Search NASA⌕ Search

SEARCH · Search NASA

Results for “Stochastic 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 721 records · Page 40

A Stochastic Programming Formulation for Disinfectant Booster Station Placement to Protect Large-Scale Water Distribution Networks

We introduce a methodology for optimally locating fixed disinfectant booster stations for response to contamination incidents in water distribution networks. In this work a stochastic programming problem considering uncertainty in both the location and time of the contamination incident is formulated, resulting in a large Mixed Integer Linear Programming problem. While the original full-space problem is intractably large, we show a series of reductions that decrease the size of the problem by up to five orders of magnitude and allow solutions of the optimal placement problem for realistically-sized water network models.

42 ENGINEERING↗

Design optimization of a scramjet under uncertainty using probabilistic learning on manifolds

Here, we demonstrate, on a scramjet combustion problem, a constrained probabilistic learning approach that augments physics-based datasets with realizations that adhere to underlying constraints and scatter. The constraints are captured and delineated through diffusion maps, while the scatter is captured and sampled through a projected stochastic differential equation. The objective function and constraints of the optimization problem are then efficiently framed as non-parametric conditional expectations. Different spatial resolutions of a large-eddy simulation filter are used to explore the robustness of the model to the training dataset and to gain insight into the significance of spatial resolution on optimal design.

42 ENGINEERING↗

Advising Big Data Transfer Over Dedicated Connections Based on Profiling Optimization

Big data transfer in next-generation scientific applications is currently commonly carried out over dedicated channels in high-performance networks (HPNs), where transport protocols play a critical role in maximizing application-level throughput. Optimizing the performance of these protocols is challenging: i) transport protocols perform differently in various network environments, and the protocol choice is not straightforward; ii) even for a given protocol in a given environment, different parameter settings of the protocol may lead to significantly different performance and oftentimes the default setting does not yield the best performance. However, it is prohibitively time-consuming to conduct exhaustive transport profiling due to the large parameter space. In this paper, we propose a PRofiling Optimization Based DAta Transfer Advisor (ProbData) to help end users determine the most effective transport method with the most appropriate parameter settings to achieve satisfactory performance for big data transfer over dedicated connections in HPNs. ProbData employs a fast profiling scheme based on the Simultaneous Perturbation Stochastic Approximation algorithm, namely, FastProf, to accelerate the exploration of the optimal operational zones of various transport methods to improve profiling efficiency. We first present a theoretical background of the optimized profiling approach in ProbData and then detail its design and implementation. The advising procedure and performance benefits of FastProf and ProbData are illustrated and evaluated by both extensive emulations based on real-life performance measurements and experiments over various physical connections in existing production HPNs.

97 MATHEMATICS AND COMPUTING↗

A characterization of robust radiation therapy treatment planning methods-from expected value to worst case optimization

Purpose: To characterize a class of optimization formulations used to handle systematic and random errors in radiation therapy, and to study the differences between the methods within this class. Methods: The class of robust methods that can be formulated as minimax stochastic programs is studied. This class generalizes many previously used methods, ranging between optimization of the expected and the worst case objective value. The robust methods are used to plan intensity-modulated proton therapy (IMPT) treatments for a case subject to systematic setup and range errors, random setup errors with and without uncertain probability distribution, and combinations thereof. As reference, plans resulting from a conventional method that uses a margin to account for errors are shown. Results: For all types of errors, target coverage robustness increased with the conservativeness of the method. For systematic errors, best case organ at risk (OAR) doses increased and worst case doses decreased with the conservativeness. Accounting for random errors of fixed probability distribution resulted in heterogeneous dose. The heterogeneities were reduced when uncertainty in the probability distribution was accounted for. Doing so, the OAR doses decreased with the conservativeness. All robust methods studied resulted in more robust target coverage and lower OAR doses than the conventional method. Conclusions: Accounting for uncertainties is essential to ensure plan quality in complex radiation therapy such as IMPT. The utilization of more information than conventional in the optimization can lead to robust target coverage and low OAR doses. Increased target coverage robustness can be achieved by more conservative methods.

60 APPLIED LIFE SCIENCES↗

Quantum optimal control of multiple weakly interacting molecular rotors in the time-dependent Hartree approximation

In this work we perform quantum optimal control simulations, based on the Time-Dependent Hartree (TDH) approximation, for systems of three to five dipole-dipole coupled OCS rotors. A control electric field is used to steer all of the individual rotors, arranged in chains and regular polygons in a plane, toward either identical or unique objectives. The goal is to explore the utility of the TDH approximation to model the field-induced dynamics of multiple interacting rotors in the weak dipole-dipole coupling regime. A stochastic hill climbing approach is employed to seek an optimal control field that achieves the desired objectives at a specified target time. We first show that multiple rotors in chain and polygon geometries can be identically oriented in the same direction; these cases do not significantly depend on the presence of the dipole-dipole interaction. Additionally, in particular geometrical arrangements, we demonstrate that individual rotors can be uniquely manipulated toward different objectives with the same field. Specifically, it is shown that for a three rotor chain, the two end rotors can be identically oriented in a specific direction while keeping the middle rotor in its ground state, and for an equilateral triangle, two rotors can be identically oriented in a specific direction while the third rotor is oriented in the opposite direction. These multirotor unique objective cases exploit the shape of the field in coordination with dipole-dipole coupling between the rotors. Comparisons to numerically exact calculations, utilizing the TDH-determined fields, are given for all optimal control studies involving systems of three rotors.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Uncertainty-Informed Operation Coordination in a Water-Energy Nexus

The widespread deployment of smart heterogeneous technologies and the growing complexity in our modern society calls for effective coordination of the interdependent lifeline networks. In particular, operation coordination of electric power and water infrastructures is urgently needed as the water system is one of the most energy-intensive networks, an interruption in which may quickly evolve into a dramatic societal concern. This paper develops a novel analytic for uncertainty-aware day-ahead operation optimization of the interconnected power and water systems (PaWS). Joint probabilistic constraint (JPC) programming is employed to capture the uncertainties in wind resources and water demand forecasts. The proposed integrated stochastic model is presented as a non-linear non-convex optimization problem, where the non-linear hydraulic constraints in the water network are linearized using piece-wise linearization technique, and the non-convexity is efficiently tackled with a solution methodology to convert the proposed model with JPCs to a tractable mixed-integer linear programming (MILP) formulation that can be quickly solved to optimality. Here, the suggested framework is applied to a 15-node commercial-scale water network jointly operated with a power transmission system using a modified IEEE 57-bus test system. The numerical results demonstrate the of the proposed stochastic framework, resulting in cost reduction (13% on average when compared to the traditional setting) and energy saving of the integrated model under different realizations of uncertain renewable energy sources (RESs) and water demand scenarios. Additionally, the scalability of the proposed model is tested on a modified IEEE 118-bus test system connected to five water networks.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Stochastic average model methods

We consider the solution of finite-sum minimization problems, such as those appearing in nonlinear least-squares or general empirical risk minimization problems. We are motivated by problems in which the summand functions are computationally expensive and evaluating all summands on every iteration of an optimization method may be undesirable. Here we present the idea of stochastic average model (SAM) methods, inspired by stochastic average gradient methods. SAM methods sample component functions on each iteration of a trust-region method according to a discrete probability distribution on component functions; the distribution is designed to minimize an upper bound on the variance of the resulting stochastic model. We present promising numerical results concerning an implemented variant extending the derivative-free model-based trust-region solver POUNDERS, which we name SAM-POUNDERS.

97 MATHEMATICS AND COMPUTING↗

Continuous-Time Public Good Contribution Under Uncertainty: A Stochastic Control Approach

In this paper we study continuous-time stochastic control problems with both monotone and classical controls motivated by the so-called public good contribution problem. That is the problem of n economic agents aiming to maximize their expected utility allocating initial wealth over a given time period between private consumption and irreversible contributions to increase the level of some public good. We investigate the corresponding social planner problem and the case of strategic interaction between the agents, i.e. the public good contribution game. We show existence and uniqueness of the social planner’s optimal policy, we characterize it by necessary and sufficient stochastic Kuhn–Tucker conditions and we provide its expression in terms of the unique optional solution of a stochastic backward equation. Similar stochastic first order conditions prove to be very useful for studying any Nash equilibria of the public good contribution game. In the symmetric case they allow us to prove (qualitative) uniqueness of the Nash equilibrium, which we again construct as the unique optional solution of a stochastic backward equation. We finally also provide a detailed analysis of the so-called free rider effect.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

H ∞ Control for Energy Dispatch in Autonomous Nanogrid With Communication Delays

This paper proposes an optimal controller and estimator for energy dispatch to balance the power supply and demand considering communication delays. The proposed algorithm involves modeling an autonomous nanogrid (ANG) consisting of distributed energy resources, energy storage systems, loads, an $H$ ∞ controller with a reference power modulation technique, and a state estimator. The ANG was developed to express the dynamic supply-demand energy balance of a nanogird system. Reference power modulation was designed to generate the desired ESS power based on the imbalanced energy. Random communication delays were modeled using a stochastic variable satisfying the Bernoulli random binary distribution. The optimal $H$ ∞ controller and estimator were developed using a linear matrix inequality approach to exponentially stabilize the closed-loop system. Simulations were performed using real daily demand forecasts obtained from the Korea Meteorological Administration to demonstrate the effectiveness of the proposed real-time optimization algorithm.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Stochastic scheduling of generating units with weekly energy storage: A hybrid decomposition approach

We propose a solution method for the large-scale stochastic unit commitment (SUC) problem with weekly-dispatched energy storage and significant weather-dependent stochastic generating capacity. Weekly storage facilities that mostly charge during weekends and discharge during weekdays require a weekly scheduling of generating units, which result in a large-scale optimization problem. This SUC problem is formulated as a two-stage stochastic model and we use the conditional value-at-risk as a risk measure. Using a Benders framework, the proposed solution method decomposes the problem into a mixed-integer linear master problem and linear and continuous subproblems. The master problem corresponds to the first-stage decisions throughout the week and includes all the commitment (binary) variables and their corresponding constraints. The subproblems correspond to the actual dispatch of the generating units on a weekly basis. Based on the success of column-and-constraint generation algorithms to solve robust optimization problems, we improve the low communication between the master problem and the subproblems in the standard Benders decomposition by adding primal variables and constraints from the subproblems to the master problem, which provides a better approximation of the recourse function. Furthermore, our computational experiments demonstrate the effectiveness of the proposed decomposition method using an instance of the South Carolina synthetic system with 90 generating units under 40 scenarios.

25 ENERGY STORAGE↗

Optimization Methods for Large-Scale Machine Learning

Here this paper provides a review and commentary on the past, present, and future of numerical optimization algorithms in the context of machine learning applications. Through case studies on text classification and the training of deep neural networks, we discuss how optimization problems arise in machine learning and what makes them challenging. A major theme of our study is that large-scale machine learning represents a distinctive setting in which the stochastic gradient (SG) method has traditionally played a central role while conventional gradient-based nonlinear optimization techniques typically falter. Based on this viewpoint, we present a comprehensive theory of a straightforward, yet versatile SG algorithm, discuss its practical behavior, and highlight opportunities for designing algorithms with improved performance. This leads to a discussion about the next generation of optimization methods for large-scale machine learning, including an investigation of two main streams of research on techniques that diminish noise in the stochastic directions and methods that make use of second-order derivative approximations.

97 MATHEMATICS AND COMPUTING↗

Do optimization methods in deep learning applications matter?

With advances in deep learning, exponential data growth and increasing model complexity, developing efficient optimization methods are attracting much research attention. Several implementations favor the use of Conjugate Gradient (CG) and Stochastic Gradient Descent (SGD) as being practical and elegant solutions to achieve quick convergence, however, these optimization processes also present many limitations in learning across deep learning applications. Recent research is exploring higher-order optimization functions as better approaches, but these present very complex computational challenges for practical use. Comparing first and higher-order optimization functions, in this paper, our experiments reveal that Levemberg-Marquardt (LM) significantly supersedes optimal convergence but suffers from very large processing time increasing the training complexity of both, classification and reinforcement learning problems. Our experiments compare off-the-shelf optimization functions(CG, SGD, LM and L-BFGS) in standard CIFAR, MNIST, CartPole and FlappyBird experiments.The paper presents arguments on which optimization functions to use and further, which functions would benefit from parallelization efforts to improve pretraining time and learning rate convergence.

Ozyildirim, Buse Melis↗

SU-F-T-182: A Stochastic Approach to Daily QA Tolerances On Spot Properties for Proton Pencil Beam Scanning

Purpose: Proton pencil beam scanning is used clinically across the United States. There are no current guidelines on tolerances for daily QA specific to pencil beam scanning, specifically related to the individual spot properties (spot width). Using a stochastic method to determine tolerances has the potential to optimize tolerances on individual spots and decrease the number of false positive failures in daily QA. Individual and global spot tolerances were evaluated. Methods: As part of daily QA for proton pencil beam scanning, a field of 16 spots (corresponding to 8 energies) is measured using an array of ion chambers (Matrixx, IBA). Each individual spot is fit to two Gaussian functions (x,y). The spot width (σ) in × and y are recorded (32 parameters). Results from the daily QA were retrospectively analyzed for 100 days of data. The deviations of the spot widths were histogrammed and fit to a Gaussian function. The stochastic spot tolerance was taken to be the mean ± 3σ. Using these results, tolerances were developed and tested against known deviations in spot width. Results: The individual spot tolerances derived with the stochastic method decreased in 30/32 instances. Using the previous tolerances (± 20% width), the daily QA would have detected 0/20 days of the deviation. Using a tolerance of any 6 spots failing the stochastic tolerance, 18/20 days of the deviation would have been detected. Conclusion: Using a stochastic method we have been able to decrease daily tolerances on the spot widths for 30/32 spot widths measured. The stochastic tolerances can lead to detection of deviations that previously would have been picked up on monthly QA and missed by daily QA. This method could be easily extended for evaluation of other QA parameters in proton spot scanning.

60 APPLIED LIFE SCIENCES↗

Improved statistical models for limited datasets in uncertainty quantification using stochastic collocation

This paper presents a data-driven framework for performing uncertainty quantification (UQ) by choosing a stochastic model that accurately describes the sources of uncertainty in a system. This model is propagated through an appropriate response surface function that approximates the behavior of this system using stochastic collocation. Given a sample of data describing the uncertainty in the inputs, our goal is to estimate a probability density function (PDF) using the kernel moment matching (KMM) method so that this PDF can be used to accurately reproduce statistics like mean and variance of the response surface function. Instead of constraining the PDF to be optimal for a particular response function, we show that we can use the properties of stochastic collocation to make the estimated PDF optimal for a wide variety of response functions. We contrast this method with other traditional procedures that rely on the Maximum Likelihood approach, like kernel density estimation (KDE) and its adaptive modification (AKDE). We argue that this modified KMM method tries to preserve what is known from the given data and is the better approach when the available data is limited in quantity. We test the performance of these methods for both univariate and multivariate density estimation by sampling random datasets from known PDFs and then measuring the accuracy of the estimated PDFs, using the known PDF as a reference. Comparing the output mean and variance estimated with the empirical moments using the raw data sample as well as the actual moments using the known PDF, we show that the KMM method performs better than KDE and AKDE in predicting these moments with greater accuracy. This improvement in accuracy is also demonstrated for the case of UQ in electrostatic and electrothermomechanical microactuators. We show how our framework results in the accurate computation of statistics in micromechanical systems.

97 MATHEMATICS AND COMPUTING↗

Linearly Adjustable International Portfolios

We present an approach to multi-stage international portfolio optimization based on the imposition of a linear structure on the recourse decisions. Multiperiod decision problems are traditionally formulated as stochastic programs. Scenario tree based solutions however can become intractable as the number of stages increases. By restricting the space of decision policies to linear rules, we obtain a conservative tractable approximation to the original problem. Local asset prices and foreign exchange rates are modelled separately, which allows for a direct measure of their impact on the final portfolio value.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Characterizing and improving the performance of molten-salt-steam heat exchangers in concentrating solar power plants

Shell-and-tube heat exchangers (HXs) for steam generation from molten salts in concentrating solar power (CSP) plants experience thermal fatigue due to significant temperature gradients and inherent transient operation. Molten salt-steam HX design lifespans exceed actual lifespans, and, as a consequence, designers overpredict plant profitability and operators neglect appropriate prescriptions to optimize these lifetimes. Here, this study refines HX lifespan estimates with data benchmarked against thermal-fluid mechanical modeling of stress and accumulated fatigue. Reduced-order thermal models of the molten salt-steam, shell-and-tube evaporator and superheater predict transient temperature profiles along the two HXs salt-steam flow paths. The modeled evaporator and superheater temperature profiles enable assessment of cyclic stresses within the HX tubesheets, where molten-salt HX failures are most common. Evaporator and superheater performance data from a current 110 MW elec commercial CSP plant provide a basis for validating the reduced-order HX models. HX life predictions derived from stochastic failure distributions serve as inputs for simulating and optimizing existing plant operations. The impact of the updated lifespans on overall plant revenue depends on operating scenarios. This study suggests that typical ramping rates for a CSP plant with a high-temperature Rankine cycle result in an evaporator and superheater life of approximately 10 and 25 years, respectively, compared to the design target of 30 years. Reduced HX lifespans decrease operational plant revenue on average by 4.6-5.1%. Furthermore, there may be as many as four HX replacements over the 30-year lifetime of the plant; and, purchase agreement loss due to failure to meet contractual production requirements can have ramifications that include the risk of bankruptcy.

14 SOLAR ENERGY↗

3D full core BEAVRS using OpenMOC with transport equivalence

Using an optimized implementation of the three-dimension method of characteristics for neutron transport, as well as a novel equivalence method for transport calculations de- signed to recover self-shielding errors from neglecting the angular dependence of reso- nant group absorption, a 3D full core light water reactor hybrid stochastic-deterministic eigenvalue calculation was achieved. This paper presents the optimizations developed and compares the transport solutions obtained. It realizes a runtime near 6,000 CPU hours for the statepoint, an improvement of an order of magnitude on previous works, with less than 1% error on pin fission/U-238 capture ratios and a few dozen pcms on the eigenvalue.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗

Three-Dimensional Full-Core BEAVRS Using OpenMOC with Transport Equivalence

Using an optimized implementation of the three-dimensional (3D) method of characteristics for neutron transport, along with a novel equivalence method for transport calculations that was designed to correct self-shielding errors from neglecting the angular dependence of resonant group absorption, a 3D full-core light water reactor hybrid stochastic-deterministic eigenvalue calculation was achieved. This paper presents the optimizations developed and compares the transport solutions obtained. For the statepoint, run times near 10 000 CPU hours are achieved—improving on previous works by an order of magnitude—with near 1% error on pin fission to 238 U capture ratios and a few dozen pcms on the eigenvalue.

22 GENERAL STUDIES OF NUCLEAR REACTORS↗