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 847 records · Page 47

HPC formulations of optimization algorithms for tensor completion

Tensor completion is a powerful tool used to estimate or recover missing values in multi-way data. It has seen great success in domains such as product recommendation and healthcare. Tensor completion is most often accomplished via low-rank sparse tensor factorization, a computationally expensive non-convex optimization problem which has only recently been studied in the context of parallel computing. In this work, we study three optimization algorithms that have been successfully applied to tensor completion: alternating least squares (ALS), stochastic gradient descent (SGD), and coordinate descent (CCD++). We explore opportunities for parallelism on shared- and distributed-memory systems and address challenges such as memory- and operation-efficiency, load balance, cache locality, and communication. Among our advancements are a communication-efficient CCD++ algorithm, an ALS algorithm rich in level-3 BLAS routines, and an SGD algorithm which combines stratification with asynchronous communication. Furthermore, we show that introducing randomization during ALS and CCD++ can accelerate convergence. We evaluate our parallel formulations on a variety of real datasets on a modern supercomputer and demonstrate speedups through 16384 cores. Further, these improvements reduce time-to-solution from hours to seconds on real-world datasets. We show that after our optimizations, ALS is advantageous on parallel systems of small-to-moderate scale, while both ALS and CCD++ provide the lowest time-to-solution on large-scale distributed systems.

97 MATHEMATICS AND COMPUTING↗

Stochastic Waveform Estimation at the Fundamental Quantum Limit

Although measuring the deterministic waveform of a weak classical force is a well-studied problem, estimating a random waveform, such as the spectral density of a stochastic signal field, is much less well understood despite it being a widespread task at the frontier of experimental physics. State-of-the-art precision sensors of random forces must account for the underlying quantum nature of the measurement but the optimal quantum protocol for interrogating such linear sensors is not known. We derive the fundamental precision limit: the extended-channel quantum Cramér-Rao bound. In the experimentally relevant regime in which losses dominate, we prove that non-Gaussian-state preparation and measurement are required to achieve this fundamental limit and we determine numerically the optimal non-Gaussian protocol. We discuss how this scheme could accelerate searches for signatures of quantum gravity, stochastic gravitational waves, and axionic dark matter.

Axions↗

Quasi-Stochastic Approximation and Off-Policy Reinforcement Learning: Preprint

The Robbins-Monro stochastic approximation algorithm is a foundation of many algorithmic frameworks for reinforcement learning and often an efficient approach to solving (or approximating the solution to) complex optimal control problems. In many cases, however, practitioners are unable to apply these techniques because of an inherent high variance. This paper aims to provide a general foundation for 'quasi-stochastic approximation,' in which all of the processes under consideration are deterministic, much like quasi-Monte Carlo for variance reduction in simulation. The variance reduction can be substantial, subject to tuning of pertinent parameters in the algorithm. This paper introduces a new coupling argument to establish the optimal rate of convergence provided the gain is sufficiently large. These results are established for linear models and tested also in nonideal settings. A major application of these general results is a new class of reinforcement learning algorithms for deterministic state space models. In this setting, the main contribution is a class of algorithms for approximating the value function for a given policy, using a different policy designed to introduce exploration.

24 POWER TRANSMISSION AND DISTRIBUTION↗

WE-H-BRA-03: Development of a Model to Include the Evolution of Resistant Tumor Subpopulations Into the Treatment Optimization Process for Schedules Involving Targeted Agents in Chemoradiation Therapy

Purpose: To develop a model that includes the process of resistance development into the treatment optimization process for schedules that include targeted therapies. Further, to validate the approach using clinical data and to apply the model to assess the optimal induction period with targeted agents before curative treatment with chemo-radiation in stage III lung cancer. Methods: Growth of the tumor and its subpopulations is modeled by Gompertzian growth dynamics, resistance induction as a stochastic process. Chemotherapy induced cell kill is modeled by log-cell kill dynamics, targeted agents similarly but restricted to the sensitive population. Radiation induced cell kill is assumed to follow the linear-quadratic model. The validation patient data consist of a cohort of lung cancer patients treated with tyrosine kinase inhibitors that had longitudinal imaging data available. Results: The resistance induction model was successfully validated using clinical trial data from 49 patients treated with targeted agents. The observed recurrence kinetics, with tumors progressing from 1.4–63 months, result in tumor growth equaling a median volume doubling time of 92 days [34–248] and a median fraction of pre-existing resistance of 0.035 [0–0.22], in agreement with previous clinical studies. The model revealed widely varying optimal time points for the use of curative therapy, reaching from ∼1m to >6m depending on the patient’s growth rate and amount of pre-existing resistance. This demonstrates the importance of patient-specific treatment schedules when targeted agents are incorporated into the treatment. Conclusion: We developed a model including evolutionary dynamics of resistant sub-populations with traditional chemotherapy and radiation cell kill models. Fitting to clinical data yielded patient specific growth rates and resistant fraction in agreement with previous studies. Further application of the model demonstrated how proper timing of chemo-radiation could minimize the probability of resistance, increasing tumor control significantly.

60 APPLIED LIFE SCIENCES↗

Ultra-thin and strong formvar-based membranes with controlled porosity for micro- and nano-scale systems

Here, we present a methodology for developing ultra-thin and strong formvar-based membranes with controlled morphologies. Formvar is a thin hydrophilic and oleophilic polymer inert to most chemicals and resistant to radiation. The formvar-based membranes are viable materials as support structures in micro- and macro-scale systems depending on thinness and porosity control. Tunable sub-micron thick porous membranes with 20%–65% porosity were synthesized by controlling the ratios of formvar, glycerol, and chloroform. This synthesis process does not require complex separation or handling methods and allows for the production of strong, thin, and porous formvar-based membranes. An expansive array of these membrane characterizations including chemical compatibility, mechanical responses, wettability, as well as the mathematical simulations as a function of porosity has been presented. The wide range of chemical compatibility allows for membrane applications in various environments, where other polymers would not be suitable. Our formvar-based membranes were found to have an elastic modulus of 7.8 GPa, a surface free energy of 50 mN m -1 and an average thickness of 125 nm. Stochastic model simulations indicate that formvar with the porosity of ~50% is the optimal membrane formulation, allowing the most material transfer across the membrane while also withstanding the highest simulated pressure loadings before tearing. Development of novel, resilient and versatile membranes with controlled porosity offers a wide range of exciting applications in the fields of nanoscience, microfluidics, and MEMS.

36 MATERIALS SCIENCE↗

Score-based deterministic density sampling

We propose a deterministic sampling framework using Score-Based Transport Modeling for sampling an unnormalized target density π given only its score ∇ log π. Our method approximates the Wasserstein gradient flow on KL($f_t$∥π) by learning the time-varying score ∇ log $f_t$ on the fly using score matching. While having the same marginal distribution as Langevin dynamics, our method produces smooth deterministic trajectories, resulting in monotone noise-free convergence. We prove that our method dissipates relative entropy at the same rate as the exact gradient flow, provided sufficient training. Numerical experiments validate our theoretical findings: our method converges at the optimal rate, has smooth trajectories, and is often more sample efficient than its stochastic counterpart. Experiments on high-dimensional image data show that our method produces high-quality generations in as few as 15 steps and exhibits natural exploratory behavior. The memory and runtime scale linearly in the sample size.

97 MATHEMATICS AND COMPUTING↗

Stretching the limits of forming processes by robust optimization: A demonstrator

Robust design of forming processes using numerical simulations is gaining attention throughout the industry. In this work, it is demonstrated how robust optimization can assist in further stretching the limits of metal forming processes. A deterministic and a robust optimization study are performed, considering a stretch-drawing process of a hemispherical cup product. For the robust optimization study, both the effect of material and process scatter are taken into account. For quantifying the material scatter, samples of 41 coils of a drawing quality forming steel have been collected. The stochastic material behavior is obtained by a hybrid approach, combining mechanical testing and texture analysis, and efficiently implemented in a metamodel based optimization strategy. The deterministic and robust optimization results are subsequently presented and compared, demonstrating an increased process robustness and decreased number of product rejects by application of the robust optimization approach.

36 MATERIALS SCIENCE↗

SU-E-T-136: Assessment of Seasonal Linear Accelerator Output Variations and Associated Impacts

Purpose: Application of process control theory in quality assurance programs promises to allow earlier identification of problems and potentially better quality in delivery than traditional paradigms based primarily on tolerances and action levels. The purpose of this project was to characterize underlying seasonal variations in linear accelerator output that can be used to improve performance or trigger preemptive maintenance. Methods: Review of runtime plots of daily (6 MV) output data acquired using in house ion chamber based devices over three years and for fifteen linear accelerators of varying make and model were evaluated. Shifts in output due to known interventions with the machines were subtracted from the data to model an uncorrected scenario for each linear accelerator. Observable linear trends were also removed from the data prior to evaluation of periodic variations. Results: Runtime plots of output revealed sinusoidal, seasonal variations that were consistent across all units, irrespective of manufacturer, model or age of machine. The average amplitude of the variation was on the order of 1%. Peak and minimum variations were found to correspond to early April and September, respectively. Approximately 48% of output adjustments made over the period examined were potentially avoidable if baseline levels had corresponded to the mean output, rather than to points near a peak or valley. Linear trends were observed for three of the fifteen units, with annual increases in output ranging from 2–3%. Conclusion: Characterization of cyclical seasonal trends allows for better separation of potentially innate accelerator behaviour from other behaviours (e.g. linear trends) that may be better described as true out of control states (i.e. non-stochastic deviations from otherwise expected behavior) and could indicate service requirements. Results also pointed to an optimal setpoint for accelerators such that output of machines is maintained within set tolerances and interventions are required less frequently.

07 ISOTOPE AND RADIATION SOURCES↗

Optimizers for stabilizing likelihood-free inference

A growing number of applications in particle physics and beyond use neural networks as unbinned likelihood ratio estimators applied to real or simulated data. Precision requirements on the inference tasks demand a high-level of stability from these networks, which are affected by the stochastic nature of training. We show how physics concepts can be used to stabilize network training through a physics-inspired optimizer. In particular, the energy conserving descent (ECD) optimization framework uses classical Hamiltonian dynamics on the space of network parameters to reduce the dependence on the initial conditions while also stabilizing the result near the minimum of the loss function. We develop a version of this optimizer known as , which has few free hyperparameters with limited ranges guided by physical reasoning. We apply to representative likelihood-ratio estimation tasks in particle physics and find on average that it out-performs the widely used Adam optimizer. We expect that ECD will be a useful tool for wide array of data-limited problems, where it is computationally expensive to exhaustively optimize hyperparameters and mitigate fluctuations with ensembling.

Monte Carlo methods↗

Quantum optimization algorithms: Energetic implications

Since the dawn of quantum computing (QC), theoretical developments like Shor's algorithm proved the conceptual superiority of QC over traditional computing. However, such quantum supremacy claims are difficult to achieve in practice because of the technical challenges of realizing noiseless qubits. In the near future, QC applications will need to rely on noisy quantum devices that offload part of their work to classical devices. One way to achieve this is by using parameterized quantum circuits in optimization or even in machine learning tasks. The energy requirements of quantum algorithms have not yet been studied extensively. Here in this article, we explore several optimization algorithms using both theoretical insights and numerical experiments to understand their impact on energy consumption. Specifically, we highlight why and how algorithms like quantum natural gradient descent, simultaneous perturbation stochastic approximations or circuit learning methods, are at least 2x to 4x more energy efficient than their classical counterparts; why feedback-based quantum optimization is energy-inefficient; and how techniques like Rosalin can improve the energy efficiency of other algorithms by a factor of ≥2 0 x. Finally, we use the NchooseK high-level programming model to run optimization problems on both gate-based quantum computers and quantum annealers. Empirical data indicate that these optimization problems run faster, have better success rates, and consume less energy on quantum annealers than on their gate-based counterparts.

97 MATHEMATICS AND COMPUTING↗

Optimal experimental design for prediction based on push-forward probability measures

Incorporating experimental data is essential for increasing the credibility of simulation-aided decision making and design. This paper presents a method which uses a computational model to guide the optimal acquisition of experimental data to produce data-informed predictions of quantities of interest (QoI). Many strategies for optimal experimental design (OED) select data that maximize some utility that measures the reduction in uncertainty of uncertain model parameters, for example the expected information gain between prior and posterior distributions of these parameters. In this paper, we seek to maximize the expected information gained from the pushforward of an initial (prior) density to the push-forward of the updated (posterior) density through the parameter-to-prediction map. The formulation presented is based upon the solution of a specific class of stochastic inverse problems which seeks a probability density that is consistent with the model and the data in the sense that the push-forward of this density through the parameter-to-observable map matches a target density on the observable data. While this stochastic inverse problem forms the mathematical basis for our approach, we develop a one-step algorithm, focused on push-forward probability measures, that leverages inference-for-prediction to bypass constructing the solution to the stochastic inverse problem. A number of numerical results are presented to demonstrate the utility of this optimal experimental design for prediction and facilitate comparison of our approach with traditional OED.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Surrogate Modeling for Capacity Planning of Charging Station Equipped With Photovoltaic Panel and Hydropneumatic Energy Storage

Due to the promising potential for environmental sustainability, there has been a significant increase of electric vehicles (EVs) and plug-in hybrid electric vehicles (PHEV) in the market. To support this increasing demand for EVs and PHEVs, challenges related to capacity planning and investment costs of public charging infrastructure must be addressed. Hence, in this paper, a capacity planning problem for charging stations is developed and aims to balance the current capital investment costs and future operational revenue. The charging station is assumed to be equipped with the solar photovoltaic (PV) panel and an energy storage system, which could be electric battery or recently invented hydropneumatic energy storage (ground-level integrated diverse energy storage (GLIDES)) system. A co-optimization model that minimizes investment and operation cost is established to determine optimal solution while considering capacity planning and following operations. EV mobility is modeled as an Erlang-loss system. Meanwhile, stochastic programming is adopted to capture uncertainties from solar radiation and charging demand of EV fleet. To provide a more general and computationally efficient model, main configuration parameters are sampled in design space and then fixed in solving the co-optimization model. Sampled parameters include EV charging slots number, PV area, capacity of energy storage system, and daily mean EV arrival number. Based on the sampled parameter combinations and its responses, black-box mappings are then constructed using surrogate models, which could provide insights for charging station placement in different practical situations. The effectiveness of the proposed surrogate modeling approach is demonstrated in numerical experiments. Here, the results indicate better profit advantage of GLIDES over battery system with the increased power capacity

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

A Binomial Stochastic Framework for Efficiently Modeling Discrete Statistics of Convective Populations

Abstract Understanding the coupling between convective clouds and the general circulation, as well as addressing the gray zone problem in convective parameterization, requires insight into the genesis and maintenance of spatial patterns in cumulus cloud populations. In this study, a simple toy model for recreating populations of interacting convective objects as distributed over a two‐dimensional Eulerian grid is formulated to this purpose. Key elements at the foundation of the model include i) a fully discrete formulation for capturing discrete behavior in convective properties at small population sample sizes, ii) object age‐dependence for representing life‐cycle effects, and iii) a prognostic number budget allowing for object interactions and co‐existence of multiple species. A primary goal is to optimize the computational efficiency of this system. To this purpose the object birth rate is represented stochastically through a spatially aware Bernoulli process. The same binomial stochastic operator is applied to horizontal advection of objects, conserving discreteness in object number. The applicability to atmospheric convection as well as behavior implied by the formulation is assessed. Various simple applications of the BiOMi model (Binomial Objects on Microgrids) are explored, suggesting that important convective behavior can be captured at low computational cost. This includes i) subsampling effects and associated powerlaw scaling in the convective gray zone, ii) stochastic predator‐prey behavior, iii) the downscale turbulent energy cascade, and iv) simple forms of spatial organization and convective memory. Consequences and opportunities for convective parameterization in next‐generation weather and climate models are discussed.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Rolling Optimization of Transmission Network Recovery and Load Restoration Considering Hybrid Wind-Storage System and Cold Load Pickup

A common solution to deal with the stochasticity introduced by fast-ramping wind power integration is to equip wind farms (WFs) with energy storage systems (ESSs) to formulate hybrid WF-ESSs. In addition to leveling off wind power fluctuations during normal operations, a hybrid WF-ESS can be a flexible power source to accumulate the power system restoration. In this paper, we propose a rolling optimization model for transmission network recovery and load restoration considering the contributions of WF-ESSs. The proposed model is formulated as a mixed integer linear programming problem that simultaneously optimizes the amount and location of restorable load blocks as well as the restoration lines. The cold load pickup features of interrupted loads considering the outage duration are modeled in detail. A chance-constrained method is employed to deal with the uncertainty of wind power, and a rolling horizon-based framework is adopted to reduce the influence of forecast error. Case studies are conducted on both New England 39-bus system and part of a provincial power system in China. The results show that the load restoration process can be significantly accelerated by employing the proposed method and contributions of hybrid WF-ESSs to power system restoration are validated.

chance-constrained optimization↗

Random coordinate descent: A simple alternative for optimizing parameterized quantum circuits

Variational quantum algorithms rely on the optimization of parameterized quantum circuits in noisy settings. The commonly used back-propagation procedure in classical machine learning is not directly applicable in this setting due to the collapse of quantum states after measurements. Thus, gradient estimations constitute a significant overhead in a gradient-based optimization of such quantum circuits. This paper introduces a random coordinate descent algorithm as a practical and easy-to-implement alternative to the full gradient descent algorithm. This algorithm only requires one partial derivative at each iteration. Motivated by the behavior of measurement noise in the practical optimization of parameterized quantum circuits, this paper presents an optimization problem setting that is amenable to analysis. Under this setting, the random coordinate descent algorithm exhibits the same level of stochastic stability as the full gradient approach, making it as resilient to noise. The complexity of the random coordinate descent method is generally no worse than that of the gradient descent and can be much better for various quantum optimization problems with anisotropic Lipschitz constants. Theoretical analysis and extensive numerical experiments validate our findings. Published by the American Physical Society 2024

Ding, Zhiyan (ORCID:000000018863403X)↗

Stochastic Dual Algorithm for Voltage Regulation in Distribution Networks with Discrete Loads

This paper considers power distribution networks with distributed energy resources and designs an incentive-based algorithm that allows the network operator and customers to pursue given operational and economic objectives while concurrently ensuring that voltages are within prescribed limits. Heterogeneous DERs with continuous and discrete control commands are considered. We address four major challenges: discrete decision variables, non-convexity due to a Stackelberg game structure, unavailability of private information from customers, and asynchronous operation. Starting from a non-convex setting, we develop a distributed stochastic dual algorithm that solves a relaxed problem, and prove that the proposed algorithm achieves the global optimal solution of the original problem on average. Feasible values for discrete decision variables are also recovered. Stability of the algorithm is analytically established and numerically corroborated.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Optimal carbon capture and storage contracts using historical CO 2 emissions levels

In an effort to reduce carbon dioxide (CO 2 ) emissions to the atmosphere, carbon capture and storage (CCS) technology has been developed to collect CO 2 from emissions generators and store it underground. Recent proposed legislation would limit the volume of emissions generated from power sources, effectively requiring some sources to participate in CCS. Both emissions sources and storage operators require incentives to enter into contracts to capture excess emissions at the source, and transport and store the CO 2 underground. As the level of emissions from power plants is stochastic and carryover into future time periods is expensive, we develop a newsvendor model to determine the optimal price and volume of these contracts to maximize the expected profit of the storage operator and encourage the participation of multiple emissions sources. Because the storage operator has a limit on the amount of CO 2 that can be injected each month, this limit affects the allocation of the optimal contract amounts between the emitters. Furthermore, the distribution of emissions and relative costs of transportation also influence the optimal policy. In addition to analytical solutions, we present data-driven methods for using correlated emissions data to determine the optimal price and volume of these contracts.

42 ENGINEERING↗

Modular supply chain optimization considering demand uncertainty to manage risk

Supply chain under demand uncertainty has been a challenging problem due to increased competition and market volatility in modern markets. Flexibility in planning decisions makes modular manufacturing a promising way to address this problem. We report the problem of multiperiod process and supply chain network design is considered under demand uncertainty. A mixed integer two-stage stochastic programming problem is formulated with integer variables indicating the process design and continuous variables to represent the material flow in the supply chain. The problem is solved using a rolling horizon approach. Benders decomposition is used to reduce the computational complexity of the optimization problem. To promote risk-averse decisions, a downside risk measure is incorporated in the model. The results demonstrate the several advantages of modular designs in meeting product demands. A pareto-optimal curve for minimizing the objectives of expected cost and downside risk is obtained.

42 ENGINEERING↗