Search NASA⌕ Search

SEARCH · Search NASA

Results for “gradient based 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 667 records · Page 37
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

torch-einshard v1.0

torch-einshard is a Python library for describing local and distributed PyTorch tensor computations with compact, einsum-like notation. Its expressions name logical axes, specify how they are sharded across a PyTorch DeviceMesh, and represent partial reductions. The library automatically performs contractions, permutations, reshaping, splitting, gathering, reduction, reduce-scatter, and repartitioning while preserving autograd. Additional features include sharding-aware FFTs, tensor rolls, halo exchange, sliding windows, 1D–3D convolutions, uneven-shard handling, parameter initialization and gradient management, and cost-based execution planning. It is designed for scientific machine learning and large-model workloads, including tensor-, sequence-, and spatial-parallel MLPs, attention, convolutions, and spectral operations. Compared with manually combining torch.einsum and distributed collectives, torch-einshard expresses both the mathematical operation and data placement in one readable formula. This reduces boilerplate and synchronization errors, keeps forward and backward communication consistent, and allows the library to select optimized collective strategies without changing model code.

Morozov, Dmitriy [Lawrence Berkeley National Labor↗
NASA NTRS1984

Efficient calculation of optimum design sensitivity

Two methods for obtaining the sensitivity of an optimized design to changes in problem parameters are presented. The first-order method uses gradient information to estimate the required sensitivity. The second-order method uses second derivatives of the design objective and constraint functions to provide a quadratic approximation to the new design which results from changing the specified parameter. It is shown that the sensitivity of the optimized design may be discontinuous with respect to the new parameter. This discontinuity is accounted for in the present methods. These methods are compared to a previous approach based on the Kuhn-Tucker necessary conditions for optimality. It is shown that the present methods provide the appropriate sensitivity information in an efficient manner. The methods are demonstrated by a structural synthesis example.

Vanderplaats, G. N.↗
NASA NTRS1988

On conjugate gradient type methods and polynomial preconditioners for a class of complex non-Hermitian matrices

Conjugate gradient type methods are considered for the solution of large linear systems Ax = b with complex coefficient matrices of the type A = T + i(sigma)I where T is Hermitian and sigma, a real scalar. Three different conjugate gradient type approaches with iterates defined by a minimal residual property, a Galerkin type condition, and an Euclidian error minimization, respectively, are investigated. In particular, numerically stable implementations based on the ideas behind Paige and Saunder's SYMMLQ and MINRES for real symmetric matrices are proposed. Error bounds for all three methods are derived. It is shown how the special shift structure of A can be preserved by using polynomial preconditioning. Results on the optimal choice of the polynomial preconditioner are given. Also, some numerical experiments for matrices arising from finite difference approximations to the complex Helmholtz equation are reported.

Freund, Roland↗
NASA NTRS1990

Trajectory optimization for an asymmetric launch vehicle

A numerical optimization technique is used to fully automate the trajectory design process for an symmetric configuration of the proposed Advanced Launch System (ALS). The objective of the ALS trajectory design process is the maximization of the vehicle mass when it reaches the desired orbit. The trajectories used were based on a simple shape that could be described by a small set of parameters. The use of a simple trajectory model can significantly reduce the computation time required for trajectory optimization. A predictive simulation was developed to determine the on-orbit mass given an initial vehicle state, wind information, and a set of trajectory parameters. This simulation utilizes an idealized control system to speed computation by increasing the integration time step. The conjugate gradient method is used for the numerical optimization of on-orbit mass. The method requires only the evaluation of the on-orbit mass function using the predictive simulation, and the gradient of the on-orbit mass function with respect to the trajectory parameters. The gradient is approximated with finite differencing. Prelaunch trajectory designs were carried out using the optimization procedure. The predictive simulation is used in flight to redesign the trajectory to account for trajectory deviations produced by off-nominal conditions, e.g., stronger than expected head winds.

Sullivan, Jeanne Marie↗
DOE OSTI2021

A novel machine learning based identification of potential adopter of rooftop solar photovoltaics

With the proliferation of rooftop solar photovoltaic installations, there is a need to proactively predict consumer potential for solar photovoltaic adoption, for improved electric utility planning and operation. Traditional analytical modeling approaches are limited to a few survey features and a larger part of the survey would remain untouched by the decision model. This article presents a novel, data-driven modeling approach that strategically prunes a large set of consumer profile features using a machine learning framework to train a model for predicting potential solar adoption. The approach utilizes the Gradient Boosting Decision Tree model through a Light Gradient Boosting framework that improves significantly over the poor prediction accuracy of the existing approaches. Model training using focal-loss based supervision is used to overcome the difficulty in identifying the potential adopters that is inherent in conventional data-driven models. In addition, to overcome possible data sparsity in a limited survey sample, a Generative Adversarial Network is presented to create synthetic user samples and its effectiveness on model performance is assessed. A Bayesian optimization approach is used to systematically arrive at the hyperparameters of the proposed model. Validation of the presented approach on a survey data collected by the National Rural Electric Cooperative Association in Virginia in 2018 demonstrates the excellent predictive capability of the machine learning based approach to modeling solar adoption reliably.

14 SOLAR ENERGY↗
DOE OSTI2026

NCCS High Performance GMRES Mixed Precision

HPG-MxP is a software package that performs a fixed number of multigrid preconditioned (using a Gauss-Seidel smoother) Generalized minimal residual (PGMRES) iterations in order to solve a possibly nonsymmetric large sparse linear system of equations. It is designed to be a benchmark to measure a computer's performance for sparse linear algebra workloads typical in scientific computing while allowing the use of mixed precision methods. The solution is required to have convergence characteristics and accuracy similar to double precision GMRES. It is based on the High Performance Conjugate Gradient Benchmark (HPCG) which restricts all implementations to use only the IEEE double precision format (FP64). The original implementation (https://github.com/hpg-mxp/hpg-mxp) was written by Ichitaro Yamazaki, Jennifer Loe, Christian Glusa, Sivasankaran Rajamanickam, Piotr Luszczek, and Jack Dongarra. Please refer to that repository for documentation on the original implementation. This version is maintained by the National Center for Computational Sciences at Oak Ridge National Laboratory. It is highly scalable and optimized for Oak Ridge Leadership Computing Facility (OLCF) systems, particularly Frontier.

Kashi, Aditya [Oak Ridge National Laboratory (ORNL↗
DOE OSTI2022

Optimizing lane reversals in transportation networks to reduce traffic congestion: A global optimization approach

This paper studies how to reduce the overall travel time of commuters in a transportation network by reversing the direction of some lanes in the network using a macroscopic network-wide perspective. Similar to the Network Design Problem, the lane reversal problem has been shown to be NP-hard given the dependence of the users’ route selection on the lane direction decision. Herein, we propose and compare three efficient methods to solve the routing and lane reversal problem jointly. First, we introduce an alternating method that decouples the routing and lane assignment problems. Second, we propose a Frank–Wolfe method that jointly takes gradient steps to adjust both the lane assignment and routing decisions. Third, we propose a convex approximation method that uses a threshold-based approach to convexify the joint routing and lane reversal objective. The convex approximation method is advantageous since it finds a global optimum solution for the approximated problem and it enables the possibility to include linear constraints. Using this method, we extend the main formulation to be able to limit a maximum number of reversed lanes, as well as to incorporate multiple origin–destination (OD) patterns. We test the proposed methods in a case study using the transportation network of Eastern Massachusetts where our results indicate an overall reduction in travel times of 4.7% by selecting the best 15 reversals. Moreover, using a small test network, we investigate the performance of the lane reversal strategies as a function of the OD demand symmetry. As expected, we observe that when the OD demand is very asymmetric (e.g., for a single OD pair, evacuations, large events), the reduction in travel times is larger than the symmetric case, reaching travel time reductions of 60%.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗
DOE OSTI2022

PETSc TSAdjoint: A Discrete Adjoint ODE Solver for First-Order and Second-Order Sensitivity Analysis

Here, we present a new software system PETSc TSAdjoint for first-order and second order adjoint sensitivity analysis of time-dependent nonlinear differential equations. The derivative calculation in PETSc TSAdjoint is essentially a high-level algorithmic differentiation process. The adjoint models are derived by differentiating the timestepping algorithms and implementing them based on the parallel infrastructure in PETSc. Full differentiation of the library code, including MPI routines, is avoided, and users do not need to derive their own adjoint models for their specific applications. PETSc TSAdjoint can compute the first-order derivative, that is, the gradient of a scalar functional, and the Hessian-vector product, which carries second-order derivative information, while requiring minimal input (a few callbacks) from the users. The adjoint model employs optimal checkpointing schemes in a manner that is transparent to users. Finally, usability, efficiency, and scalability are demonstrated through examples from a variety of applications.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Convergence and Quantum Advantage of Trotterized MERA for Strongly-Correlated Systems

Strongly-correlated quantum many-body systems are difficult to study and simulate classically. We recently proposed a variational quantum eigensolver (VQE) based on the multiscale entanglement renormalization ansatz (MERA) with tensors constrained to certain Trotter circuits. Here, we determine the scaling of computation costs for various critical spin chains which substantiates a polynomial quantum advantage in comparison to classical MERA simulations based on exact energy gradients or variational Monte Carlo. Algorithmic phase diagrams suggest an even greater separation for higher-dimensional systems. Hence, the Trotterized MERA VQE is a promising route for the efficient investigation of strongly-correlated quantum many-body systems on quantum computers. Furthermore, we show how the convergence can be substantially improved by building up the MERA layer by layer in the initialization stage and by scanning through the phase diagram during optimization. For the Trotter circuits being composed of single-qubit and two-qubit rotations, it is experimentally advantageous to have small rotation angles. We find that the average angle amplitude can be reduced considerably with negligible effect on the energy accuracy. Benchmark simulations suggest that the structure of the Trotter circuits for the TMERA tensors is not decisive; in particular, brick-wall circuits and parallel random-pair circuits yield very similar energy accuracies.

Miao, Qiang [Duke Quantum Center, Duke University,↗
NASA NTRS1994

Incremental triangulation by way of edge swapping and local optimization

This document is intended to serve as an installation, usage, and basic theory guide for the two dimensional triangulation software 'HARLEY' written for the Silicon Graphics IRIS workstation. This code consists of an incremental triangulation algorithm based on point insertion and local edge swapping. Using this basic strategy, several types of triangulations can be produced depending on user selected options. For example, local edge swapping criteria can be chosen which minimizes the maximum interior angle (a MinMax triangulation) or which maximizes the minimum interior angle (a MaxMin or Delaunay triangulation). It should be noted that the MinMax triangulation is generally only locally optical (not globally optimal) in this measure. The MaxMin triangulation, however, is both locally and globally optical. In addition, Steiner triangulations can be constructed by inserting new sites at triangle circumcenters followed by edge swapping based on the MaxMin criteria. Incremental insertion of sites also provides flexibility in choosing cell refinement criteria. A dynamic heap structure has been implemented in the code so that once a refinement measure is specified (i.e., maximum aspect ratio or some measure of a solution gradient for the solution adaptive grid generation) the cell with the largest value of this measure is continually removed from the top of the heap and refined. The heap refinement strategy allows the user to specify either the number of cells desired or refine the mesh until all cell refinement measures satisfy a user specified tolerance level. Since the dynamic heap structure is constantly updated, the algorithm always refines the particular cell in the mesh with the largest refinement criteria value. The code allows the user to: triangulate a cloud of prespecified points (sites), triangulate a set of prespecified interior points constrained by prespecified boundary curve(s), Steiner triangulate the interior/exterior of prespecified boundary curve(s), refine existing triangulations based on solution error measures, and partition meshes based on the Cuthill-McKee, spectral, and coordinate bisection strategies.

Wiltberger, N. Lyn↗
NASA NTRS2019

Modeling Phenological Controls on Carbon Dynamics in Dryland Sagebrush Ecosystems

Dryland ecosystems play an important role in determining how precipitation anomalies affect terrestrial carbonfluxes at regional to global scales. Thus, to understand how climate change may affect the global carbon cycle,we must also be able to understand and model its effects on dryland vegetation. Dynamic Global VegetationModels (DGVMs) are an important tool for modeling ecosystem dynamics, but they often struggle to reproduceseasonal patterns of plant productivity. Because the phenological niche of many plant species is linked to bothtotal productivity and competitive interactions with other plants, errors in how process-based models representphenology hinder our ability to predict climate change impacts. This may be particularly problematic in drylandecosystems where many species have developed a complex phenology in response to seasonal variability in bothmoisture and temperature. Here, we examine how uncertainty in key parameters as well as the structure ofexisting phenology routines affect the ability of a DGVM to match seasonal patterns of leaf area index (LAI) andgross primary productivity (GPP) across a temperature and precipitation gradient. First, we optimized modelparameters using a combination of site-level eddy covariance data and remotely-sensed LAI data. Second, wemodified the model to include a semi-deciduous phenology type and added flexibility to the representation ofgrass phenology. While optimizing parameters reduced model bias, the largest gains in model performance wereassociated with the development of our new representation of phenology. This modified model was able to bettercapture seasonal patterns of both leaf area index (R2=0.75) and gross primary productivity (R2=0.84), thoughits ability to estimate total annual GPP depended on using eddy covariance data for optimization. The new modelalso resulted in a more realistic outcome of modeled competition between grass and shrubs. These findingsdemonstrate the importance of improving how DGVMs represent phenology in order to accurately forecastclimate change impacts in dryland ecosystems.

Ecosystem model Phenology Parameter optimization E↗
DOE OSTI2025

Nonvariational ADAPT algorithm for quantum simulations

We explore a nonvariational quantum state preparation approach combined with the ADAPT operator selection strategy in the application of preparing the ground state of a desired target Hamiltonian. In this algorithm, energy gradient measurements determine both the operators and the gate parameters in the quantum circuit construction. We compare this nonvariational algorithm with ADAPT-VQE and with feedback-based quantum algorithms in terms of the rate of energy reduction, the circuit depth, and the measurement cost in molecular simulation. We find that, despite using deeper circuits, this new algorithm reaches chemical accuracy at a similar measurement cost to ADAPT-VQE. Since it does not rely on a classical optimization subroutine, it may provide robustness against circuit parameter errors due to imperfect control or gate synthesis.

Tang'S, Ho Lun [Virginia Polytechnic Inst. and Sta↗
DOE OSTI2021

Separation of Lanthanide Isotopes from Mixed Fission Product Samples

The measurement of radioactive fission products from nuclear events has important implications for nuclear data production, environmental monitoring, and nuclear forensics. In a previous paper, the authors reported the optimization of an intra-group lanthanide separation using LN extraction resin from Eichrom Technologies®, Inc. and a nitric acid gradient. In this work, the method was demonstrated for the separation and quantification of multiple short-lived fission product lanthanide isotopes from a fission product sample produced from the thermal irradiation of highly enriched uranium. The separations were performed in parallel in quadruplicate with reproducible results and high decontamination factors for 153 Sm, 156 Eu, and 161 Tb. Based on the results obtained here, the fission yields for 144 Ce, 153 Sm, 156 Eu, and 161 Tb are consistent with published fission yields. This work demonstrates the effectiveness of the separations for the intended application of short-lived lanthanide fission product analysis requiring high decontamination factors.

38 RADIATION CHEMISTRY, RADIOCHEMISTRY, AND NUCLEA↗
NASA NTRS1995

Flight test validation of a frequency-based system identification method on an F-15 aircraft

A frequency-based performance identification approach was evaluated using flight data from the NASA F-15 Highly Integrated Digital Electronic Control aircraft. The approach used frequency separation to identify the effectiveness of multiple controls simultaneously as an alternative to independent control identification methods. Fourier transformations converted measured control and response data into frequency domain representations. Performance gradients were formed using multiterm frequency matching of control and response frequency domain models. An objective function was generated using these performance gradients. This function was formally optimized to produce a coordinated control trim set. This algorithm was applied to longitudinal acceleration and evaluated using two control effectors: nozzle throat area and inlet first ramp. Three criteria were investigated to validate the approach: simultaneous gradient identification, gradient frequency dependency, and repeatability. This report describes the flight test results. These data demonstrate that the approach can accurately identify performance gradients during simultaneous control excitation independent of excitation frequency.

Schkolnik, Gerard S.↗
DOE OSTI2024

Metric Learning to Accelerate Convergence of Operator Splitting Methods

Recent developments in machine learning have led to promising advances in accelerating the solution of constrained optimization problems. Increasing demand for real-time decision-making capabilities in applications such as artificial intelligence and optimal control has led to a variety of proposed strategies for learning to produce fast solutions to optimization problems. For example, recent works have shown that it is possible to accelerate the convergence of optimization algorithms by learning to select their parameters, such as gradient descent stepsizes. This work proposes a new approach, in which the underlying metric spaces of proximal operator splitting algorithms are learned to maximize convergence rate. While prior works in optimization theory have derived optimal metrics in simple cases, no such result exists for many practical problem forms including general Quadratic Programming (QP). This paper shows how differentiable optimization can enable the end-to-end learning of proximal metrics, enhancing the convergence of proximal algorithms for QP problems beyond what is possible based on known theory. Additionally, the results illustrate a strong connection between the learned proximal metrics and active constraints at the optima, leading to an interpretation in which the predicted proximal metrics can be viewed as a form of active set prediction.

King, Ethan [BATTELLE (PACIFIC NW LAB)]↗
NASA NTRS1984

An evaluation and force gradient determination of mechanically linked reversible sidestick controllers for General Aviation aircraft

In connection with the increase in air traffic, IFR (Instrument Flight Rules) flight in the air traffic control system has become very demanding. It has, therefore, become imperative to optimize the pilot's skills in his management of the various aircraft systems. The present investigation is concerned with the human factors aspects of the use of sidesticks in direct mechanical linkage (reversible) control systems in a production General Aviation (G.A.) aircraft. A total of 140 fifteen to twenty minute flight tasks were flown on the NASA Langley G.A. motion base simulator. The study involved a comparison of three locations of the sidestick, left side, center, and right side, and the standard yoke. It was found that the sidestick is preferable to the standard yoke. However, some of the design and installation features of the sidestick are critical for pilot acceptance.

Bergeron, H. P.↗
DOE OSTI2026

Coarse-Grained Simulations of Polyrotaxane Hydrogels under Quiescent and Shear Conditions

Cyclodextrin-based polyrotaxanes (PR) form hydrogels in water through cyclodextrin (CD) aggregation and crystallization. These networks can break under shear flows, making them versatile platforms for extrusion-based 3D printing. To optimize the material properties of 3D-printed PR gels, a microscopic understanding of the structural evolution during 3D printing is necessary. Here, we employ coarse-grained (CG) simulations to reveal the PR assembly process at the molecular level. Our simulations reproduce the experimental crystal morphologies of PR at varying concentrations and chain lengths under quiescent conditions. Using nonequilibrium simulations, we show shear flow ruptures crystalline domain connectivity in PR gels and stacks the lamellae in the gradient direction, allowing the materials to flow during 3D printing. After the cessation of flow, the anisotropic crystal alignment and absence of available dangling PR diminish the intercrystal connectivity. The printed materials are therefore mechanically weaker than the pristine hydrogels, in agreement with experimental results. Nonetheless, by relating the microscopic structural evolution with viscoelastic properties of PR gels and solutions, we elucidate how flow conditions and sample composition affect the 3D printing performance of PR hydrogels.

Smith, Cameron D. [Dartmouth College, Hanover, NH ↗
DOE OSTI2020

Molecular Design of a Highly Stable Single-Ion Conducting Polymer Gel Electrolyte

Single-ion conducting (SIC) polymer electrolytes with a high Li transference number ( t Li + ) have shown the capability to enable enhanced battery performance and safety by avoiding liquid–electrolyte leakage and suppressing Li dendrite formation. However, issues of insufficient ionic conductivity, low electrochemical stability, and poor polymer/electrode interfacial contact have greatly hindered their commercial use. In this work, a Li-containing boron-centered fluorinated SIC polymer gel electrolyte (LiBFSIE) was rationally designed to achieve a high t Li + and high electrochemical stability. Owing to the low dissociation energy of the boron-centered anion and Li + , the as-prepared LiBFSIE exhibited an ionic conductivity of 2 × 10 –4 S/cm at 35 °C, which is exclusively contributed by Li ions owing to a high t Li + of 0.93. Both simulation and experimental approaches were applied to investigate the ion diffusion and concentration gradient in the LiBFSIE and non-cross-linked dual-ion systems. Typical rectangular Li stripping/plating voltage profiles demonstrated the uniform Li deposition assisted by LiBFSIE. The interfacial contact and electrolyte infiltration were further optimized with an in situ UV–vis-initiated polymerization method together with the electrode materials. By virtue of the high electrochemical stability of LiBFSIE, the cells achieved a promising average Coulombic efficiency of 99.95% over 200 cycles, which is higher than that of liquid–electrolyte-based cells. No obvious capacity fading was observed, indicating the long-term stability of LiBFSIE for lithium metal batteries.

36 MATERIALS SCIENCE↗