Search NASA⌕ Search

SEARCH · Search NASA

Results for “Computational 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 1,675 records · Page 93

Breakup dynamics in a pressure-swirl injector for urea-water solution applications: A computational study

The co-optimization of in-cylinder combustion and after-treatment technology has become a major aspect in engine design and development, with the goal of meeting the increasingly restrictive emission regulations in the transportation industry. Selective Catalytic Reduction is a robust technology to control the emission of NO x , and the injection of urea in water solution is the exhaust tailpipe is a key aspect of its operation. The proposed work uses high-fidelity Computational Fluid Dynamics to characterize the atomization dynamics of the liquid jet in relevant cross-flow conditions. The study focuses on a commercial low-pressure (9 bar) pressure-swirl injector which is characterized in its internal geometry through high-resolution X-ray micro-computational tomography. The internal two-phase flow has been modeled according to the volume-of-fluid approach in a large eddy simulation framework and validated against near-nozzle X-ray radiography measurement. Moreover, characterizing the breakup dynamics for the swirling hollow cone formation, and assessing the influence of the cross-flow in the breakup dynamics was completed. The results have been reported proposing Re-Oh maps and probability density functions of the spray kinematics. Higher cross-flow momentum generates an increase in the jet intact length and a reduction of the liquid droplet diameters. The axial momentum of the jet is affected by the cross-flow already in the near-nozzle region, determining a relevant deviation of the spray velocities. In conclusion, this work aims to inform the initialization of Eulerian-Lagrangian spray models through the assignment of droplet kinematics and static one-way coupling between volume-of-fluid results and Lagrangian spray parcels, to be used for system-size domain simulations.

33 ADVANCED PROPULSION SYSTEMS↗

Development and validation of a second-order thermal network model for residential buildings

Heating, Ventilation, and Air Conditioning (HVAC) systems can maintain the space air temperature of residential buildings, either directly by heating/cooling the air, or indirectly via heat transfer to and from the building structure that acts as a thermal mass. Hence, HVAC systems can help achieve load shifting, peak load reduction, and/or energy cost saving, thus enabling grid-interactive HVAC operation. A home thermal model that can accurately reflect the dynamics of the space air and interior wall surface temperatures, is therefore valuable. This paper develops such a model using the standard RC (resistance-capacitance) approach. The model contains a virtual envelope node and an internal space node and is thus second-order. A hybrid parameter identification scheme, made up of the least-squares and optimal search methods, is also developed. The proposed model and scheme were validated using data collected from a test home. It was found that a modest amount of training data was sufficient to yield reliable parameter estimates and accurate prediction. It was also found that when making 24-hour-ahead prediction of the space air temperature, both methods had comparable performances when the training data began in a transition season. However, when they began in an HVAC season, the optimal search method performed better. Furthermore, the least-squares method is recommended during a transition season due to its lower computational burden, while the optimal search method is recommended during an HVAC season due to its better estimation performance.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Advanced modeling of $\mathrm{DER}$s and $\mathrm{UPFC}$ devices with Sparse Tableau Formulation for $\mathrm{ACOPF}$

Distributed energy resources (DERs) are projected to be significant contributors in electric power grid modernization efforts. Yet many distribution and transmission system operators are concerned about potential reliability and security impacts of high penetration of DERs, and cite the need of greater visibility and controllability of decision variables associated with these DERs. Tractable modeling of DERs in AC optimal power flow (ACOPF) is one of the barriers to their integration into the existing operation, planning and market paradigm. Similarly, as more intermittent renewable resources are deployed, power flow control becomes more valuable, and accurate representation of Flexible AC Transmission System (FACTS) devices in ACOPF will likely grow in importance. Here this paper focuses on the detailed representation of substations with the high penetration of DERs, and on the Unified Power Flow Controller (UPFC) FACTS devices, in OPF. To this end, a novel Sparse Tableau Formulation (STF) for ACOPF is used to accurately model such features as substation reconfiguration to accommodate DERs, and to represent decision variables appropriate to the voltage source converters within UPFCs. Numerical studies are provided to demonstrate the impacts of detailed DER and UPFC representations on optimal operation, and the computational efficiency.

42 ENGINEERING↗

Extended Lagrangian Born-Oppenheimer molecular dynamics in the limit of vanishing self-consistent field optimization

We present an efficient general approach to first principles molecular dynamics simulations based on extended Lagrangian Born-Oppenheimer molecular dynamics [A. M. N. Niklasson, Phys. Rev. Lett. 100, 123004 (2008)] in the limit of vanishing self-consistent field optimization. The reduction of the optimization requirement reduces the computational cost to a minimum, but without causing any significant loss of accuracy or long-term energy drift. The optimization-free first principles molecular dynamics requires only one single diagonalization per time step, but is still able to provide trajectories at the same level of accuracy as “exact,” fully converged, Born-Oppenheimer molecular dynamics simulations. The optimization-free limit of extended Lagrangian Born-Oppenheimer molecular dynamics therefore represents an ideal starting point for robust and efficient first principles quantum mechanical molecular dynamics simulations.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

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↗

Development of array-type prompt gamma measurement system for in vivo range verification in proton therapy

Purpose: In vivo range verification is one of the most important parts of proton therapy to fully utilize its benefits delivering high radiation dose to tumor, while sparing the normal tissue with the so-called Bragg peak. Currently, however, range verification method is not used in clinics. The purpose of the present study is to optimize and evaluate the configuration of an array-type prompt gamma measurement system on determining distal dose edge for in vivo range verification of proton therapy. Methods: To effectively measure the prompt gammas against the background gammas, the Monte Carlo simulations with the MCNPX code were employed in optimizing the configuration of the measurement system, and the Monte Carlo method was also used to understand the effect of the background gammas, mainly neutron capture gammas, in the measured gamma distribution. To reduce the effect of the background gammas, the optimized energy window of 4-10 MeV in measuring the prompt gammas was employed. A parameterized source was used to maximize computation speed in the optimization study. A simplified test measurement system, using only one detector moving from one measurement location to the next, was constructed and applied to therapeutic proton beams of 80-220 MeV. For accurate determination of the distal dose edge, the sigmoidal curve-fitting method was applied to the measured distributions of the prompt gammas, and then, the location of the half-value between the maximum and minimum value in the curve-fitting was determined as the distal dose edge and compared with the beam range assessed by the proton dose distribution. Results: The parameterized source term employed in optimization process improved the calculation speed by up to {approx}300 times. The optimization study indicates that an array-type measurement system with 3, 2, 2, and 150 mm for scintillator thickness, slit width, septal thickness, and slit length, respectively, can effectively measure the prompt gamma distributions minimizing the contribution of background gammas. The present results show that a few hundred counts of prompt gammas can be easily obtained by measuring 10 s at each measurement location for proton beams of {approx}4 nA. The distal dose edges determined by the prompt gamma distribution are 5.45, 14.73, and 27.74 cm for proton beams of 5.17 (80 MeV), 14.99 (150 MeV), and 27.38 (220 MeV) cm, respectively. Conclusions: The results show that the array-type measurement system can measure prompt gamma distributions from a therapeutic proton beam within a short measurement time, and that the distal dose edge can be determined within a few millimeters of error without using any sophisticated analysis.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

The Benefits and Weaknesses of Containerizing Software for HPC

Containerization technology has emerged as a transformative tool for software engineers, offering consistent development and deployment environments, simplifying dependency management, and enhancing scalability and portability across diverse systems. However, its application in High-Performance Computing (HPC) presents unique challenges, including the management of virtualization overhead, the need for efficient resource allocation, and the maintenance of optimal performance for compute-intensiv

Ho, Eric Victor [Sandia National Laboratories (SNL↗

A differentiable approach to the maximum independent set problem using dataless neural networks

The success of machine learning solutions for reasoning about discrete structures has brought attention to its adoption within combinatorial optimization algorithms. Such approaches generally rely on supervised learning by leveraging datasets of the combinatorial structures of interest drawn from some distribution of problem instances. Reinforcement learning has also been employed to find such structures. Here, in this paper, we propose a different approach in that no data is required for training the neural networks that produce the solution. In this sense, what we present is not a machine learning solution, but rather one that is dependent on neural networks and where backpropagation is applied to a loss function defined by the structure of the neural network architecture as opposed to a training dataset. In particular, we reduce the popular combinatorial optimization problem of finding a maximum independent set to a neural network and employ a dataless training scheme to refine the parameters of the network such that those parameters yield the structure of interest. Additionally, we propose a universal graph reduction procedure to handle large-scale graphs. The reduction exploits community detection for graph partitioning and is applicable to any graph type and/or density. Experimental results on both real and synthetic graphs demonstrate that our proposed method performs on par or outperforms state-of-the-art learning-based methods in terms of the size of the found set without requiring any training data.

97 MATHEMATICS AND COMPUTING↗

An approach to learn regulation to maximize growth and entropy production rates in metabolism

Elucidating cell regulation remains a challenging task due to the complexity of metabolism and the difficulty of experimental measurements. Here we present a method for prediction of cell regulation to maximize cell growth rate while maintaining the solvent capacity of the cell. Prediction is formulated as an optimization problem using a thermodynamic framework that can leverage experimental data. We develop a formulation and variable initialization procedure that allows for computing solutions of the optimization with an interior point method. The approach is applied to photoheterotrophic growth of Rhodospirilium rubrum using ethanol as a carbon source, which has applications to biosynthesis of ethylene production. Growth is captured as the rate of synthesis of amino acids into proteins, and synthesis of nucleotide triphoshaptes into RNA and DNA. The method predicts regulation that produces a high rate of protein and RNA synthesis while DNA synthesis is reduced close to zero in agreement with production of DNA being turned off for much of the cell cycle.

59 BASIC BIOLOGICAL SCIENCES↗

Intelligent System Partitioning for Agent-Based Security Constrained Optimal Power Flow

This project developed scalable, computationally efficient algorithms to solve realistic large-scale power system optimization problems as part of a larger series of competitions run by ARPA-E. These problems are important because the secure and reliable operation of the power grid, especially under increased uncertainty and variability, is growing increasingly challenging. The economic feasibility of the proposed methods developed by our team is quite low, considering it’s a purely software-based solution to operate power grids more efficiently. The technical effectiveness, as evidenced by our performance in the competition, balances heuristics and approximations to provide a tradeoff between speed and accuracy.

24 POWER TRANSMISSION AND DISTRIBUTION↗

Robustified time-optimal control of uncertain structural dynamic systems

A new approach for computing open-loop time-optimal control inputs for uncertain linear dynamical systems is developed. In particular, the single-axis, rest-to-rest maneuvering problem of flexible spacecraft in the presence of uncertainty in model parameters is considered. Robustified time-optimal control inputs are obtained by solving a parameter optimization problem subject to robustness constraints. A simple dynamical system with a rigid-body mode and one flexible mode is used to illustrate the concept.

Liu, Qiang↗

pnnl/Pathway_Controlled_Optimization

A method for prediction of cell regulation to maximize cell growth rate while maintaining the solvent capacity of the cell. Prediction is formulated as an optimization problem using a thermodynamic framework that can leverage experimental data. We develop a formulation and variable initialization procedure that allows for computing solutions of the optimization with an interior point method

Central, PNNL Developer↗

DeepHyper: A Python Package for Massively Parallel Hyperparameter Optimization in Machine Learning

Machine learning models are increasingly applied across scientific disciplines, yet their effectiveness often hinges on heuristic decisions—such as data transformations, training strategies, and model architectures—that are not learned by the models themselves. Automating the selection of these heuristics and analyzing their sensitivity is crucial for building robust and efficient learning workflows. DeepHyper addresses this challenge by democratizing hyperparameter optimization, providing accessible tools to streamline and enhance machine learning workflows from a laptop to the largest supercomputer in the world. Building on top of hyperparameter optimization, it unlocks new capabilities around ensembles of models for improved accuracy and uncertainty quantification. All of these organized around efficient parallel computing.

ensemble↗

Numerical study of sound propagation in a jet flow

An improved computer oriented solution method for problems involving the propagation of sound through a nonuniform jet flow is developed. The method seeks to optimize the use of computer resources such as core storage space and central memory time. Complete formulation details are presented for a jet flow model consisting of a fixed point source on the jet center line in the potential core.

Padula, S. L.↗

Automated divertor target design by adjoint shape sensitivity analysis and a one-shot method

As magnetic confinement fusion progresses towards the development of first reactor-scale devices, computational tokamak divertor design is a topic of high priority. Presently, edge plasma codes are used in a forward approach, where magnetic field and divertor geometry are manually adjusted to meet design requirements. Due to the complex edge plasma flows and large number of design variables, this method is computationally very demanding. On the other hand, efficient optimization-based design strategies have been developed in computational aerodynamics and fluid mechanics. Such an optimization approach to divertor target shape design is elaborated in the present paper. A general formulation of the design problems is given, and conditions characterizing the optimal designs are formulated. Using a continuous adjoint framework, design sensitivities can be computed at a cost of only two edge plasma simulations, independent of the number of design variables. Furthermore, by using a one-shot method the entire optimization problem can be solved at an equivalent cost of only a few forward simulations. The methodology is applied to target shape design for uniform power load, in simplified edge plasma geometry.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

HyperSpace: Distributed Bayesian Hyperparameter Optimization

As machine learning models continue to increase in complexity, so does the potential number of free model parameters commonly known as hyperparameters. While there has been considerable progress toward finding optimal configurations of these hyperparameters, many optimization procedures are treated as black boxes. We believe optimization methods should not only return a set of optimized hyperparameters, but also give insight into the effects of model hyperparameter settings. To this end, we present HyperSpace, a parallel implementation of Bayesian sequential model-based optimization. HyperSpace leverages high performance computing (HPC) resources to better understand unknown, potentially non-convex hyperparameter search spaces. We show that it is possible to learn the dependencies between model hyperparameters through the optimization process. By partitioning large search spaces and running many optimization procedures in parallel, we also show that it is possible to discover families of good hyperparameter settings over a variety of models including unsupervised clustering, regression, and classification tasks.

Computer Science↗

Accelerating Radiation Computations for Dynamical Models With Targeted Machine Learning and Code Optimization

Abstract Atmospheric radiation is the main driver of weather and climate, yet due to a complicated absorption spectrum, the precise treatment of radiative transfer in numerical weather and climate models is computationally unfeasible. Radiation parameterizations need to maximize computational efficiency as well as accuracy, and for predicting the future climate many greenhouse gases need to be included. In this work, neural networks (NNs) were developed to replace the gas optics computations in a modern radiation scheme (RTE+RRTMGP) by using carefully constructed models and training data. The NNs, implemented in Fortran and utilizing BLAS for batched inference, are faster by a factor of 1–6, depending on the software and hardware platforms. We combined the accelerated gas optics with a refactored radiative transfer solver, resulting in clear‐sky longwave (shortwave) fluxes being 3.5 (1.8) faster to compute on an Intel platform. The accuracy, evaluated with benchmark line‐by‐line computations across a large range of atmospheric conditions, is very similar to the original scheme with errors in heating rates and top‐of‐atmosphere radiative forcings typically below 0.1 K day −1 and 0.5 W m −2 , respectively. These results show that targeted machine learning, code restructuring techniques, and the use of numerical libraries can yield material gains in efficiency while retaining accuracy.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Viscous Aerodynamic Shape Optimization with Installed Propulsion Effects

Aerodynamic shape optimization is demonstrated to tailor the under-track pressure signature of a conceptual low-boom supersonic aircraft. Primarily, the optimization reduces nearfield pressure waveforms induced by propulsion integration effects. For computational efficiency, gradient-based optimization is used and coupled to the discrete adjoint formulation of the Reynolds-averaged Navier Stokes equations. The engine outer nacelle, nozzle, and vertical tail fairing are axi-symmetrically parameterized, while the horizontal tail is shaped using a wing-based parameterization. Overall, 48 design variables are coupled to the geometry and used to deform the outer mold line. During the design process, an inequality drag constraint is enforced to avoid major compromise in aerodynamic performance. Linear elastic mesh morphing is used to deform volume grids between design iterations. The optimization is performed at Mach 1.6 cruise, assuming standard day altitude conditions at 51,707-ft. To reduce uncertainty, a coupled thermodynamic engine cycle model is employed that captures installed inlet performance effects on engine operation.

computational fluid dynamics↗