Search NASA⌕ Search

SEARCH · Search NASA

Results for “algorithmic differentiation”

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

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

At least 721 records · Page 40
All sourcesNASA NTRSDOE OSTI
NASA NTRS2010

High-resolution three-dimensional imaging radar

A three-dimensional imaging radar operating at high frequency e.g., 670 GHz, is disclosed. The active target illumination inherent in radar solves the problem of low signal power and narrow-band detection by using submillimeter heterodyne mixer receivers. A submillimeter imaging radar may use low phase-noise synthesizers and a fast chirper to generate a frequency-modulated continuous-wave (FMCW) waveform. Three-dimensional images are generated through range information derived for each pixel scanned over a target. A peak finding algorithm may be used in processing for each pixel to differentiate material layers of the target. Improved focusing is achieved through a compensation signal sampled from a point source calibration target and applied to received signals from active targets prior to FFT-based range compression to extract and display high-resolution target images. Such an imaging radar has particular application in detecting concealed weapons or contraband.

Cooper, Ken B.↗
DOE OSTI2011

DIMENSION REDUCTION NUMERICAL CLOSURE METHOD FOR ADVECTION-DIFFUSION-REACTION SYSTEMS

Many natural physical processes exhibit a multiscale behavior, that is the same processes allow different mathematical description on different scales. The microscale description is usually based on fundamental conservation laws that form a closed system of ordinary differential equations (ODEs) or partial differential equations (PDEs) but the numerical discretization of these equations may produce a system of ODEs with an enormous number of unknowns. Furthermore, a time integration of the microscale equations usually requires time steps that are smaller than the observation time by many orders of magnitude. A direct solution of these ODEs can be extremely expensive. This necessitates development of advanced algorithms for model (or dimension) reduction. Often, we are only interested in the average behavior of the system rather than the exact solution of the ODEs. Here we developed a novel dimension reduction method that gives an approximate solution of the ODEs and gives an accurate prediction of the average behavior. The method relies on a computational closure of averaged evolution balance equations. The computational closure is achieved via short bursts of microscale simulation. The dimension reduction model was used to simulate flow and transport with mixing controlled reactions and mineral precipitation. A pore-scale model was used as a microscale model. A good agreement between microscale simulations and the accelerated microscale simulations confirms the accuracy and computational efficiency of the dimension reduction model. The method significant accelerates microscale simulations, while providing accurate approximation of the solution and accurate prediction of the average behavior of the system.

99 GENERAL AND MISCELLANEOUS↗
DOE OSTI2014

Monte Carlo based beam model using a photon MLC for modulated electron radiotherapy

Purpose: Modulated electron radiotherapy (MERT) promises sparing of organs at risk for certain tumor sites. Any implementation of MERT treatment planning requires an accurate beam model. The aim of this work is the development of a beam model which reconstructs electron fields shaped using the Millennium photon multileaf collimator (MLC) (Varian Medical Systems, Inc., Palo Alto, CA) for a Varian linear accelerator (linac). Methods: This beam model is divided into an analytical part (two photon and two electron sources) and a Monte Carlo (MC) transport through the MLC. For dose calculation purposes the beam model has been coupled with a macro MC dose calculation algorithm. The commissioning process requires a set of measurements and precalculated MC input. The beam model has been commissioned at a source to surface distance of 70 cm for a Clinac 23EX (Varian Medical Systems, Inc., Palo Alto, CA) and a TrueBeam linac (Varian Medical Systems, Inc., Palo Alto, CA). For validation purposes, measured and calculated depth dose curves and dose profiles are compared for four different MLC shaped electron fields and all available energies. Furthermore, a measured two-dimensional dose distribution for patched segments consisting of three 18 MeV segments, three 12 MeV segments, and a 9 MeV segment is compared with corresponding dose calculations. Finally, measured and calculated two-dimensional dose distributions are compared for a circular segment encompassed with a C-shaped segment. Results: For 15 × 34, 5 × 5, and 2 × 2 cm{sup 2} fields differences between water phantom measurements and calculations using the beam model coupled with the macro MC dose calculation algorithm are generally within 2% of the maximal dose value or 2 mm distance to agreement (DTA) for all electron beam energies. For a more complex MLC pattern, differences between measurements and calculations are generally within 3% of the maximal dose value or 3 mm DTA for all electron beam energies. For the two-dimensional dose comparisons, the differences between calculations and measurements are generally within 2% of the maximal dose value or 2 mm DTA. Conclusions : The results of the dose comparisons suggest that the developed beam model is suitable to accurately reconstruct photon MLC shaped electron beams for a Clinac 23EX and a TrueBeam linac. Hence, in future work the beam model will be utilized to investigate the possibilities of MERT using the photon MLC to shape electron beams.

62 RADIOLOGY AND NUCLEAR MEDICINE↗
DOE OSTI2015

Enhancing adaptive sparse grid approximations and improving refinement strategies using adjoint-based a posteriori error estimates

In this paper we present an algorithm for adaptive sparse grid approximations of quantities of interest computed from discretized partial differential equations. We use adjoint-based a posteriori error estimates of the physical discretization error and the interpolation error in the sparse grid to enhance the sparse grid approximation and to drive adaptivity of the sparse grid. Utilizing these error estimates provides significantly more accurate functional values for random samples of the sparse grid approximation. We also demonstrate that alternative refinement strategies based upon a posteriori error estimates can lead to further increases in accuracy in the approximation over traditional hierarchical surplus based strategies. Throughout this paper we also provide and test a framework for balancing the physical discretization error with the stochastic interpolation error of the enhanced sparse grid approximation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
NASA NTRS1974

Finite element solution theory for three-dimensional boundary flows

A finite element algorithm is derived for the numerical solution of a three-dimensional flow field described by a system of initial-valued, elliptic boundary value partial differential equations. The familiar three-dimensional boundary layer equations belong to this description when diffusional processes in only one coordinate direction are important. The finite element algorithm transforms the original description into large order systems of ordinary differential equations written for the dependent variables discretized at node points of an arbitrarily irregular computational lattice. The generalized elliptic boundary conditions is piecewise valid for each dependent variable on boundaries that need not explicitly coincide with coordinate surfaces. Solutions for sample problems in laminar and turbulent boundary flows illustrate favorable solution accuracy, convergence, and versatility.

Baker, A. J.↗
DOE OSTI2024

Binary Quantum Control Optimization with Uncertain Hamiltonians

Optimizing the controls of quantum systems plays a crucial role in advancing quantum technologies. The time-varying noises in quantum systems and the widespread use of inhomogeneous quantum ensembles raise the need for high-quality quantum controls under uncertainties. In this paper, we consider a stochastic discrete optimization formulation of a discretized binary optimal quantum control problem involving Hamiltonians with predictable uncertainties. We propose a sample-based reformulation that optimizes both risk-neutral and risk-averse measurements of control policies, and solve these with two gradient-based algorithms using sum-up-rounding approaches. Furthermore, we discuss the differentiability of the objective function and prove upper bounds of the gaps between the optimal solutions to binary control problems and their continuous relaxations. We conduct numerical simulations on various sized problem instances based on two applications of quantum pulse optimization; we evaluate different strategies to mitigate the impact of uncertainties in quantum systems. In conclusion, we demonstrate that the controls of our stochastic optimization model achieve significantly higher quality and robustness compared with the controls of a deterministic model.

conditional value-at-risk (CVaR)↗
DOE OSTI2024

GPU-enabled extreme-scale turbulence simulations: Fourier pseudo-spectral algorithms at the exascale using OpenMP offloading

Fourier pseudo-spectral methods for nonlinear partial differential equations are of wide interest in many areas of advanced computational science, including direct numerical simulation of three-dimensional (3-D) turbulence governed by the Navier-Stokes equations in fluid dynamics. This paper presents a new capability for simulating turbulence at a new record resolution up to 35 trillion grid points, on the world's first exascale computer, Frontier, comprising AMD MI250x GPUs with HPE's Slingshot interconnect and operated by the US Department of Energy's Oak Ridge Leadership Computing Facility (OLCF). Key programming strategies designed to take maximum advantage of the machine architecture involve performing almost all computations on the GPU which has the same memory capacity as the CPU, performing all-to-all communication among sets of parallel processes directly on the GPU, and targeting GPUs efficiently using OpenMP offloading for intensive number-crunching including 1-D Fast Fourier Transforms (FFT) performed using AMD ROCm library calls. With 99% of computing power on Frontier being on the GPU, leaving the CPU idle leads to a net performance gain via avoiding the overhead of data movement between host and device except when needed for some I/O purposes. Memory footprint including the size of communication buffers for MPI_ALLTOALL is managed carefully to maximize the largest problem size possible for a given node count. Detailed performance data including separate contributions from different categories of operations to the elapsed wall time per step are reported for five grid resolutions, from 2048 3 on a single node to 32768 3 on 4096 or 8192 nodes out of 9408 on the system. Both 1D and 2D domain decompositions which divide a 3D periodic domain into slabs and pencils respectively are implemented. The present code suite (labeled by the acronym GESTS, GPUs for Extreme Scale Turbulence Simulations) achieves a figure of merit (in grid points per second) exceeding goals set in the Center for Accelerated Application Readiness (CAAR) program for Frontier. The performance attained is highly favorable in both weak scaling and strong scaling, with notable departures only for 2048 3 where communication is entirely intra-node, and for 32768 3 , where a challenge due to small message sizes does arise. Communication performance is addressed further using a lightweight test code that performs all-to-all communication in a manner matching the full turbulence simulation code. Performance at large problem sizes is affected by both small message size due to high node counts as well as dragonfly network topology features on the machine, but is consistent with official expectations of sustained performance on Frontier. Overall, although not perfect, the scalability achieved at the extreme problem size of 32768 3 (and up to 8192 nodes — which corresponds to hardware rated at just under 1 exaflop/sec of theoretical peak computational performance) is arguably better than the scalability observed using prior state-of-the-art algorithms on Frontier's predecessor machine (Summit) at OLCF. New science results for the study of intermittency in turbulence enabled by this code and its extensions are to be reported separately in the near future.

3D fast Fourier transform↗
DOE OSTI2019

Along-the-channel modeling and analysis of PEFCs at low stoichiometry: Development of a 1+2D model

Water management remains a key challenge in polymer-electrolyte fuel cells. In this paper, a pseudo 3-D (1+2D) model is developed to account better for changes of water management along the channel, as well as verify the possibilities of using differential cells for data capture and translation to integral cell performance. An accurate 2-D membrane-electrode-assembly model is developed for differential cell modeling, which is combined with an along-the-channel stepping algorithm to account for down the channel changes in pressure, temperature, reactant concentration, and relative humidity. Variations in cell performance along the channel due to changes in operating conditions are characterized quantitatively and optimized, where drier feed conditions demonstratively require such an approach. In conclusion, the study identifies gaps between differential and integral cells including those related to flow velocity and highlights the need for better models to understand and link integral cell performance and water management.

30 DIRECT ENERGY CONVERSION↗
NASA NTRS1999

Backup Attitude Control Algorithms for the MAP Spacecraft

The Microwave Anisotropy Probe (MAP) is a follow-on to the Differential Microwave Radiometer (DMR) instrument on the Cosmic Background Explorer (COBE) spacecraft. The MAP spacecraft will perform its mission, studying the early origins of the universe, in a Lissajous orbit around the Earth-Sun L(sub 2) Lagrange point. Due to limited mass, power, and financial resources, a traditional reliability concept involving fully redundant components was not feasible. This paper will discuss the redundancy philosophy used on MAP, describe the hardware redundancy selected (and why), and present backup modes and algorithms that were designed in lieu of additional attitude control hardware redundancy to improve the odds of mission success. Three of these modes have been implemented in the spacecraft flight software. The first onboard mode allows the MAP Kalman filter to be used with digital sun sensor (DSS) derived rates, in case of the failure of one of MAP's two two-axis inertial reference units. Similarly, the second onboard mode allows a star tracker only mode, using attitude and derived rate from one or both of MAP's star trackers for onboard attitude determination and control. The last backup mode onboard allows a sun-line angle offset to be commanded that will allow solar radiation pressure to be used for momentum management and orbit stationkeeping. In addition to the backup modes implemented on the spacecraft, two backup algorithms have been developed in the event of less likely contingencies. One of these is an algorithm for implementing an alternative scan pattern to MAP's nominal dual-spin science mode using only one or two reaction wheels and thrusters. Finally, an algorithm has been developed that uses thruster one shots while in science mode for momentum management. This algorithm has been developed in case system momentum builds up faster than anticipated, to allow adequate momentum management while minimizing interruptions to science. In this paper, each mode and algorithm will be discussed, and simulation results presented.

ODonnell, James R., Jr.↗
DOE OSTI2019

A Leray regularized ensemble-proper orthogonal decomposition method for parameterized convection-dominated flows

Abstract Partial differential equations (PDEs) are often dependent on input quantities that are uncertain. To quantify this uncertainty PDEs must be solved over a large ensemble of parameters. Even for a single realization this can be a computationally intensive process. In the case of flows governed by the Navier–Stokes equations, an efficient method has been devised for computing an ensemble of solutions. To further reduce the computational cost of this method, an ensemble-proper orthogonal decomposition (POD) method was recently proposed. The main contribution of this work is the introduction of POD spatial filtering for ensemble-POD methods. The POD spatial filter makes possible the construction of the Leray ensemble-POD model, which is a regularized-reduced order model for the numerical simulation of convection-dominated flows of moderate Reynolds number. The Leray ensemble-POD model employs the POD spatial filter to smooth (regularize) the convection term in the Navier–Stokes equations, and diminishes the numerical inaccuracies produced by the ensemble-POD method in the numerical simulation of convection-dominated flows. Specifically, for the numerical simulation of a convection-dominated two-dimensional flow between two offset cylinders, we show that the Leray ensemble-POD method better reflects the dynamics of the benchmark results than the ensemble-POD scheme. The second contribution of this work is a new numerical discretization of the variable viscosity ensemble algorithm in which the average viscosity is replaced with the maximum viscosity. It is shown that this new numerical discretization is significantly more stable than those in current use. Furthermore, error estimates for the novel Leray ensemble-POD algorithm with this new numerical discretization are also proven.

Mathematics↗
NASA NTRSDate not supplied

Assessment of Cloud Fraction Derived from a New Geostationary Satellite Cloud Retrieval Algorithm for CERES and Progress Towards Cross-platform Continuity

For over two decades The Clouds and the Earth’s Radiant Energy System (CERES) project has produced long-term records of top-of-atmosphere (TOA) and surface irradiances for detecting changes in the Earth’s radiation budget and advancing understanding of how clouds contribute to those changes. Accurate characterization of the spatial and temporal distributions of clouds are a critical component for producing CERES datasets. The CERES Cloud Working Group (CWG) derives cloud properties from both geostationary (GEO) and low-Earth orbit (LEO) satellite sensors in order to provide complete global coverage at hourly temporal resolution. However, the use of multiple sensors to provide this spatiotemporal coverage throughout a long-term record presents some challenges, because the various sensors generally have different spectral band characteristics (e.g., spectral band width and response) and spatial resolution. These differences can result in spatial artifacts at the coverage boundary between two sensors or temporal artifacts when one sensor replaces another in the record. For the upcoming Edition 5 release of CERES products, the CWG is developing cloud retrieval algorithms which utilize only spectral bands common to most modern passive satellite radiometers and account for differences in spectral width and response. The goal with this approach is to provide global cloud properties for CERES with greater cross-platform consistency than the previous Edition 4 products and thus mitigate artifacts which are evident at the interface of two sensors. This study focuses on retrieval of total cloud fraction from various GEO sensors, and we use Cloud-Aerosol Lidar with Orthogonal Polarization (CALIOP) observations to assess the accuracy of total cloud fraction derived from the various sensors. The Edition 5 cloud mask algorithm depends on estimates of cloud-free TOA spectral radiances to differentiate clear and cloudy conditions, and CALIOP is used to assess the accuracy of those estimates when cloud-free conditions are indeed observed.

CALIOP↗
DOE OSTI2026

ZEUS: An Efficient GPU Optimization Method Integrating PSO, BFGS, and Automatic Differentiation

We introduce a novel, efficient computational method, ZEUS, for numerical optimization, and provide an open-source implementation. It has four key ingredients: (1) particle swarm optimization (PSO), (2) the use of the Broyden-Fletcher-Goldfarb-Shanno (BFGS) method, (3) automatic differentiation (AD), and (4) GPUs. Our approach addresses the computational challenges inherent in high-dimensional, non-convex optimization problems. In the first phase of the algorithm, we get a potentially good set of starting points using PSO. Thereafter, we run BFGS independently in parallel from these starting points. BFGS is one of the best-performing algorithms for numerical optimization. However, it requires the gradient of the function being optimized. ZEUS integrates automatic differentiation into BFGS thus avoiding the need for the user to calculate derivatives explicitly. The use of GPUs allows ZEUS to speed up the calculations substantially. We carry out systematic studies to explore the trade-offs between the number of PSO iterations taken, starting points, and BFGS iteration depth. We show that a handful of iterations of PSO can improve global convergence when combined with BFGS. We also present performance studies using common test functions. The source code can be found at https://github.com/fnal-numerics/global-optimizer-gpu.

Soos, Dominik [Old Dominion U.]↗
NASA NTRS1977

Finite element analysis of low speed viscous and inviscid aerodynamic flows

A weak interaction solution algorithm was established for aerodynamic flow about an isolated airfoil. Finite element numerical methodology was applied to solution of each of differential equations governing potential flow, and viscous and turbulent boundary layer and wake flow downstream of the sharp trailing edge. The algorithm accounts for computed viscous displacement effects on the potential flow. Closure for turbulence was accomplished using both first and second order models. The COMOC finite element fluid mechanics computer program was modified to solve the identified equation systems for two dimensional flows. A numerical program was completed to determine factors affecting solution accuracy, convergence and stability for the combined potential, boundary layer, and parabolic Navier-Stokes equation systems. Good accuracy and convergence are demonstrated. Each solution is obtained within the identical finite element framework of COMOC.

Baker, A. J.↗
NASA NTRS1988

Real-time optical laboratory solution of parabolic differential equations

An optical laboratory matrix-vector processor is used to solve parabolic differential equations (the transient diffusion equation with two space variables and time) by an explicit algorithm. This includes optical matrix-vector nonbase-2 encoded laboratory data, the combination of nonbase-2 and frequency-multiplexed data on such processors, a high-accuracy optical laboratory solution of a partial differential equation, new data partitioning techniques, and a discussion of a multiprocessor optical matrix-vector architecture.

Casasent, David↗
NASA NTRS1989

Asynchronous multilevel adaptive methods for solving partial differential equations on multiprocessors - Performance results

The fast adaptive composite grid method (FAC) is an algorithm that uses various levels of uniform grids (global and local) to provide adaptive resolution and fast solution of PDEs. Like all such methods, it offers parallelism by using possibly many disconnected patches per level, but is hindered by the need to handle these levels sequentially. The finest levels must therefore wait for processing to be essentially completed on all the coarser ones. A recently developed asynchronous version of FAC, called AFAC, completely eliminates this bottleneck to parallelism. This paper describes timing results for AFAC, coupled with a simple load balancing scheme, applied to the solution of elliptic PDEs on an Intel iPSC hypercube. These tests include performance of certain processes necessary in adaptive methods, including moving grids and changing refinement. A companion paper reports on numerical and analytical results for estimating convergence factors of AFAC applied to very large scale examples.

Mccormick, S.↗
NASA NTRS2021

Retrieval Algorithm for the Column CO2 Mixing Ratio from Pulsed Multi-Wavelength Lidar Measurements

The retrieval algorithm for the column mixing ratio of CO2 from the measurements of a pulsed multi-wavelength integrated path differential absorption (IPDA) lidar is described. The lidar samples the shape of the 1572.33 nm CO2 absorption line at 15 or 30 wavelengths. The algorithm uses a least-squares fit between the CO2 line shape computed from a layered 10 atmosphere model to that sampled by the lidar. In addition to the column average CO2 dry air mole fraction (XCO2), several other parameters are also solved simultaneously from the fit. These include the Doppler shift in the received laser signal wavelengths, the product of the surface reflectivity and atmospheric transmission and a linear trend in the lidar receiver’s spectral response. The algorithm can also be used to solve for the average water vapor mixing ratio, which causes a secondary absorption in the wings of the CO2 absorption line under high humidity conditions. The least-squares fit is linearized about the 15 expected XCO2 value which allows the use of a standard linear least-squares fitting method and software tools. The standard deviation of the retrieved XCO2 is obtained from covariance matrix of the fit. An averaging kernel is defined similarly to that used for passive trace-gas sounding. Examples are presented of using the algorithm to retrieve XCO2 from the measurements from NASA Goddard’s airborne CO2 Sounder lidar made at a constant altitude and during spiral-down maneuvers.

Xiaoli Sun↗
DOE OSTI2022

Learning Stochastic Parametric Differentiable Predictive Control Policies

We present a scalable unsupervised learning-based method for obtaining explicit control policies for model predictive control problems for stochastic linear systems with additive uncertainties subject to nonlinear chance constraints. We call the proposed method stochastic parametric differentiable predictive control (SP-DPC), which extends the recently proposed deterministic DPC policy optimization algorithm. We formulate the SP-DPC as a deterministic approximation to the stochastic parametric constrained optimal control problem via independent sampling of the problem's parameters and uncertainties. This formulation allows us to directly compute the policy gradients via automatic differentiation of the problem's value function, evaluated over sampled parameters and uncertainties. In particular, the computed expectation of the problem's value function is backpropagated through the finite-time closed-loop system rollouts parametrized by a known nominal system dynamics model and neural control policy. We also provide theoretical probabilistic guarantees on closed-loop stability and chance constraints satisfaction for systems controlled by learned neural policies. We demonstrate the computational efficiency and scalability of the proposed policy optimization algorithm in three numerical examples, including systems with a large number of states or subject to nonlinear constraints.

Drgona, Jan↗
DOE OSTI2013

A dynamically bi-orthogonal method for time-dependent stochastic partial differential equations II: Adaptivity and generalizations

This is part II of our paper in which we propose and develop a dynamically bi-orthogonal method (DyBO) to study a class of time-dependent stochastic partial differential equations (SPDEs) whose solutions enjoy a low-dimensional structure. In part I of our paper [9], we derived the DyBO formulation and proposed numerical algorithms based on this formulation. Some important theoretical results regarding consistency and bi-orthogonality preservation were also established in the first part along with a range of numerical examples to illustrate the effectiveness of the DyBO method. In this paper, we focus on the computational complexity analysis and develop an effective adaptivity strategy to add or remove modes dynamically. Our complexity analysis shows that the ratio of computational complexities between the DyBO method and a generalized polynomial chaos method (gPC) is roughly of order O((m/N{sub p}){sup 3}) for a quadratic nonlinear SPDE, where m is the number of mode pairs used in the DyBO method and N{sub p} is the number of elements in the polynomial basis in gPC. The effective dimensions of the stochastic solutions have been found to be small in many applications, so we can expect m is much smaller than N{sub p} and computational savings of our DyBO method against gPC are dramatic. The adaptive strategy plays an essential role for the DyBO method to be effective in solving some challenging problems. Another important contribution of this paper is the generalization of the DyBO formulation for a system of time-dependent SPDEs. Several numerical examples are provided to demonstrate the effectiveness of our method, including the Navier–Stokes equations and the Boussinesq approximation with Brownian forcing.

97 MATHEMATICS AND COMPUTING↗