Search NASA⌕ Search

SEARCH · Search NASA

Results for “variational 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 685 records · Page 38

Approximate Bayesian Model Inversion for PDEs with Heterogeneous and State-Dependent Coefficients

We present two approximate Bayesian inference methods for parameter estimation in partial differential equation (PDE) models with space-dependent and state-dependent parameters. We demonstrate that these methods provide accurate and cost-effective alternatives to Markov Chain Monte Carlo simulation. We assume a parameterized Gaussian prior on the unknown functions, and approximate the posterior density by a parameterized multivariate Gaussian density. The parameters of the prior and posterior are estimated from sparse observations of the PDE model's states and the unknown functions themselves by maximizing the evidence lower bound (ELBO), a lower bound on the log marginal likelihood of the observations. The first method, Laplace-EM, employs the expectation maximization algorithm to maximize the ELBO, with a Laplace approximation of the posterior on the E-step, and minimization of a Kullback-Leibler divergence on the M-step. The second method, DSVI-EB, employs the doubly stochastic variational inference (DSVI) algorithm, in which the ELBO is maximized via gradient-based stochastic optimization, with nosiy gradients computed via simple Monte Carlo sampling and Gaussian backpropagation. We apply these methods to identifying diffusion coefficients in linear and nonlinear diffusion equations, and we find that both methods provide accurate estimates of posterior densities and the hyperparameters of Gaussian priors. While the Laplace-EM method is more accurate, it requires computing Hessians of the physics model. The DSVI-EB method is found to be less accurate but only requires gradients of the physics model.

approximate Bayesian inference, model inversion, v↗

One directional polarized neutron reflectometry with optimized reference layer method

In the past decade, several neutron reflectometry methods for determining the modulus and phase of the complex reflection coefficient of an unknown multilayer thin film have been worked out among which the method of variation of surroundings and reference layers are of highest interest. These methods were later modified for measurement of the polarization of the reflected beam instead of the measurement of the intensities. In their new architecture, these methods not only suffered from the necessity of change of experimental setup but also another difficulty was added to their experimental implementations. This deficiency was related to the limitations of the technology of the neutron reflectometers that could only measure the polarization of the reflected neutrons in the same direction as the polarization of the incident beam. As the instruments are limited, the theory has to be optimized so that the experiment could be performed. In a recent work, we developed the method of variation of surroundings for one directional polarization analysis. In this new work, the method of reference layer with polarization analysis has been optimized to determine the phase and modulus of the unknown film with measurement of the polarization of the reflected neutrons in the same direction as the polarization of the incident beam.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Using multiple data sources to investigate foraging niche partitioning in sympatric obligate avian scavengers

As carrion feeders competing for a limited and ephemeral resource, avian scavengers are ideal model organisms to study mechanisms of niche partitioning. Previous work has focused on observations of species’ interactions at carcasses, and although fruitful, the inclusion of behavior beyond that at carcasses, such as movement patterns, could yield a more comprehensive understanding of mechanisms of foraging niche partitioning. Our goal was to assess how differences in physiological, morphological, and social characteristics between sympatric black vultures ( Coragyps atratus ) and turkey vultures ( Cathartes aura ) manifest interspecific variation in movement patterns, foraging habitats, and carcass use to reduce direct competition. Both species are obligate scavengers that co-occur across much of the New World. Turkey vultures are solitary foragers with enhanced olfactory capabilities and wings optimized for energy-efficient low-altitude soaring. Black vultures are aggressive, heavier bodied, and forage socially. We assessed interspecific variation in carcass exploitation using experimental carcass trials encompassing a range of habitats, carcass sizes, and seasons, and used GPS telemetry to compare how soaring conditions influenced daily space use and diurnal patterns of movement activity for each species. Turkey vultures occurred more often at small (i.e., rabbit) carcasses, and those obscured by forest cover than black vultures, and were the first vultures recorded at 94% of carcasses visited by both species. Both species increased movement rates and daily ranges when soaring conditions were favorable; however, turkey vultures had higher travel rates regardless of soaring conditions. Our results suggest the olfactory sense of turkey vultures allows them to locate smaller and visually obscured carcasses more efficiently than black vultures, facilitated by wing morphology that allows them to forage over large areas efficiently. Socially foraging black vultures appear specialized at locating larger carcasses in open habitats, which does not require as much foraging flight as turkey vultures. Our study illustrates how inclusion of movement ecology into niche partitioning studies provides a more complete understanding of the mechanisms of coexistence in avian scavenger guilds. This approach may provide important information to guide conservation efforts, such as supplemental feeding, of imperiled vulture species.

54 ENVIRONMENTAL SCIENCES↗

A New Hybrid Quantum-Classical Algorithm for Solving the Unit Commitment Problem

Solving problems related to planning and operations of large-scale power systems is challenging on classical computers due to their inherent nature as mixed-integer and nonlinear problems. Quantum computing provides new avenues to approach these problems. We develop a hybrid quantum-classical algorithm for the Unit Commitment (UC) problem in power systems which aims at minimizing the total cost while optimally allocating generating units to meet the hourly demand of the power loads. The hybrid algorithm combines a variational quantum algorithm (VQA) with a classical Benders-type heuristic. The resulting algorithm computes approximate solutions to UC in three stages: i) a collection of UC vectors capable meeting the power demand with lowest possible operating costs is generated based on VQA; ii) a classical sequential least squares programming (SLSQP) routine is leveraged to find the optimal power level corresponding to a predetermined number of candidate vectors; iii) in the last stage, the approximate solution of UC along with generating units power level combination is given. To demonstrate the effectiveness of the presented method, three different systems with 3 generating units, 10 generating units, and 26 generating units were tested for different time periods. In addition, convergence of the hybrid quantum-classical algorithm for select time periods is proven out on IonQ's Forte system.

Aboumrad, Willie [IonQ, Inc]↗

Optimizing weak lensing mass estimates for cluster profile uncertainty

Weak lensing measurements of cluster masses are necessary for calibrating mass-observable relations (MORs) to investigate the growth of structure and the properties of dark energy. However, the measured cluster shear signal varies at fixed mass M 200m due to inherent ellipticity of background galaxies, intervening structures along the line of sight, and variations in the cluster structure due to scatter in concentrations, asphericity and substructure. We use N-body simulated halos to derive and evaluate a weak lensing circular aperture mass measurement M ap that minimizes the mass estimate variance <(M ap - M 200m ) 2 > in the presence of all these forms of variability. Depending on halo mass and observational conditions, the resulting mass estimator improves on M ap filters optimized for circular NFW-profile clusters in the presence of uncorrelated large scale structure (LSS) about as much as the latter improve on an estimator that only minimizes the influence of shape noise. Optimizing for uncorrelated LSS while ignoring the variation of internal cluster structure puts too much weight on the profile near the cores of halos, and under some circumstances can even be worse than not accounting for LSS at all. As a result, we discuss the impact of variability in cluster structure and correlated structures on the design and performance of weak lensing surveys intended to calibrate cluster MORs.

79 ASTRONOMY AND ASTROPHYSICS↗

DAKOTA (Design Analysis Kit for Optimization and Terascale) v.6.14

Sandia's Dakota software supports science and engineering transformation through advanced exploration of simulations. Specifically, Dakota manages and analyzes ensembles of simulations to provide broader and deeper perspective for analysts and decision makers. This enables them to: (1) enhance understanding of risk, (2) improve products, and (3) assess simulation credibility. In its simplest mode, Dakota can automate typical parameter variation studies through a generic interface to a computational model. Dakota also delivers advanced parametric analysis techniques enabling design exploration, optimization, model calibration, risk analysis, and quantification of margins and uncertainty with such models. It directly supports verification and validation activities. The algorithms implemented in Dakota aim to address challenges in performing these analyses with complex science and engineering models from desktop to high performance computers. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525 SAND2021-10468 O

Adams, BrianM.↗

DAKOTA (Design Analysis Kit for Optimization and Terascale) v.6.16

Sandia's Dakota software supports science and engineering transformation through advanced exploration of simulations. Specifically, Dakota manages and analyzes ensembles of simulations to provide broader and deeper perspective for analysts and decision makers. This enables them to: (1) enhance understanding of risk, (2) improve products, and (3) assess simulation credibility. In its simplest mode, Dakota can automate typical parameter variation studies through a generic interface to a computational model. Dakota also delivers advanced parametric analysis techniques enabling design exploration, optimization, model calibration, risk analysis, and quantification of margins and uncertainty with such models. It directly supports verification and validation activities. The algorithms implemented in Dakota aim to address challenges in performing these analyses with complex science and engineering models from desktop to high performance computers. Sandia National Laboratories is a multimission laboratory managed and operated by National Technology & Engineering Solutions of Sandia, LLC, a wholly owned subsidiary of Honeywell International Inc., for the U.S. Department of Energy’s National Nuclear Security Administration under contract DE-NA0003525 SAND2021-10468 O

Adams, BrianM.↗

Path-space variational inference for non-equilibrium coarse-grained systems

In this paper we discuss information-theoretic tools for obtaining optimized coarse-grained molecular models for both equilibrium and non-equilibrium molecular simulations. The latter are ubiquitous in physicochemical and biological applications, where they are typically associated with coupling mechanisms, multi-physics and/or boundary conditions. In general the non-equilibrium steady states are not known explicitly as they do not necessarily have a Gibbs structure. The presented approach can compare microscopic behavior of molecular systems to parametric and non-parametric coarse-grained models using the relative entropy between distributions on the path space and setting up a corresponding path-space variational inference problem. The methods can become entirely data-driven when the microscopic dynamics are replaced with corresponding correlated data in the form of time series. Furthermore, we present connections and generalizations of force matching methods in coarse-graining with path-space information methods. We demonstrate the enhanced transferability of information-based parameterizations to different observables, at a specific thermodynamic point, due to information inequalities. We discuss methodological connections between information-based coarse-graining of molecular systems and variational inference methods primarily developed in the machine learning community. However, we note that the work presented here addresses variational inference for correlated time series due to the focus on dynamics. The applicability of the proposed methods is demonstrated on high-dimensional stochastic processes given by overdamped and driven Langevin dynamics of interacting particles.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Variational Dirichlet Blur Kernel Estimation

Blind image deconvolution involves two key objectives: 1) latent image and 2) blur estimation. For latent image estimation, we propose a fast deconvolution algorithm, which uses an image prior of nondimensional Gaussianity measure to enforce sparsity and an undetermined boundary condition methodology to reduce boundary artifacts. For blur estimation, a linear inverse problem with normalization and nonnegative constraints must be solved. However, the normalization constraint is ignored in many blind image deblurring methods, mainly because it makes the problem less tractable. In this paper, we show that the normalization constraint can be very naturally incorporated into the estimation process by using a Dirichlet distribution to approximate the posterior distribution of the blur. Making use of variational Dirichlet approximation, we provide a blur posterior approximation that considers the uncertainty of the estimate and removes noise in the estimated kernel. Furthermore, experiments with synthetic and real data demonstrate that the proposed method is very competitive to the state-of-the-art blind image restoration methods.

97 MATHEMATICS AND COMPUTING↗

Deep reinforcement learning assisted co-optimization of Volt-VAR grid service in distribution networks

With the increasing penetration of distributed energy resources in distribution networks, Volt-VAR control and optimization (VVC/VVO) have become very important to ensure an acceptable quality of service to all customers. System operators can rely on slow-responding utility devices, including capacitor banks and on-load tap changing transformers, along with fast-responding battery and photovoltaic (PV) inverters for the VVC/VVO implementation. Because of variations in response time of these two classes of devices, and different control actions (discrete versus continuous), coordinated and optimal scheduling and operation have become of utmost importance. Here, this paper develops a look-ahead deep reinforcement learning (DRL)-based multi-objective VVO technique to improve the voltage profile of active distribution networks, decrease network and inverter power loss, and save the operational cost of the grid. It proposes a deep deterministic policy gradient (DDPG)-based approach to schedule the optimal reactive and/or active power set-points of fast-responding inverters, and a deep Q-network (DQN)-based DRL agent to schedule the discrete decisions variables of slow-responding assets. The reactive power output of PV and battery smart inverters are scheduled at 30-minute intervals and the capacitors’ commitment status is scheduled with several hour intervals. The proposed framework is validated on the modified IEEE 34-bus and 123-bus test cases with embedded PV and PV-plus-storage. To validate the efficacy of the proposed VVO, it is compared with several scenarios, including the base case without VVO, localized droop control of DERs, DDPG-only, and twin delayed DDPG (TD3) agent-based DRL techniques. The results justify the superior performance of the proposed method to improve the voltage profile, reduce network power loss, and minimize the look-ahead grid operational cost while minimizing the undesirable power losses in inverters as a result of power factor adjustments.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Swarm Intelligence Based Optimal Design of Local Volt/Var Control Function for Distributed Energy Resources

The increasing penetration of renewable based distributed energy resources (DERs) in distribution network (DN) leads to larger and more frequent voltage variation in distributions network (DN), thus posing challenges on voltage control. Real-time local voltage control method is a promising solution for the above issue. However, the local voltage control function needs to be customized and optimized according to real distribution system condition. In this paper, a swarm intelligence based Volt/Var control optimal design method (SO-VVC) is proposed to optimize the control function. Compared with existing approaches, the proposed method can not only represent the nonlinear behaviour of power flow but is also computation efficient. The performance of the proposed SO-VVC is demonstrated by case studies on a modified IEEE-123 bus system.

Zhang, Zhengfa [University of Tennessee, Knoxville↗

Mitigating the noise of DESI mocks using analytic control variates

In order to address fundamental questions related to the expansion history of the Universe and its primordial nature with the next generation of galaxy experiments, we need to model reliably large-scale structure observables such as the correlation function and the power spectrum. Cosmological N-body simulations provide a reference through which we can test our models, but their output suffers from sample variance on large scales. Fortunately, this is the regime where accurate analytic approximations exist. To reduce the variance, which is key to making optimal use of these simulations, we can leverage the accuracy and precision of such analytic descriptions using Control Variates (CV). The power of control variates stems from utilizing inexpensive but highly correlated surrogates of the statistics one wishes to measure. The stronger the correlation between the surrogate and the statistic of interest, the larger the variance reduction delivered by the method. We apply two control variate formulations to mock catalogs generated in anticipation of upcoming data from the Dark Energy Spectroscopic Instrument (DESI) to test the robustness of its analysis pipeline. Our CV-reduced measurements offer a factor of 5-10 improvement in the measurement error compared with the raw measurements. We explore the relevant properties of the galaxy samples that dictate this reduction and comment on the improvements we find on some of the derived quantities relevant to Baryon Acoustic Oscillation (BAO) analysis.

79 ASTRONOMY AND ASTROPHYSICS↗

Patient-based radiographic exposure factor selection: a systematic review

Digital technology has wider exposure latitude and post-processing algorithms which can mask the evidence of underexposure and overexposure. Underexposure produces noisy, grainy images which can impede diagnosis and overexposure results in a greater radiation dose to the patient. These exposure errors can result from inaccurate adjustment of exposure factors in response to changes in patient thickness. This study aims to identify all published radiographic exposure adaptation systems which have been, or are being, used in general radiography and discuss their applicability to digital systems. Studies in EMBASE, MEDLINE, CINAHL and SCOPUS were systematically reviewed. Some of the search terms used were exposure adaptation, exposure selection, exposure technique, 25% rule, 15% rule, DuPont™ Bit System and radiography. A manual journal-specific search was also conducted in The Radiographer and Radiologic Technology. Studies were included if they demonstrated a system of altering exposure factors to compensate for variations in patients for general radiography. Studies were excluded if they focused on finding optimal exposures for an ‘average’ patient or focused on the relationship between exposure factors and dose. The database search uncovered 11 articles and the journal-specific search uncovered 13 articles discussing systems of exposure adaptation. They can be categorised as simple one-step guidelines, comprehensive charts and computer programs. Only two papers assessed the efficacy of exposure adjustment systems. No literature compares the efficacy of exposure adaptations system for film/screen radiography with digital radiography technology nor is there literature on a digital specific exposure adaptation system.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

A Stochastic Maximum Principle for a Stochastic Differential Game of a Mean-Field Type

We construct a stochastic maximum principle (SMP) which provides necessary conditions for the existence of Nash equilibria in a certain form of N-agent stochastic differential game (SDG) of a mean-field type. The information structure considered for the SDG is of a possible asymmetric and partial type. To prove our SMP we take an approach based on spike-variations and adjoint representation techniques, analogous to that of S. Peng (SIAM J. Control Optim. 28(4):966-979, 1990) in the optimal stochastic control context. In our proof we apply adjoint representation procedures at three points. The first-order adjoint processes are defined as solutions to certain mean-field backward stochastic differential equations, and second-order adjoint processes of a first type are defined as solutions to certain backward stochastic differential equations. Second-order adjoint processes of a second type are defined as solutions of certain backward stochastic equations of a type that we introduce in this paper, and which we term conditional mean-field backward stochastic differential equations. From the resulting representations, we show that the terms relating to these second-order adjoint processes of the second type are of an order such that they do not appear in our final SMP equations. A comparable situation exists in an article by R. Buckdahn, B. Djehiche, and J. Li (Appl. Math. Optim. 64(2):197-216, 2011) that constructs a SMP for a mean-field type optimal stochastic control problem; however, the approach we take of using these second-order adjoint processes of a second type to deal with the type of terms that we refer to as the second form of quadratic-type terms represents an alternative to a development, to our setting, of the approach used in their article for their analogous type of term.

97 MATHEMATICS AND COMPUTING↗

Hydrokinetic energy conversion using flow induced oscillations of single-cylinder with large passive turbulence control

Various types of flow-induced oscillations (FIOs) have been implemented in development of marine hydrokinetic (MHK) energy converters. With passive turbulence control (PTC), energy harvesting starts at a flow speed of about 0.5 m/s. However, there is worldwide MHK energy available in even slower currents. In the present study, the effect of damping on FIO and power extraction is investigated for a converter with large turbulence stimulation (PTC) consisting of straight strips with a height of 15% of the cylinder diameter and placed symmetrically on the cylinder surface. The oscillating amplitude decreases, as the damping ratio increases, with unchanged sinusoidal pattern of the displacement time-history. The frequency ratio is also affected by damping especially in the VIV initial branch and transition region between VIV and galloping. An important flow characteristic of the large-PTC cylinder is that a recirculation region is formed behind the PTC, causing appreciable disturbance to the flow past the cylinder. Power can be harvested in the whole FIO range and the harnessed power maximum appears at the largest inflow velocity tested. However, the optimum of harnessing efficiency is located at the beginning of the VIV upper branch. The gap between VIV and galloping is bridged when large PTC is used, eliminating the drop in power and efficiency even at higher damping, which would be a weakness of regular-PTC cylinder for energy harvesting. Finally, the mechanism behind the variation of harnessing efficiency with inflow velocity and damping ratio is revealed, and the optimality criterion for the converter design is discussed.

16 TIDAL AND WAVE POWER↗

Deformation of pores in response to uniaxial and hydrostatic stress cycling in Marcellus Shale: Implications for gas recovery

One of the main challenges during gas production from shale reservoirs is low recovery rate. One contributing factor to this outcome is an insufficient understanding of pore systems, especially pore behaviour following changes in reservoir conditions or resulting from gas production practices. Because the pressure in the producing well can be controlled, understanding the effects of pressure variation on the pore size distribution and methane trapping is necessary to help design optimal conditions to improve the gas recovery rate. This work is the first systematic study of sub-millimeter pore deformation in shale caused by uniaxial and hydrostatic stress up to 100 MPa. Overmature samples from the Middle Devonian Marcellus Shale were analyzed using neutron scattering (SANS and USANS) techniques to interpret the response of nanopores to stress cycling of magnitude and duration compatible with the hydraulic fracturing treatments. Experiments reported here are performed at a series of uniaxial pressures up to 100 MPa and at hydrostatic pressures of deuterated methane 0 and 50 MPa. Since at the original depth of the shale samples’ burial of 2184 m the hydrostatic pressure is approximately 27 MPa and the lithostatic pressure is about 55 MPa, the experimental conditions reasonably well simulate the reservoir pressure regime. Our SANS and USANS results show that different pore sizes are affected by uniaxial stress in different ways. Specifically, in the pore size range from 1 nm to 800 nm, a decrease of pore density with pressure is observed, with the most depleted being mesopores of about 100 nm in diameter. The observed decrease is likely related to deformation of kerogen, followed by a loss of pore nano-volume, as well as methane trapped in the micropores. For pores larger than 5 μm, USANS data suggest that the negative trend is reversed at above 74 MPa, and the number density of large macropores may increase with increased stress even above the original value. The increased number of macropores at high pressure may create new interconnected conduits for gas migration, resulting in a better recovery rate. Another important finding of this study is an irreversible rearrangement of pore size distribution taking place after pressure cycling. Furthermore, this irreversible reorganization of pore size distribution should be taken into account during management of well production to maximize recovery rate.

03 NATURAL GAS↗

{sup 18}F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography-Based Radiotherapy Target Volume Definition in Non-Small-Cell Lung Cancer: Delineation by Radiation Oncologists vs. Joint Outlining With a PET Radiologist?

Purpose: {sup 18}F-Fluorodeoxyglucose positron emission tomography/computed tomography (PET/CT) has benefits in target volume (TV) definition in radiotherapy treatment planning (RTP) for non-small-cell lung cancer (NSCLC); however, an optimal protocol for TV delineation has not been determined. We investigate volumetric and positional variation in gross tumor volume (GTV) delineation using a planning PET/CT among three radiation oncologists and a PET radiologist. Methods and Materials: RTP PET/CT scans were performed on 28 NSCLC patients (Stage IA-IIIB) of which 14 patients received prior induction chemotherapy. Three radiation oncologists and one PET radiologist working with a fourth radiation oncologist independently delineated the GTV on CT alone (GTV{sub CT}) and on fused PET/CT images (GTV{sub PETCT}). The mean percentage volume change (PVC) between GTV{sub CT} and GTV{sub PETCT} for the radiation oncologists and the PVC between GTV{sub CT} and GTV{sub PETCT} for the PET radiologist were compared using the Wilcoxon signed-rank test. Concordance index (CI) was used to assess both positional and volume change between GTV{sub CT} and GTV{sub PETCT} in a single measurement. Results: For all patients, a significant difference in PVC from GTV{sub CT} to GTV{sub PETCT} exists between the radiation oncologist (median, 5.9%), and the PET radiologist (median, -0.4%, p = 0.001). However, no significant difference in median concordance index (comparing GTV{sub CT} and GTV{sub FUSED} for individual cases) was observed (PET radiologist = 0.73; radiation oncologists = 0.66; p = 0.088). Conclusions: Percentage volume changes from GTV{sub CT} to GTV{sub PETCT} were lower for the PET radiologist than for the radiation oncologists, suggesting a lower impact of PET/CT in TV delineation for the PET radiologist than for the oncologists. Guidelines are needed to standardize the use of PET/CT for TV delineation in RTP.

62 RADIOLOGY AND NUCLEAR MEDICINE↗