 6 MV. The contribution from Auger electrons was investigated by comparing the two GNP simulation scenarios while activating and deactivating atomic de-excitation processes in Geant4.Results: The physical dose enhancement ratio (DER) of GNP was calculated using the Monte Carlo model. The model has demonstrated that the DER depends on the amount of gold and the position of the gold cluster within the cell. Individual cell regions experienced statistically significant (p < 0.05) change in absorbed dose (DER between 1 and 10) depending on the type of gold geometry used. The DER resulting from gold clusters attached to the cell nucleus had the more significant effect of the two cases (DER {approx} 55). The DER value calculated at 6 MV was shown to be at least an order of magnitude smaller than the DER values calculated for the 80 kVp spectrum. Based on simulations, when 80 kVp photons are used, Auger electrons have a statistically insignificant (p < 0.05) effect on the overall dose increase in the cell. The low energy of the Auger electrons produced prevents them from propagating more than 250-500 nm from the gold cluster and, therefore, has a negligible effect on the overall dose increase due to GNP.Conclusions: The results presented in the current work show that the primary dose enhancement is due to the production of additional photoelectrons.

61 RADIATION PROTECTION AND DOSIMETRY↗

Investigating the limits of PET/CT imaging at very low true count rates and high random fractions in ion-beam therapy monitoring

Purpose: External beam radiotherapy with protons and heavier ions enables a tighter conformation of the applied dose to arbitrarily shaped tumor volumes with respect to photons, but is more sensitive to uncertainties in the radiotherapeutic treatment chain. Consequently, an independent verification of the applied treatment is highly desirable. For this purpose, the irradiation-induced β{sup +}-emitter distribution within the patient is detected shortly after irradiation by a commercial full-ring positron emission tomography/x-ray computed tomography (PET/CT) scanner installed next to the treatment rooms at the Heidelberg Ion-Beam Therapy Center (HIT). A major challenge to this approach is posed by the small number of detected coincidences. This contribution aims at characterizing the performance of the used PET/CT device and identifying the best-performing reconstruction algorithm under the particular statistical conditions of PET-based treatment monitoring. Moreover, this study addresses the impact of radiation background from the intrinsically radioactive lutetium-oxyorthosilicate (LSO)-based detectors at low counts. Methods: The authors have acquired 30 subsequent PET scans of a cylindrical phantom emulating a patientlike activity pattern and spanning the entire patient counting regime in terms of true coincidences and random fractions (RFs). Accuracy and precision of activity quantification, image noise, and geometrical fidelity of the scanner have been investigated for various reconstruction algorithms and settings in order to identify a practical, well-suited reconstruction scheme for PET-based treatment verification. Truncated listmode data have been utilized for separating the effects of small true count numbers and high RFs on the reconstructed images. A corresponding simulation study enabled extending the results to an even wider range of counting statistics and to additionally investigate the impact of scatter coincidences. Eventually, the recommended reconstruction scheme has been applied to exemplary postirradiation patient data-sets. Results: Among the investigated reconstruction options, the overall best results in terms of image noise, activity quantification, and accurate geometrical recovery were achieved using the ordered subset expectation maximization reconstruction algorithm with time-of-flight (TOF) and point-spread function (PSF) information. For this algorithm, reasonably accurate (better than 5%) and precise (uncertainty of the mean activity below 10%) imaging can be provided down to 80 000 true coincidences at 96% RF. Image noise and geometrical fidelity are generally improved for fewer iterations. The main limitation for PET-based treatment monitoring has been identified in the small number of true coincidences, rather than the high intrinsic random background. Application of the optimized reconstruction scheme to patient data-sets results in a 25% − 50% reduced image noise at a comparable activity quantification accuracy and an improved geometrical performance with respect to the formerly used reconstruction scheme at HIT, adopted from nuclear medicine applications. Conclusions: Under the poor statistical conditions in PET-based treatment monitoring, improved results can be achieved by considering PSF and TOF information during image reconstruction and by applying less iterations than in conventional nuclear medicine imaging. Geometrical fidelity and image noise are mainly limited by the low number of true coincidences, not the high LSO-related random background. The retrieved results might also impact other emerging PET applications at low counting statistics.

60 APPLIED LIFE SCIENCES↗

Automated segmentation of dental CBCT image with prior-guided sequential random forests

Purpose: Cone-beam computed tomography (CBCT) is an increasingly utilized imaging modality for the diagnosis and treatment planning of the patients with craniomaxillofacial (CMF) deformities. Accurate segmentation of CBCT image is an essential step to generate 3D models for the diagnosis and treatment planning of the patients with CMF deformities. However, due to the image artifacts caused by beam hardening, imaging noise, inhomogeneity, truncation, and maximal intercuspation, it is difficult to segment the CBCT. Methods: In this paper, the authors present a new automatic segmentation method to address these problems. Specifically, the authors first employ a majority voting method to estimate the initial segmentation probability maps of both mandible and maxilla based on multiple aligned expert-segmented CBCT images. These probability maps provide an important prior guidance for CBCT segmentation. The authors then extract both the appearance features from CBCTs and the context features from the initial probability maps to train the first-layer of random forest classifier that can select discriminative features for segmentation. Based on the first-layer of trained classifier, the probability maps are updated, which will be employed to further train the next layer of random forest classifier. By iteratively training the subsequent random forest classifier using both the original CBCT features and the updated segmentation probability maps, a sequence of classifiers can be derived for accurate segmentation of CBCT images. Results: Segmentation results on CBCTs of 30 subjects were both quantitatively and qualitatively validated based on manually labeled ground truth. The average Dice ratios of mandible and maxilla by the authors’ method were 0.94 and 0.91, respectively, which are significantly better than the state-of-the-art method based on sparse representation (p-value < 0.001). Conclusions: The authors have developed and validated a novel fully automated method for CBCT segmentation.

61 RADIATION PROTECTION AND DOSIMETRY↗

Rapid quantification of mutant fitness in diverse bacteria by sequencing randomly bar-coded transposons

Transposon mutagenesis with next-generation sequencing (TnSeq) is a powerful approach to annotate gene function in bacteria, but existing protocols for TnSeq require laborious preparation of every sample before sequencing. Thus, the existing protocols are not amenable to the throughput necessary to identify phenotypes and functions for the majority of genes in diverse bacteria. Here, we present a method, random bar code transposon-site sequencing (RB-TnSeq), which increases the throughput of mutant fitness profiling by incorporating random DNA bar codes into Tn5 and mariner transposons and by using bar code sequencing (BarSeq) to assay mutant fitness. RB-TnSeq can be used with any transposon, and TnSeq is performed once per organism instead of once per sample. Each BarSeq assay requires only a simple PCR, and 48 to 96 samples can be sequenced on one lane of an Illumina HiSeq system. We demonstrate the reproducibility and biological significance of RB-TnSeq with Escherichia coli , Phaeobacter inhibens , Pseudomonas stutzeri , Shewanella amazonensis , and Shewanella oneidensis . To demonstrate the increased throughput of RB-TnSeq, we performed 387 successful genome-wide mutant fitness assays representing 130 different bacterium-carbon source combinations and identified 5,196 genes with significant phenotypes across the five bacteria. In P. inhibens , we used our mutant fitness data to identify genes important for the utilization of diverse carbon substrates, including a putative D-mannose isomerase that is required for mannitol catabolism. RB-TnSeq will enable the cost-effective functional annotation of diverse bacteria using mutant fitness profiling. A large challenge in microbiology is the functional assessment of the millions of uncharacterized genes identified by genome sequencing. Transposon mutagenesis coupled to next-generation sequencing (TnSeq) is a powerful approach to assign phenotypes and functions to genes. However, the current strategies for TnSeq are too laborious to be applied to hundreds of experimental conditions across multiple bacteria. Here, we describe an approach, random bar code transposon-site sequencing (RB-TnSeq), which greatly simplifies the measurement of gene fitness by using bar code sequencing (BarSeq) to monitor the abundance of mutants. We performed 387 genome-wide fitness assays across five bacteria and identified phenotypes for over 5,000 genes. RB-TnSeq can be applied to diverse bacteria and is a powerful tool to annotate uncharacterized genes using phenotype data.

59 BASIC BIOLOGICAL SCIENCES↗

River sinuosity describes a continuum between randomness and ordered growth

River channels are among the most common landscape features on Earth. An essential characteristic of channels is sinuosity: their tendency to take a circuitous path, which is quantified as along-stream length divided by straight-line length. River sinuosity is interpreted as a characteristic that either forms randomly at channel inception or develops over time as meander bends migrate. Studies tend to assume the latter and thus have used river sinuosity as a proxy for both modern and ancient environmental factors including climate, tectonics, vegetation, and geologic structure. But no quantitative criterion for planform expression has distinguished between random, initial sinuosity and that developed by ordered growth through channel migration. This ambiguity calls into question the utility of river sinuosity for understanding Earth’s history. We propose a quantitative framework to reconcile these competing explanations for river sinuosity. Using a coupled analysis of modeled and natural channels, we show that while a majority of observed sinuosity is consistent with randomness and limited channel migration, rivers with sinuosity ≥1.5 likely formed their geometry through sustained, ordered growth due to channel migration. This criterion frames a null hypothesis for river sinuosity that can be applied to evaluate the significance of environmental interpretations in landscapes shaped by rivers. The quantitative link between sinuosity and channel migration further informs strategies for preservation and restoration of riparian habitat and guides predictions of fluvial deposits in the rock record and in remotely sensed environments from the seafloor to planetary surfaces.

58 GEOSCIENCES↗

Color Randomization of Fast Gluon-Gluon Pairs in the Quark-Gluon Plasma

We study the color randomization of two-gluon states produced after splitting of a primary fast gluon in the quark-gluon plasma. We find that for the LHC conditions the color randomization of the gg pairs is rather slow. At jet energies E = 100 and 500 GeV, for typical jet path length in the plasma in central Pb+Pb collisions, the SU(3)-multiplet averaged color Casimir operator of the gg pair differs considerably from its value 2N{sub c} for a fully randomized gg state. Our calculations of the energy dependence for generation of the nearly collinear decuplet gg states, that can lead to the baryon jet fragmentation, show that the contribution of the anomalous decuplet color states to the baryon production should become small at p{sub T} ≳ 10 GeV.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Efficient broadcast on random geometric graphs

A Randon Geometric Graph (RGG) is constructed by distributing n nodes uniformly at random in the unit square and connecting two nodes if their Euclidean distance is at most r, for some prescribed r. They analyze the following randomized broadcast algorithm on RGGs. At the beginning, there is only one informed node. Then in each round, each informed node chooses a neighbor uniformly at random and informs it. They prove that this algorithm informs every node in the largest component of a RGG in {Omicron}({radical}n/r) rounds with high probability. This holds for any value of r larger than the critical value for the emergence of a giant component. In particular, the result implies that the diameter of the giant component is {Theta}({radical}n/r).

42 ENGINEERING↗

Stochastic basis adaptation and spatial domain decomposition for partial differential equations with random coefficients

We present a novel approach for high-dimensional stochastic partial differential equations that decomposes the computational domain into smaller disjoint subdomains and reduces the random dimension in each subdomain to save computational costs. The proposed approach combines spatial domain decomposition and stochastic basis adaptation in each subdomain. We adapt the stochastic basis in each subdomain such that the local solution has a low-dimensional random space representation. To compute the stochastic solution in each subdomain, we use the Neumann-Neumann algorithm, where we first estimate the interface solution and then evaluate the interior solution in each subdomain using the interface solution as a boundary condition. The interior solutions in each subdomain are computed independently of each other, which reduces the operation count from $\approx O(N^\alpha)$ to $\approx O(M^\alpha),$ where $N$ is the total number of degrees of freedom, $M$ is the number of degrees of freedom in each subdomain, and the exponent $\alpha>1$ depends on the uncertainty quantification method used. In addition, the localized nature of solutions makes the proposed approach highly parallelizable. We illustrate the accuracy and efficiency of the approach for a steady-state diffusion equation with a random space-dependent diffusion coefficient.

uncertainty quantification, basis adaptation↗

Surrogate-Based Autotuning for Randomized Sketching Algorithms in Regression Problems

Algorithms from Randomized Numerical Linear Algebra (RandNLA) are known to be effective in handling high-dimensional computational problems, providing high-quality empirical performance as well as strong probabilistic guarantees. However, their practical application is complicated by the fact that the user needs to set various algorithm-specific tuning parameters which are different from those used in traditional NLA. This paper demonstrates how a surrogate-based autotuning approach can be used to address fundamental problems of parameter selection in RandNLA algorithms. In particular, we provide a detailed investigation of surrogate-based autotuning for sketch-and-precondition (SAP)-based randomized least squares methods, which have been one of the great success stories in modern RandNLA. Empirical results show that our surrogate-based autotuning approach can achieve near-optimal performance with much less tuning cost than a random search (up to about 7.6x fewer trials of different parameter configurations). Moreover, while our experiments focus on least squares, our results demonstrate a general-purpose autotuning pipeline applicable to any kind of RandNLA algorithm.

Cho, Younghyun↗

Efficient Training of Deep Neural Operator Networks via Randomized Sampling

Neural operators (NOs) employ deep neural networks to learn the mappings between infinitedimensional function spaces. Deep operator network (DeepONet), a popular NO architecture, has demonstrated success in the real-time prediction of complex dynamics across various scientific and engineering applications. In this work, we introduce a random sampling technique to be adopted during the training of DeepONet, aimed at improving the generalization ability of the model, while significantly reducing the computational time. The proposed approach targets the trunk network of the DeepONet model that outputs the basis functions corresponding to the spatiotemporal locations of the bounded domain on which the physical system is defined. While constructing the loss function, DeepONet training traditionally considers a uniform grid of spatiotemporal points at which all the output functions are evaluated for each iteration. This approach leads to a larger batch size, resulting in poor generalization and increased memory demands, due to the limitations of the stochastic gradient descent (SGD) optimizer. The proposed random sampling over the inputs of the trunk net mitigates these challenges, improving generalization and reducing the memory requirements during training, resulting in significant computational gains. We validate our hypothesis through three benchmark examples, demonstrating substantial reductions in training time while achieving comparable or lower overall test errors relative to the traditional training approach. Our results indicate that incorporating randomization in the trunk network inputs during training enhances the efficiency and robustness of DeepONet, offering a promising avenue for improving the framework’s performance in modeling complex physical systems.

Karumuri, Sharmila [Department of Civil & Systems ↗

COBRA:COMPUTED-TOMOGRAPHY BASED RANDOM-FIELD APPROXIMATION

SF-25-115 COBRA (COmputed-tomography Based Random-field Approximation) is a Python application for generating statistically equivalent random fields from CT-scan imagery. It leverages Karhunen–Loève expansions to model microstructural variability, enabling users to: Preprocess CT scans (filtering and Gaussian transformation); Fit covariance kernels fromempirical data; Solve eigenproblems to obtain KL modes; Sample random fields onsistent with fitted statistics; Postprocess samples back into the physical domain.

Hu, Tianchen↗

Comparing Algorithms for Graph Isomorphism Using Discrete- and Continuous-Time Quantum Random Walks

Berry and Wang [Phys. Rev. A 83, 042317 (2011)] show numerically that a discrete-time quan- tum random walk of two noninteracting particles is able to distinguish some non-isomorphic strongly regular graphs from the same family. Here we analytically demonstrate how it is possible for these walks to distinguish such graphs, while continuous-time quantum walks of two noninteracting parti- cles cannot. We show analytically and numerically that even single-particle discrete-time quantum random walks can distinguish some strongly regular graphs, though not as many as two-particle noninteracting discrete-time walks. Additionally, we demonstrate how, given the same quantum random walk, subtle di erences in the graph certi cate construction algorithm can nontrivially im- pact the walk's distinguishing power. We also show that no continuous-time walk of a xed number of particles can distinguish all strongly regular graphs when used in conjunction with any of the graph certi cates we consider. We extend this constraint to discrete-time walks of xed numbers of noninteracting particles for one kind of graph certi cate; it remains an open question as to whether or not this constraint applies to the other graph certi cates we consider.

97 MATHEMATICS AND COMPUTING↗

Traceable Device-Independent Randomness as a Service

We implement a traceable protocol for device-independent random number generation. Our demonstration involves transparently integrating a device-independent RNG into a public randomness beacon, and feeding 7434 random number pulses into the beacon over 39 days.

Kavuri, Gautam A.↗

Optical filters made from random metasurfaces using Bayesian optimization

Abstract We theoretically investigate the ability to design optical filters from a single material and a single layer of randomly dispersed resonant dielectric particles, defining a random metasurface. Using a Bayesian and generalized Mie inverse-design approach, we design particle radii distributions that give rise to longpass, shortpass, bandpass, and bandstop spectral bands in the infrared. The optical response is shown to be directly related to electric and magnetic multipole scattering of the constituent particles and their near field coupling. We discuss the effect of the particle size distribution and particle–particle coupling interactions on filter design in random systems lacking long-range order.

Materials Science↗

Development of a Random Field Model for Gas Plume Detection in Multiple LWIR Images.

This report develops a random field model that describes gas plumes in LWIR remote sensing images. The random field model serves as a prior distribution that can be combined with LWIR data to produce a posterior that determines the probability that a gas plume exists in the scene and also maps the most probable location of any plume. The random field model is intended to work with a single pixel regression estimator--a regression model that estimates gas concentration on an individual pixel basis.

Heasler, Patrick G.↗