Search NASA⌕ Search

SEARCH · Search NASA

Results for “reduced set methods”

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

DP-TwoLevel: two-stage gradient subspace learning for differentially private federated learning

Federated learning (FL) enables collaborative model training across distributed data sources without sharing raw data, but faces fundamental challenges in communication efficiency and privacy. Differentially private (DP) training mitigates information leakage but introduces noise that degrades model performance, especially in high-dimensional settings. We propose DP-TwoLevel, a hierarchical gradient projection method that improves utility under fixed DP constraints by exploiting low-dimensional structure in model updates. Our approach learns a two-level PCA-based representation of gradients and applies DP noise in a reduced-dimensional subspace, thereby lowering the effective noise magnitude while preserving dominant signal components. We evaluate the method across three datasets (MNIST, Fashion-MNIST, CIFAR-10) and three privacy regimes (ϵ∈0.5, 1.0, 2.0). Across nine experimental settings, DP-TwoLevel consistently outperforms DP-FedAvg, achieving an average accuracy improvement of 9.44%, with larger gains observed in lower ϵ(higher-noise) regimes (up to +22.31%). We further analyze scalability across models ranging from 100K to 1.49M parameters and identify a variance-based success criterion: performance remains strong when the projection preserves more than 75% of gradient variance, degrades in a marginal regime (65–75%), and fails below this threshold. Our results demonstrate that structure-aware dimensionality reduction can significantly improve the privacy–utility tradeoff in FL without modifying formal privacy guarantees. We also provide empirical evidence of scaling limitations for global projections and motivate per-layer extensions for larger models.

Kotevska, Olivera [ORNL] (ORCID:0000000316772243)↗

On the statistics of biased tracers in the Effective Field Theory of Large Scale Structures

With the completion of the Planck mission, in order to continue to gather cosmological information it has become crucial to understand the Large Scale Structures (LSS) of the universe to percent accuracy. The Effective Field Theory of LSS (EFTofLSS) is a novel theoretical framework that aims to develop an analytic understanding of LSS at long distances, where inhomogeneities are small. We further develop the description of biased tracers in the EFTofLSS to account for the effect of baryonic physics and primordial non-Gaussianities, finding that new bias coefficients are required. Then, restricting to dark matter with Gaussian initial conditions, we describe the prediction of the EFTofLSS for the one-loop halo-halo and halo-matter two-point functions, and for the tree-level halo-halo-halo, matter-halo-halo and matter-matter-halo three-point functions. Several new bias coefficients are needed in the EFTofLSS, even though their contribution at a given order can be degenerate and the same parameters contribute to multiple observables. We develop a method to reduce the number of biases to an irreducible basis, and find that, at the order at which we work, seven bias parameters are enough to describe this extremely rich set of statistics. We then compare with the output of an N-body simulation where the normalization parameter of the linear power spectrum is set to σ 8 = 0.9. For the lowest mass bin, we find percent level agreement up to k ≃ 0.3 h Mpc –1 for the one-loop two-point functions, and up to k ≃ 0.15 h Mpc –1 for the tree-level three-point functions, with the k-reach decreasing with higher mass bins. In conclusion, this is consistent with the theoretical estimates, and suggests that the cosmological information in LSS amenable to analytical control is much more than previously believed.

79 ASTRONOMY AND ASTROPHYSICS↗

A promising limited angular computed tomography reconstruction via segmentation based regional enhancement and total variation minimization

X-ray computed tomography (CT) is a powerful and common inspection technique used for the industrial non-destructive testing. However, large-sized and heavily absorbing objects cause the formation of artifacts because of either the lack of specimen penetration in specific directions or the acquisition of data from only a limited angular range of views. Although the sparse optimization-based methods, such as the total variation (TV) minimization method, can suppress artifacts to some extent, reconstructing the images such that they converge to accurate values remains difficult because of the deficiency in continuous angular data and inconsistency in the projections. To address this problem, we use the idea of regional enhancement of the true values and suppression of the illusory artifacts outside the region to develop an efficient iterative algorithm. This algorithm is based on the combination of regional enhancement of the true values and TV minimization for the limited angular reconstruction. In this algorithm, the segmentation approach is introduced to distinguish the regions of different image knowledge and generate the support mask of the image. A new regularization term, which contains the support knowledge to enhance the true values of the image, is incorporated into the objective function. Then, the proposed optimization model is solved by variable splitting and the alternating direction method efficiently. A compensation approach is also designed to extract useful information from the initial projections and thus reduce false segmentation result and correct the segmentation support and the segmented image. The results obtained from comparing both simulation studies and real CT data set reconstructions indicate that the proposed algorithm generates a more accurate image than do the other reconstruction methods. The experimental results show that this algorithm can produce high-quality reconstructed images for the limited angular reconstruction and suppress the illusory artifacts caused by the deficiency in valid data.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Interpolative Separable Density Fitting Decomposition for Accelerating Hybrid Density Functional Calculations with Applications to Defects in Silicon

We present a new efficient way to perform hybrid density functional theory (DFT)-based electronic structure calculations. The new method uses an interpolative separable density fitting (ISDF) procedure to construct a set of numerical auxiliary basis vectors and a compact approximation of the matrix consisting of products of occupied orbitals represented in a large basis set such as the planewave basis. Such an approximation allows us to reduce the number of Poisson solves from O ($N$ $^{2}_{ϵ}$) to O ( N ϵ ) when we apply the exchange operator to occupied orbitals in an iterative method for solving the Kohn–Sham equations, where Ne is the number of electrons in the system to be studied. We show that the ISDF procedure can be carried out in O ($N$ $^{3}_{ϵ}$) operations, with a much smaller preconstant compared to methods used in existing approaches. When combined with the recently developed adaptively compressed exchange (ACE) operator formalism, which reduces the number of times the exchange operator needs to be updated, the resulting ACE-ISDF method significantly reduces the computational cost associated with the exchange operator by nearly 2 orders of magnitude compared to existing approaches for a large silicon system with 1000 atoms. We demonstrate that the ACE-ISDF method can produce accurate energies and forces for insulating and metallic systems and that it is possible to obtain converged hybrid functional calculation results for a 1000-atom bulk silicon within 10 min on 2000 computational cores. We also show that ACE-ISDF can scale to 8192 computational cores for a 4096-atom bulk silicon system. We use the ACE-ISDF method to geometrically optimize a 1000-atom silicon system with a vacancy defect using the HSE06 functional and computes its electronic structure. Here, we find that that the computed energy gap from the HSE06 functional is much closer to the experimental value compared to that produced by semilocal functionals in the DFT calculations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Parallelized POD-based suboptimal economic model predictive control of a state-constrained Boussinesq approximation

Motivated by an energy efficient building application, we want to optimize a quadratic cost functional subject to the Boussinesq approximation of the Navier-Stokes equations and to bilateral state and control constraints. Since the computation of such an optimal solution is numerically costly, we design an efficient strategy to compute a sub-optimal (but applicationally acceptable) solution with significantly reduced computational effort. We employ an economic Model Predictive Control (MPC) strategy to obtain a feedback control. The MPC sub-problems are based on a linear-quadratic optimal control problem subjected to mixed control and state constraints and a convection-diffusion equation, reduced with proper orthogonal decomposition. Finally, to solve each sub-problem, we apply a primal-dual active set strategy. The method can be fully parallelized, which enables the solution of large problems with real-world parameters.

97 MATHEMATICS AND COMPUTING↗

Stochastic Microgrid Scheduling With Chance‐Constrained Resilience Consideration

Traditionally, it is assumed that microgrids transition seamlessly from grid‐connected operation to islanded mode in the event of sudden main grid outages. In reality, the islanding process, especially unintentional islanding, is rarely seamless. Instead, it is subject to voltage and frequency fluctuations caused by the instantaneous disconnection of the point of common coupling (PCC) switch, variations in loads and renewable generation output and even the protection tripping of distributed energy resources (DERs). To mitigate these fluctuations and facilitate a smooth islanding process, we propose a stochastic microgrid scheduling model that incorporates chance‐constrained resilience measures. Specifically, the resilience measure is defined as the probability of successful islanding (PSI), that is, the probability that a microgrid can mitigate the generation‐demand imbalance caused by the disconnection of the PCC switch, variations in load and renewable generation and DER tripping. This measure is modelled using chance constraints. Unlike existing reliability and resilience indices, which typically neglect the possibility of microgrid/DER failure under extreme events and assume their survival while primarily focussing on reducing impact duration or magnitude, the proposed PSI‐based framework explicitly addresses microgrid and DER survival during the islanding transition. The formulated nonlinear chance constraints are approximated using a multiinterval approach and equivalently represented as a mixed‐integer linear programming (MILP) formulation. Case study results validate the proposed method, showing that the PSI estimation error is reduced to less than 8%, compared to approximately 28% with existing methods. Various sensitivity analyses on the DER tripping rate and PSI settings were performed to validate the robustness of the proposed method. In particular, the necessity of accounting for DER tripping in the PSI calculation was demonstrated.

chance constrained optimization↗

Fast uncertainty estimates in deep learning interatomic potentials

Deep learning has emerged as a promising paradigm to give access to highly accurate predictions of molecular and material properties. A common short-coming shared by current approaches, however, is that neural networks only give point estimates of their predictions and do not come with predictive uncertainties associated with these estimates. Existing uncertainty quantification efforts have primarily leveraged the standard deviation of predictions across an ensemble of independently trained neural networks. This incurs a large computational overhead in both training and prediction, resulting in order-of-magnitude more expensive predictions. Here, we propose a method to estimate the predictive uncertainty based on a single neural network without the need for an ensemble. This allows us to obtain uncertainty estimates with virtually no additional computational overhead over standard training and inference. We demonstrate that the quality of the uncertainty estimates matches those obtained from deep ensembles. We further examine the uncertainty estimates of our methods and deep ensembles across the configuration space of our test system and compare the uncertainties to the potential energy surface. Finally, we study the efficacy of the method in an active learning setting and find the results to match an ensemble-based strategy at order-of-magnitude reduced computational cost.

Chemistry↗

Inference of the optimal pretarget electron beam parameters in a Monte Carlo virtual linac model through simulated annealing

The purpose of this study was to develop an efficient method to determine the optimal intensity distribution of the pretarget electron beam in a Monte Carlo (MC) accelerator model able to most accurately reproduce a set of measured photon field profiles for a given accelerator geometry and nominal photon beam energy. The method has the ability to reduce the number of simulations required to commission a MC accelerator model and has achieved better agreement with measurement than other methods described in literature. The method begins from a cylindrically symmetric pretarget electron beam (radius of 0.5 cm) of uniform intensity. This beam is subdivided into annular regions of fluence for which each region is individually transported through the accelerator head and into a water phantom. A simulated annealing search is then performed to determine the optimal combination of weights of the annular fluences that provide a best match between the measured dose distributions and the weighted sum of annular dose distributions for particular pretarget electron energy. When restricted to Gaussian intensity distributions, the optimization determined an optimal FWHM=1.34 mm for 18.0 MeV electrons, with a RMSE=0.49% on 40x40 cm{sup 2} lateral profiles. When allowed to deviate from Gaussian intensities a further reduction in RMSE was achieved. For our Clinac 21 EX accelerator MC model (based on the 1996 Varian Oncology Systems, Monte Carlo Project package), the optimal unrestricted intensity distribution was found to be a Gaussian-like solution (18.0 MeV, FWHM=1.10 mm, 40x40 cm{sup 2} profile, and RMSE=0.15%) with the presence of an extra focal halo contribution on the order of 10% of the maximum Gaussian intensity. Using the optimally derived intensity, 10x10 and 4x4 cm{sup 2} profiles were found to be in agreement with measurement with a maximum RMSE=0.49%. The optimized Gaussian and unrestricted values of the electron beam FWHM were both within the range of those inferred by focal spot image measurements performed by Jaffray et al.[''X-ray sources of medical linear accelerators: Focal and extra-focal radiation,'' Med. Phys. 20, 1417-1427 (1993)]. The inference of an extra focal pretarget electron component may be an indicator of a deficiency in the MC model and needs further investigation.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

A method for normalizing pathology images to improve feature extraction for quantitative pathology

Purpose: With the advent of digital slide scanning technologies and the potential proliferation of large repositories of digital pathology images, many research studies can leverage these data for biomedical discovery and to develop clinical applications. However, quantitative analysis of digital pathology images is impeded by batch effects generated by varied staining protocols and staining conditions of pathological slides. Methods: To overcome this problem, this paper proposes a novel, fully automated stain normalization method to reduce batch effects and thus aid research in digital pathology applications. Their method, intensity centering and histogram equalization (ICHE), normalizes a diverse set of pathology images by first scaling the centroids of the intensity histograms to a common point and then applying a modified version of contrast-limited adaptive histogram equalization. Normalization was performed on two datasets of digitized hematoxylin and eosin (H&E) slides of different tissue slices from the same lung tumor, and one immunohistochemistry dataset of digitized slides created by restaining one of the H&E datasets. Results: The ICHE method was evaluated based on image intensity values, quantitative features, and the effect on downstream applications, such as a computer aided diagnosis. For comparison, three methods from the literature were reimplemented and evaluated using the same criteria. The authors found that ICHE not only improved performance compared with un-normalized images, but in most cases showed improvement compared with previous methods for correcting batch effects in the literature. Conclusions: ICHE may be a useful preprocessing step a digital pathology image processing pipeline.

61 RADIATION PROTECTION AND DOSIMETRY↗

Conservative Multimoment Transport along Characteristics for Discontinuous Galerkin Methods

A set of algorithms based on characteristic discontinuous Galerkin methods is presented for tracer transport on the sphere. The algorithms are designed to reduce message passing interface communication volume per unit of simulated time relative to current methods generally, and to the spectral element scheme employed by the U.S. Department of Energy's Exascale Earth System Model (E3SM) specifically. Two methods are developed to enforce discrete mass conservation when the transport schemes are coupled to a separate dynamics solver; constrained transport and Jacobian-combined transport. A communication-efficient method is introduced to enforce tracer consistency between the transport scheme and dynamics solver; this method also provides the transport scheme's shape preservation capability. A subset of the algorithms derived here is implemented in E3SM and shown to improve transport performance by a factor of 2.2 for the model's standard configuration with 40 tracers at the strong scaling limit of one element per core. Read More: https://epubs.siam.org/doi/10.1137/18M1165943

97 MATHEMATICS AND COMPUTING↗

Gravitational collapse of scalar fields via spectral methods

In this paper we present a new numerical code based on the Galerkin method to integrate the field equations for the spherical collapse of massive and massless scalar fields. By using a spectral decomposition in terms of the radial coordinate, the field equations were reduced to a finite set of ordinary differential equations in the space of modes associated with the Galerkin expansion of the scalar field, together with algebraic sets of equations connecting modes associated with the metric functions. The set of ordinary differential equations with respect to the null coordinate is then integrated using an eighth-order Runge-Kutta method. The numerical tests have confirmed the high accuracy and fast convergence of the code. As an application we have evaluated the whole spectrum of black hole masses which ranges from infinitesimal to large values obtained after varying the amplitude of the initial scalar field distribution. We have found strong numerical evidence that this spectrum is described by a nonextensive distribution law.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗

Radial differential interior tomography and its image reconstruction with differentiated backprojection and projection onto convex sets

Purpose: Interior tomography has been recognized as one of the most effective approaches in computed tomography (CT) to reduce radiation dose rendered to patients. In this work, the authors propose and evaluate an imaging method of radial differential interior tomography.Methods: In interior tomography, an x-ray beam is collimated to only irradiate the region of interest (ROI) with suspected lesions while the surrounding area/volume of normal tissues/organs is spared. In the proposed imaging method of radial differential interior tomography, the outcome is a ROI image that has gone through a radial differential filtering. The image reconstruction algorithm for the radial differential interior tomography is kept in the fashion of differentiated backprojection and projection onto convex sets, but the required a priori knowledge in a small round area becomes zero and may be more readily available in practice.Results: Using the projection data simulated by computer and acquired by CT scanner, the authors evaluate and verify the performance of the proposed radial differential interior tomography method and its associated image reconstruction algorithm. The preliminary results show that the proposed imaging method can generate an image that is the radial differentiation of a conventional tomographic image and is robust over noise that inevitably exist in practice.Conclusions: It is believed that the proposed imaging method may find its utility in advanced clinical applications wherein a ROI-based image processing and analysis is required for lesion visualization, characterization, and diagnosis.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

SU-F-207-02: Use of Postmortem Subjects for Subjective Image Quality Assessment in Abdominal CT Protocols with Iterative Reconstruction

Purpose: New radiation dose reduction technologies are emerging constantly in the medical imaging field. The latest of these technologies, iterative reconstruction (IR) in CT, presents the ability to reduce dose significantly and hence provides great opportunity for CT protocol optimization. However, without effective analysis of image quality, the reduction in radiation exposure becomes irrelevant. This work explores the use of postmortem subjects as an image quality assessment medium for protocol optimizations in abdominal CT. Methods: Three female postmortem subjects were scanned using the Abdomen-Pelvis (AP) protocol at reduced minimum tube current and target noise index (SD) settings of 12.5, 17.5, 20.0, and 25.0. Images were reconstructed using two strengths of iterative reconstruction. Radiologists and radiology residents from several subspecialties were asked to evaluate 8 AP image sets including the current facility default scan protocol and 7 scans with the parameters varied as listed above. Images were viewed in the soft tissue window and scored on a 3-point scale as acceptable, borderline acceptable, and unacceptable for diagnosis. The facility default AP scan was identified to the reviewer while the 7 remaining AP scans were randomized and de-identified of acquisition and reconstruction details. The observers were also asked to comment on the subjective image quality criteria they used for scoring images. This included visibility of specific anatomical structures and tissue textures. Results: Radiologists scored images as acceptable or borderline acceptable for target noise index settings of up to 20. Due to the postmortem subjects’ close representation of living human anatomy, readers were able to evaluate images as they would those of actual patients. Conclusion: Postmortem subjects have already been proven useful for direct CT organ dose measurements. This work illustrates the validity of their use for the crucial evaluation of image quality during CT protocol optimization, especially when investigating the effects of new technologies.

60 APPLIED LIFE SCIENCES↗

A probabilistic graphical model based stochastic input model construction

Model reduction techniques have been widely used in modeling of high-dimensional stochastic input in uncertainty quantification tasks. However, the probabilistic modeling of random variables projected into reduced-order spaces presents a number of computational challenges. Due to the curse of dimensionality, the underlying dependence relationships between these random variables are difficult to capture. In this work, a probabilistic graphical model based approach is employed to learn the dependence by running a number of conditional independence tests using observation data. Thus a probabilistic model of the joint PDF is obtained and the PDF is factorized into a set of conditional distributions based on the dependence structure of the variables. The estimation of the joint PDF from data is then transformed to estimating conditional distributions under reduced dimensions. To improve the computational efficiency, a polynomial chaos expansion is further applied to represent the random field in terms of a set of standard random variables. This technique is combined with both linear and nonlinear model reduction methods. Numerical examples are presented to demonstrate the accuracy and efficiency of the probabilistic graphical model based stochastic input models. - Highlights: • Data-driven stochastic input models without the assumption of independence of the reduced random variables. • The problem is transformed to a Bayesian network structure learning problem. • Examples are given in flows in random media.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

libwfa: Wavefunction analysis tools for excited and open‐shell electronic states

Abstract An open‐source software library for wavefunction analysis, libwfa, provides a comprehensive and flexible toolbox for post‐processing excited‐state calculations, featuring a hierarchy of interconnected visual and quantitative analysis methods. These tools afford compact graphical representations of various excited‐state processes, provide detailed insight into electronic structure, and are suitable for automated processing of large data sets. The analysis is based on reduced quantities, such as state and transition density matrices (DMs), and allows one to distill simple molecular orbital pictures of physical phenomena from intricate correlated wavefunctions. The implemented descriptors provide a rigorous link between many‐body wavefunctions and intuitive physical and chemical models, for example, exciton binding, double excitations, orbital relaxation, and polyradical character. A broad range of quantum‐chemical methods is interfaced with libwfa via a uniform interface layer in the form of DMs. This contribution reviews the structure of libwfa and highlights its capabilities by several representative use cases. This article is categorized under: Software > Quantum Chemistry Theoretical and Physical Chemistry > Spectroscopy

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Comparison of integral equations used to study ${T}_{cc}^{+}$ for a stable D *

We perform a detailed comparison between three formalisms used in recent studies of DD* scattering at heavier-than-physical pion masses, which aim to understand the properties of the doubly-charmed tetraquark, ${T}_{cc}^{+}$ (3875). These methods are the three-particle relativistic field theory (RFT) formalism, the two-body Lippmann-Schwinger (LS) equation with chiral effective field theory potentials, and the two-particle relativistic framework proposed by Baião Raposo and Hansen (BRH approach). In a simplified single-channel setting, we derive the conditions under which the infinite-volume integral equations from the RFT and BRH approaches reduce to the LS form. We present numerical examples showing that differences between these methods can be largely removed by adjusting short-range couplings. We also address a number of technical issues in the RFT approach.

Hadronic Spectroscopy↗

Two locations, two times, and the element set

A robust analytical formulation is developed to apply classical initial orbital determination to artificial satellites whose locations are uncertain to about 1 cu km and separated in time by no more than 30 min. An analytical simplification reduces Gauss's method, iteration on the semilatus rectum, iteration on the true anomaly, and the Lambert-Euler technique, to the solution of a single equation in one unknown, instead of the usual coupled triplet of three equations in three unknowns. The method is demonstrated for all common artificial satellite orbits over a variety of time intervals between the two location vectors, and for a varied set of position and distance errors.

Taff, L. G.↗

Method for phase retrieval to reduce a sampling requirement when imaging a dynamic process

A method for retrieving phase information in a coherent diffraction imaging process includes acquiring a plurality of 3D data sets, each 3D data set corresponding to one of a plurality of time states, and reconstructing a 3D image of the object at a given time state using the 3D data set from all of the time states. Each 3D data set is acquired by: illuminating an object positioned in a first position with a coherent beam; measuring a first 2D diffraction pattern using an area detector; rotating the object around a tilt axis thereof to a second position that is different from the first position; re-illuminating the object positioned in the second position with the coherent beam; re-measuring a second 2D diffraction pattern using the area detector; and repeating the rotating, re-illuminating and re-measuring steps such that each 3D data set includes a predetermined number of diffraction patterns.

36 MATERIALS SCIENCE↗