SEARCH · Engineering Papers

Results for “optimization methods”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 1,441 records · Page 80

All sources NASA NTRS DOE OSTI

DOE OSTI2015

Geminal embedding scheme for optimal atomic basis set construction in correlated calculations

We introduce an efficient method to construct optimal and system adaptive basis sets for use in electronic structure and quantum Monte Carlo calculations. The method is based on an embedding scheme in which a reference atom is singled out from its environment, while the entire system (atom and environment) is described by a Slater determinant or its antisymmetrized geminal power (AGP) extension. The embedding procedure described here allows for the systematic and consistent contraction of the primitive basis set into geminal embedded orbitals (GEOs), with a dramatic reduction of the number of variational parameters necessary to represent the many-body wave function, for a chosen target accuracy. Within the variational Monte Carlo method, the Slater or AGP part is determined by a variational minimization of the energy of the whole system in presence of a flexible and accurate Jastrow factor, representing most of the dynamical electronic correlation. The resulting GEO basis set opens the way for a fully controlled optimization of many-body wave functions in electronic structure calculation of bulk materials, namely, containing a large number of electrons and atoms. We present applications on the water molecule, the volume collapse transition in cerium, and the high-pressure liquid hydrogen.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

DOE OSTI2022

DRAS: Deep Reinforcement Learning for Cluster Scheduling in High Performance Computing

Cluster schedulers are crucial in high-performance computing (HPC). They determine when and which user jobs should be allocated to available system resources. Existing cluster scheduling heuristics are developed by human experts based on their experience with specific HPC systems and workloads. However, the increasing complexity of computing systems and the highly dynamic nature of application workloads have placed tremendous burden on manually designed and tuned scheduling heuristics. More aggressive optimization and automation are needed for cluster scheduling in HPC. In this work, we present an automated HPC scheduling agent named DRAS (Deep Reinforcement Agent for Scheduling) by leveraging deep reinforcement learning. DRAS is built on a hierarchical neural network incorporating special HPC scheduling features such as resource reservation and backfilling. An efficient training strategy is presented to enable DRAS to rapidly learn the target environment. Once being provided a specific scheduling objective given by the system manager, DRAS automatically learns to improve its policy through interaction with the scheduling environment and dynamically adjusts its policy as workload changes. We implement DRAS into a HPC scheduling platform called CQGym. CQGym provides a common platform allowing users to flexibly evaluate DRAS and other scheduling methods such as heuristic and optimization methods. Furthermore, the experiments using CQGym with different production workloads demonstrate that DRAS outperforms the existing heuristic and optimization approaches by up to 50%.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

Two Dimensional Topology Optimization of Heat Exchangers with the Density and Level-Set Methods

We design heat exchangers using two topology optimization approaches: the density, i.e. volume fraction and level set methods. Our goal is to maximize the heat exchange between two fluids in separate channels while constraining the pressure drop across each channel. The heat exchanger is modeled with a coupled thermal-flow formulation. The flow is governed by an isothermal and incompressible Stokes-Brinkman equation and the heat transfer is governed by a convection-diffusion equation with high Peclet number. We solve one set of Stokes-Brinkman equations per fluid. Each Brinkman term in the flow equation serves to model the other phase as a solid, thereby preventing mixing. We first represent the solid and fluid phases using a volume fraction variable and apply a SIMP-like penalization in the Brinkman term to drive the optimization to a discrete design. The cost and constraint function derivatives are automatically calculated with the library pyadjoint and the optimization is performed by the Method of Moving Asymptotes. In a second optimization formulation, we use the level set approach to define the interface that separates the two fluids. Pyadjoint calculates the shape derivatives of the cost and constraint functions and the Hamilton-Jacobi advects the interface, allowing for topological changes. We present results in two dimensions and discuss the advantages and disadvantages of each approach.

42 ENGINEERING↗

DOE OSTI2024

Optimized Viewing Techniques to Minimize Radiation Damage From X-ray Imaging Systems

X-ray inspection of ball grid arrays (BGAs) is typically performed at one or more viewing angles to examine adhesion sites for errors such as voids, joint cracking, or head-in-pillow. During this inspection process, the cir cuit board assembly is subject to ionizing radiation exposure, which can cause trapped charge within oxide layers of semiconductor devices. Some x-ray machines allow for programmable inspection routines, which could be used to optimize radiation exposure to semiconductor components. Using Monte Carlo methods, x-ray inspection of a BGA was simulated to determine a range of acceptable viewing angles. Dose rates to circuit board components were estimated at each inspection angle to determine the view resulting in optimized radiation exposure. Results showed that for each BGA, the maximum unobstructed viewing times without exceeding a 5 Gy dose limit to a single part ranged from 82 to 94 minutes. Using a radiation cost function method, optimized viewing across all components was found. Here, it was observed that for a consistent dose limit applied to silicon-based components, performing inspection with BGAs facing the x-ray source was optimal. A third method was applied, assigning individual dose limits based on empirical data from the NASA Goddard Space Flight Center radiation database. This method showed that optimized viewing maximizes the distance between the radiation source and highly sensitive components. It was also observed that cumulative effects from viewing two BGAs will influence viewing angles, causing the optimal view of one BGA to exist nearly 180° from the other.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

DOE OSTI2014

TU-C-17A-12: Towards a Passively Optimized Phase-Space Monte Carlo (POPMC) Treatment Planning Method: A Proof of Principle

Purpose: The advent of on-line/off-line adaptive, and biologically-conformal radiation therapy has led to a need for treatment planning solutions that utilize voxel-specific penalties, requiring optimization over a large solution space that is performed quickly, and the dose in each voxel calculated accurately. This work proposes a “passive” optimization framework, which is executed concurrently during Monte Carlo dose calculation, evaluating the cost/benefit of each history during transport, and creates a passively optimized fluence map. Methods: The Monte Carlo code Geant4 v9.6 was used for this study. The standard voxel geometry implementation was modified to support the passive optimization framework, with voxel-specific optimization parameters. Dose-benefit functions, which will increase a particle history’s weight upon dose deposition, were defined in a central collection of voxels to effectively create target structures. Histories that deposit energy to voxels are reweighted based on a voxel’s dose multiplied by its cost/benefit value. Upon full termination of each history, the dose contributions of that history are reweighted to reflect a contribution proportional to the history’s final weight. A parallel-planar 1.25 MeV photon fluence is transported through the geometry, and re-weighted at each dose deposition step. The resulting weight is tallied with the incident spatial and directional coordinates in a phase-space distribution. Results: A uniform incident fluence was reweighted during MC dose calculations to create an optimized fluence map which would generate dose profiles in target volumes that exhibit the same dose characteristics as the prescribed optimization parameters. An optimized dose profile, calculated concurrently with the phase-space, reflects the resulting dose distribution. Conclusion: This study demonstrated the feasibility of passively optimizing an incident fluence map during Monte Carlo dose calculations. The flexibility of the voxel-specific optimization framework allows a variety and combination of optimization parameters to be calculated for each voxel at every transportation step. This work is partially supported by Varian. This work is partially supported by the NIH.

07 ISOTOPE AND RADIATION SOURCES↗

DOE OSTI2015

CALIBRATION OF SEMI-ANALYTIC MODELS OF GALAXY FORMATION USING PARTICLE SWARM OPTIMIZATION

We present a fast and accurate method to select an optimal set of parameters in semi-analytic models of galaxy formation and evolution (SAMs). Our approach compares the results of a model against a set of observables applying a stochastic technique called Particle Swarm Optimization (PSO), a self-learning algorithm for localizing regions of maximum likelihood in multidimensional spaces that outperforms traditional sampling methods in terms of computational cost. We apply the PSO technique to the SAG semi-analytic model combined with merger trees extracted from a standard Lambda Cold Dark Matter N-body simulation. The calibration is performed using a combination of observed galaxy properties as constraints, including the local stellar mass function and the black hole to bulge mass relation. We test the ability of the PSO algorithm to find the best set of free parameters of the model by comparing the results with those obtained using a MCMC exploration. Both methods find the same maximum likelihood region, however, the PSO method requires one order of magnitude fewer evaluations. This new approach allows a fast estimation of the best-fitting parameter set in multidimensional spaces, providing a practical tool to test the consequences of including other astrophysical processes in SAMs.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2018

Parallel multiple-chain DRAM MCMC for large-scale geosteering inversion and uncertainty quantification

Geosteering is the proactive control of a wellbore placement based on the downhole measurements, aiming at maximizing economic production from the well. We report the recent development of azimuthal resistivity logging-whiledrilling (LWD) tools have a much larger depth of detection, thus sets higher demands for geosteering inversion as more unknown parameters need to be taken into account in the earth model to be inverted. For complicated nonlinear problems, traditional deterministic inversion methods are more likely to be trapped near local optima. In general, Markov chain Monte Carlo (MCMC) inversion methods are more capable of finding the global optimal solution and providing additional uncertainty analyses by sampling from the target distribution. However, MCMC methods usually have the problem of slow convergence. Even though optimization methods like delayed rejection (DR) and adaptive Metropolis (AM) were proposed to speed up the convergence, using MCMC methods to solve geosteering inversion problems may still incur unacceptable time cost. To reduce the sampling cost of MCMC methods, we use parallel multiple-chain DRAM MCMC methods to solve geosteering inverse problems and estimate the corresponding uncertainty. A clustering method, density-based spatial clustering of applications with noise (DBSCAN), is applied to select the best solution among many results. Lastly, the simulation results show that running many relatively short MCMC chains for one problem can obtain a similar result as running a long single chain. Besides, avoiding communications between multiple Markov chains during the sampling process yields almost linear scalability.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2023

Sparsity for Gradient-Based Optimization of Wind Farm Layouts

Optimizing wind farm layouts is an important step in designing an efficient wind farm. Optimizing wind farm layouts is also a difficult task due to computation times increasing with the number of turbines present in the farm. The most computationally expensive part of gradient-based optimization is calculating the gradient. In order to reduce the expense of gradient calculation, we performed a study on the use of sparsity in wind farm layout optimization. This paper presents the findings of the sparsity study and provides a method to use sparsity in wind farm layout optimization. We tested this sparsity method by optimizing multiple farms with sparse methods and compared the results to optimizations of the same farms using traditional methods. By using the sparse method to optimize multiple farms, we found that the objective results were comparable between sparse and traditional methods and that sparse methods were 4 times faster than traditional methods on average. We expect more speedups with improved methodology and larger wind farms. By using sparse methods, it is possible to solve the wind farm layout optimization problem more efficiently, thus allowing for a more thorough study of the wind farm layout design space without excessive computational costs. Further work is required to refine the method and prepare for testing on real-world wind farm layout applications.

gradient↗

DOE OSTI2024

Terrain-Relative Navigation with Neuro-Inspired Elevation Encoding

Terrain-relative navigation (TRN) encompasses a wide variety of algorithms that perform localization with respect to the terrain below a flying vehicle. In traditional approaches, measurements of the terrain are matched to a map carried onboard. This work presents a terrain-relative navigation filter with a position measurement inspired by neural activity associated with positioning in nature. Here, the filter is shown to produce accurate position measurements that outperform popular optimization and template matching methods given poor prior knowledge of the position. The proposed method is also better-suited to distributed implementation than optimization-based methods.

33 ADVANCED PROPULSION SYSTEMS↗

DOE OSTI2018

Survey of Multifidelity Methods in Uncertainty Propagation, Inference, and Optimization

In many situations across computational science and engineering, multiple computational models are available that describe a system of interest. These different models have varying evaluation costs and varying fidelities. Typically, a computationally expensive high-fidelity model describes the system with the accuracy required by the current application at hand, while lower-fidelity models are less accurate but computationally cheaper than the high-fidelity model. Outer-loop applications, such as optimization, inference, and uncertainty quantification, require multiple model evaluations at many different inputs, which often leads to computational demands that exceed available resources if only the high-fidelity model is used. This work surveys multifidelity methods that accelerate the solution of outer-loop applications by combining high-fidelity and low-fidelity model evaluations, where the low-fidelity evaluations arise from an explicit low-fidelity model (e.g., a simplified physics approximation, a reduced model, a data-fit surrogate) that approximates the same output quantity as the high-fidelity model. The overall premise of these multifidelity methods is that low-fidelity models are leveraged for speedup while the high-fidelity model is kept in the loop to establish accuracy and/or convergence guarantees. We categorize multifidelity methods according to three classes of strategies: adaptation, fusion, and filtering. The paper reviews multifidelity methods in the outer-loop contexts of uncertainty propagation, inference, and optimization.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2019

Quantum Optimal Control Using High Performance Computing

This feasibility study developed numerical methods for the optimal control problem for realizing logical gates in closed quantum systems, where the evolution of the state vector is governed by the time dependent Schrodinger equation. The number of parameters in the control functions is made independent of the number of time steps by expanding them in terms of B-spline basis functions, with and without carrier waves. We use an interior point gradient-based technique from the IPOPT package to minimize the gate infidelity subject to amplitude constraints on the control functions. The symplectic Stromer-Verlet scheme is used to integrate a real-valued formulation of Schrodinger's equation in time and the gradient of the gate infidelity is obtained by solving the corresponding adjoint equation. This allows all components of the gradient to be calculated at the cost of solving three Schrodinger systems, independently of the number of parameters in the control functions. The correctness of the method is verified and applied to Hamiltonians that model the dynamics of coupled super-conducting qubits.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

DOE OSTI2018

An efficient algorithm for the projection of a point on the intersection of two hyperplanes and a box in $\mathbb {R}^n$

Here, we present an efficient strongly polynomial algorithm for the projection of a point on the intersection of two hyperplanes and a box in $\mathbb {R}^n$ . Interior point methods are the most efficient algorithm in the literature to solve this problem. While efficient in practice, the complexity of interior-point methods is bounded by a polynomial in the dimension of the problem and in the accuracy of the solution. Moreover, their efficiency is highly dependent on a series of parameters depending on the specific method chosen (especially for nonlinear problems), such as step size, barrier parameter, accuracy, among others. We propose a new method based on the KKT optimality conditions. In this method, we write the problem as a function of the Lagrangian multipliers of the hyperplanes and seek to find the pair of multipliers that corresponds to the optimal solution. We prove that the algorithm has complexity $O(n^2 \log n)$ .

42 ENGINEERING↗

DOE OSTI2019

Exact Optimal Power Dispatch in Unbalanced Distribution Systems With High PV Penetration

Smart inverters provide additional control capability to help optimize the operation of distribution systems. This paper proposes a framework for exact optimal active and reactive power dispatch of distributed photovoltaic (PV) generation, switched capacitors, and tap changers in large multi-phase unbalanced distribution systems. The objectives of the optimal dispatch are minimization of the energy loss, PV active power curtailment, and capacitor and tap changer switching operations, in addition to elimination of voltage violations and reverse power flow. The optimization problem is formulated in rectangular coordinates as a nonlinear, nonconvex problem. Effective computational strategies are proposed to allow the application of predictorcorrector primal-dual interior point method to solve optimization problems in real-time with a large number of constraints and variables, including discrete variables corresponding to switched capacitors and tap changers. The accuracy of the numerical solution and the ability to implement the proposed framework are validated using the unbalanced multi-phase IEEE 34-bus and EPRI 2,998-bus distribution systems with 15-minute load and PV data. The results show a significant loss reduction and complete removal of both voltage violations and reverse power flow.

Nguyen, Quan (ORCID:0000000347769557)↗

DOE OSTI2025

Scientific frontiers of agrivoltaic cropping systems

Agrivoltaic (AV) systems integrate agriculture with electricity conversion through photovoltaic (PV) modules. Compared with conventional ground-mounted PV systems, AV systems can reduce land-use competition and offer agronomic and economic advantages, such as more stable crop production and additional farm income. However, AV systems can decrease agricultural performance and are typically 20-90% costlier to install than conventional PV systems. Here, in this Review, we analyse the implementation of AV cropping systems to preserve agricultural activities and highlight challenges and barriers. The global electricity potential of AV systems is ~66-385 PWh annually, depending on PV technology and installation density, if deployed in the most suitable areas, without accounting for grid availability. Scaling up has been hindered by crop selection for shading conditions, decreased energy conversion per unit of land area and issues with social acceptance, landscape impact and environmental sustainability. These issues can be addressed by developments such as wavelength-selective PV; system configurations, such as optimizing module spacing to reduce shading; and operational methods, such as optimizing tracking strategies and integrating agricultural infrastructure. Cross-sector policies can support AV systems by addressing the needs of diverse stakeholders over shared land resources. Further development will require collaboration among the design, performance, deployment and systems research communities.

14 SOLAR ENERGY↗

DOE OSTI2023

Equidistant optimization of elliptical superconducting rf standing wave cavities

A record accelerating rate was achieved earlier in standing wave (SW) SRF cavities when their shape was optimized for a lower peak surface magnetic field sacrificing the peak surface electric field. In view of new materials with higher limiting magnetic fields, expected for SRF cavities, in the first line of the Nb 3 Sn, the approach to optimization of cavity shape should be revised. A method of equidistant optimization, offered earlier for traveling wave (TW) cavities is applied to SW cavities. It is shown here that without limitation by magnetic field, the maximal accelerating rate is defined not only by limitations of the electric field but to a significant degree by the cavity shape. For example, for a cavity with the aperture radius of $R$ $a$ = 35 mm, the minimal ratio of the peak surface electric field to the accelerating rate is about $E$ pk / $E$ acc = 1.54. So, with the maximal surface field experimentally achieved $E$ pk ~ 125 MV / m, the maximal achievable accelerating rate is about 80 MeV / m even if there are no restrictions by the magnetic field. Optimized cavity shapes with and without limitations by a magnetic field are presented. Another opportunity—optimization for a low magnetic field, is opening for the same material, Nb 3 Sn, with the purpose of having a high-quality factor and increased accelerating rate that can be used for industrial linacs with cryocooler-based cooling scheme.

43 PARTICLE ACCELERATORS↗

DOE OSTI2021

Randomized Sketching Algorithms for Low-Memory Dynamic Optimization

This paper develops a novel limited-memory method to solve dynamic optimization problems. The memory requirements for such problems often present a major obstacle, particularly for problems with PDE constraints such as optimal flow control, full waveform inversion, and optical tomography. In these problems, PDE constraints uniquely determine the state of a physical system for a given control; the goal is to find the value of the control that minimizes an objective. While the control is often low dimensional, the state is typically more expensive to store. This paper suggests using randomized matrix approximation to compress the state as it is generated and shows how to use the compressed state to reliably solve the original dynamic optimization problem. Concretely, the compressed state is used to compute approximate gradients and to apply the Hessian to vectors. The approximation error in these quantities is controlled by the target rank of the sketch. This approximate first- and second-order information can readily be used in any optimization algorithm. As an example, we develop a sketched trust-region method that adaptively chooses the target rank using a posteriori error information and provably converges to a stationary point of the original problem. Numerical experiments with the sketched trust-region method show promising performance on challenging problems such as the optimal control of an advection-reaction-diffusion equation and the optimal control of fluid flow past a cylinder.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2024

Integration of Decentralized Graph-Based Multi-Agent Reinforcement Learning with Digital Twin for Traffic Signal Optimization

Machine learning (ML) methods, particularly Reinforcement Learning (RL), have gained widespread attention for optimizing traffic signal control in intelligent transportation systems. However, existing ML approaches often exhibit limitations in scalability and adaptability, particularly within large traffic networks. This paper introduces an innovative solution by integrating decentralized graph-based multi-agent reinforcement learning (DGMARL) with a Digital Twin to enhance traffic signal optimization, targeting the reduction of traffic congestion and network-wide fuel consumption associated with vehicle stops and stop delays. In this approach, DGMARL agents are employed to learn traffic state patterns and make informed decisions regarding traffic signal control. The integration with a Digital Twin module further facilitates this process by simulating and replicating the real-time asymmetric traffic behaviors of a complex traffic network. The evaluation of this proposed methodology utilized PTV-Vissim, a traffic simulation software, which also serves as the simulation engine for the Digital Twin. The study focused on the Martin Luther King (MLK) Smart Corridor in Chattanooga, Tennessee, USA, by considering symmetric and asymmetric road layouts and traffic conditions. Comparative analysis against an actuated signal control baseline approach revealed significant improvements. Experiment results demonstrate a remarkable 55.38% reduction in Eco_PI, a developed performance measure capturing the cumulative impact of stops and penalized stop delays on fuel consumption, over a 24 h scenario. In a PM-peak-hour scenario, the average reduction in Eco_PI reached 38.94%, indicating the substantial improvement achieved in optimizing traffic flow and reducing fuel consumption during high-demand periods. These findings underscore the effectiveness of the integrated DGMARL and Digital Twin approach in optimizing traffic signals, contributing to a more sustainable and efficient traffic management system.

42 ENGINEERING↗

DOE OSTI2024

Integration of Decentralized Graph-Based Multi-Agent Reinforcement Learning with Digital Twin for Traffic Signal Optimization

actuated signal control↗