Search NASA⌕ Search

SEARCH · Search NASA

Results for “Differentiable Programming”

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 217 records · Page 12
All sourcesNASA NTRSDOE OSTI
NASA NTRS1973

FORTRAN 4 computer program for calculating critical speeds of rotating shafts

A FORTRAN 4 computer program, written for the IBM DCS 7094/7044 computer, that calculates the critical speeds of rotating shafts is described. The shaft may include bearings, couplings, extra masses (nonshaft mass), and disks for the gyroscopic effect. Shear deflection is also taken into account, and provision is made in the program for sections of the shaft that are tapered. The boundary conditions at the ends of the shaft can be fixed (deflection and slope equal to zero) or free (shear and moment equal to zero). The fixed end condition enables the program to calculate the natural frequencies of cantilever beams. Instead of using the lumped-parameter method, the program uses continuous integration of the differential equations of beam flexure across different shaft sections. The advantages of this method over the usual lumped-parameter method are less data preparation and better approximation of the distribution of the mass of the shaft. A main feature of the program is the nature of the output. The Calcomp plotter is used to produce a drawing of the shaft with superimposed deflection curves at the critical speeds, together with all pertinent information related to the shaft.

Trivisonno, R. J.↗
NASA NTRS1974

A study of the Sabatier-methanation reaction kinetics

The kinetics of the Sabatier methanation reaction, the reduction of carbon dioxide with hydrogen to methane and water, was investigated for 58 percent nickel on kieselguhr catalyst and 20 percent ruthenium on alumina catalyst. Differential rate data from an experimental program were correlated with a power function rate equation both for forward and reverse reactions. The kinetic parameters of activation energy, frequency rate constant and reaction order were determined for the rate equation. The values of these parameters were obtained from an Arrhenius plot of the experimental differential rate data. Also the carbon monoxide side reaction effect was measured and included in the correlation of parameters. The reaction was found to fit the rate equation experimentally within the temperature range 421 K, where the reaction effectively begins, the 800 K where the reaction rate drops and departs from the rate equation form.

Verostko, C. E.↗
NASA NTRS2007

Phototransistors Development and their Applications to Lidar

Custom-designed two-micron phototransistors have been developed using Liquid Phase Epitaxy (LPE), Molecular Beam Epitaxy (MBE) and Metal-Organic Chemical Vapor Deposition (MOCVD) techniques under Laser Risk Reduction Program (LRRP). The devices were characterized in the Detector Characterization Laboratory at NASA Langley Research Center. It appears that the performance of LPE- and MBE-grown phototransistors such as responsivity, noise-equivalent-power, and gain, are better than MOCVD-grown devices. Lidar tests have been conducted using LPE and MBE devices under the 2-micrometer CO2 Differential Absorption Lidar (DIAL) Instrument Incubator Program (IIP) at the National Center for Atmospheric Research (NCAR), Boulder, Colorado. The main focus of these tests was to examine the phototransistors performances as compared to commercial InGaAs avalanche photodiode by integrating them into the Raman-shifted Eye-safe Aerosol Lidar (REAL) operating at 1.543 micrometers. A simultaneous measurement of the atmospheric backscatter signals using the LPE phototransistors and the commercial APD demonstrated good agreement between these two devices. On the other hand, simultaneous detection of lidar backscatter signals using MBE-grown phototransistor and InGaAs APD, showed a general agreement between these two devices with a lower performance than LPE devices. These custom-built phototransistors were optimized for detection around 2-micrometer wavelength while the lidar tests were performed at 1.543 micrometers. Phototransistor operation at 2-micron will improve the performance of a lidar system operating at that wavelength. Measurements include detecting hard targets (Rocky Mountains), atmospheric structure consisting of cirrus clouds and boundary layer. These phototransistors may have potential for high sensitivity differential absorption lidar measurements of carbon dioxide and water vapor at 2.05-micrometers and 1.9-micrometers, respectively.

Abedin, M. N.↗
DOE OSTI2021

Automatic Differentiation in MetaPhysicL and Its Applications in MOOSE

Efficient solution via Newton’s method of nonlinear systems of equations requires an accurate representation of the Jacobian, corresponding to the derivatives of the component residual equations with respect to the degrees of freedom. In practice these systems of equations often arise from spatial discretization of partial differential equations used to model physical phenomena. These equations may involve domain motion or material equations that are complex functions of the systems’ degrees of freedom. Computing the Jacobian by hand in these situations is arduous and prone to error. Finite difference approximations of the Jacobian or its action are prone to truncation error, especially in multiphysics settings. Symbolic differentiation packages may be used, but often result in an excessive number of terms in realistic model scenarios. An alternative to symbolic and numerical differentiation is automatic differentiation (AD), which propagates derivatives with every elementary operation of a computer program, corresponding to continual application of the chain rule. Automatic differentiation offers the guarantee of an exact Jacobian at a relatively small overhead cost. In this work, we outline the adoption of AD in the Multiphysics Object Oriented Simulation Environment (MOOSE) via the MetaPhysicL package. We describe the application of MOOSE’s AD capability to several sets of physics that were previously infeasible to model via hand-coded or Jacobian-free simulation techniques, including arbitrary Lagrangian-Eulerian and level-set simulations of laser melt pools, phase-field simulations with free energies provided through neural networks, and metallic nuclear fuel simulations that require inner Newton loop calculation of nonlinear material properties.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
NASA NTRS1989

Surface characterization and testing II; Proceedings of the Meeting, San Diego, CA, Aug. 10, 11, 1989

Various papers on surface characterization and testing are presented. Individual topics addressed include: simple Hartmann test data interpretation, optimum configuration of the Offner null corrector, system for phase-shifting interferometry in the presence of vibration, fringe variation and visibility in speckle-shearing interferometry, functional integral representation of rough surfaces, calibration of surface heights in an interferometric optical profiler, image formation in common path differential profilometers, SEM of optical surfaces, measuring surface profiles with scanning tunneling microscopes, surface profile measurements of curved parts, high-resolution optical profiler, scanning heterodyne interferometer with immunity from microphonics, real-time crystal axis measurements of semiconductor materials, radial metrology with a panoramic annular lens, surface analysis for the characterization of defects in thin-film processes, Spacelab Optical Viewport glass assembly optical test program for the Starlab mission, scanning differential intensity and phase system for optical metrology.

Greivenkamp, John E.↗
DOE OSTI2024

Scrutinizing Variables for Checkpoint Using Automatic Differentiation

Checkpoint/Restart (C/R) saves the running state of the programs periodically, which consumes considerable time and system resources. We observe that not every piece of data is involved in the computation in typical HPC applications; such unused data should be excluded from checkpointing for better storage and compute efficiency. We propose a systematic approach that leverages automatic differentiation (AD) to scrutinize every element within variables (e.g., arrays) necessary for checkpointing. This allows us to identify critical and uncritical elements and eliminate uncritical elements from checkpointing. Specifically, we inspect every single element within a variable necessary for checkpointing with an AD tool to determine whether the element has an impact on the application output or not. We validate our approach with all benchmarks from the NPB suite. We visualize the distribution of critical and uncritical elements within a variable with respect to its binary impact (yes or no) on the application output.

Huang, Xin [Kobe University]↗
NASA NTRS1979

Numerical calculations of high-altitude differential charging: Preliminary results

A two dimensional simulation program was constructed in order to obtain theoretical predictions of floating potential distributions on geostationary spacecraft. The geometry was infinite-cylindrical with angle dependence. Effects of finite spacecraft length on sheath potential profiles can be included in an approximate way. The program can treat either steady-state conditions or slowly time-varying situations, involving external time scales much larger than particle transit times. Approximate, locally dependent expressions were used to provide space charge, density profiles, but numerical orbit-following is used to calculate surface currents. Ambient velocity distributions were assumed to be isotropic, beam-like, or some superposition of these.

Laframboise, J. G.↗
DOE OSTI2022

Source-to-Source Automatic Differentiation of OpenMP Parallel Loops

This article presents our work toward correct and efficient automatic differentiation of OpenMP parallel worksharing loops in forward and reverse mode. Automatic differentiation is a method to obtain gradients of numerical programs, which are crucial in optimization, uncertainty quantification, and machine learning. The computational cost to compute gradients is a common bottleneck in practice. For applications that are parallelized for multicore CPUs or GPUs using OpenMP, one also wishes to compute the gradients in parallel. Here, we propose a framework to reason about the correctness of the generated derivative code, from which we justify our OpenMP extension to the differentiation model. We implement this model in the automatic differentiation tool Tapenade and present test cases that are differentiated following our extended differentiation procedure. Performance of the generated derivative programs in forward and reverse mode is better than sequential, although our reverse mode often scales worse than the input programs.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Proximal Galerkin: A Structure-Preserving Finite Element Method for Pointwise Bound Constraints

The proximal Galerkin finite element method is a high-order, low iteration complexity, nonlinear numerical method that preserves the geometric and algebraic structure of pointwise bound constraints in infinite-dimensional function spaces. This paper introduces the proximal Galerkin method and applies it to solve free boundary problems, enforce discrete maximum principles, and develop a scalable, mesh-independent algorithm for optimal design with pointwise bound constraints. This paper also introduces the latent variable proximal point (LVPP) algorithm, from which the proximal Galerkin method derives. When analyzing the classical obstacle problem, we discover that the underlying variational inequality can be replaced by a sequence of second-order partial differential equations (PDEs) that are readily discretized and solved with, e.g., the proximal Galerkin method. Throughout this work, we arrive at several contributions that may be of independent interest. These include (1) a semilinear PDE we refer to as the entropic Poisson equation; (2) an algebraic/geometric connection between high-order positivity-preserving discretizations and certain infinite-dimensional Lie groups; and (3) a gradient-based, bound-preserving algorithm for two-field, density-based topology optimization. The complete proximal Galerkin methodology combines ideas from nonlinear programming, functional analysis, tropical algebra, and differential geometry and can potentially lead to new synergies among these areas as well as within variational and numerical analysis. Open-source implementations of our methods accompany this work to facilitate reproduction and broader adoption.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1990

An adaptive grid algorithm for one-dimensional nonlinear equations

Richards' equation, which models the flow of liquid through unsaturated porous media, is highly nonlinear and difficult to solve. Step gradients in the field variables require the use of fine grids and small time step sizes. The numerical instabilities caused by the nonlinearities often require the use of iterative methods such as Picard or Newton interation. These difficulties result in large CPU requirements in solving Richards equation. With this in mind, adaptive and multigrid methods are investigated for use with nonlinear equations such as Richards' equation. Attention is focused on one-dimensional transient problems. To investigate the use of multigrid and adaptive grid methods, a series of problems are studied. First, a multigrid program is developed and used to solve an ordinary differential equation, demonstrating the efficiency with which low and high frequency errors are smoothed out. The multigrid algorithm and an adaptive grid algorithm is used to solve one-dimensional transient partial differential equations, such as the diffusive and convective-diffusion equations. The performance of these programs are compared to that of the Gauss-Seidel and tridiagonal methods. The adaptive and multigrid schemes outperformed the Gauss-Seidel algorithm, but were not as fast as the tridiagonal method. The adaptive grid scheme solved the problems slightly faster than the multigrid method. To solve nonlinear problems, Picard iterations are introduced into the adaptive grid and tridiagonal methods. Burgers' equation is used as a test problem for the two algorithms. Both methods obtain solutions of comparable accuracy for similar time increments. For the Burgers' equation, the adaptive grid method finds the solution approximately three times faster than the tridiagonal method. Finally, both schemes are used to solve the water content formulation of the Richards' equation. For this problem, the adaptive grid method obtains a more accurate solution in fewer work units and less computation time than required by the tridiagonal method. The performance of the adaptive grid method tends to degrade as the solution process proceeds in time, but still remains faster than the tridiagonal scheme.

Gutierrez, William E.↗
NASA NTRS2005

Advanced Weather Awareness and Reporting Enhancements

AWARE (Aviation Weather Awareness and Reporting Enhancements) was a NASA Cooperative Research and Development program conducted jointly by Rockwell Scientific, Rockwell Collins, and NASA. The effort culminated in an enhanced weather briefing and reporting tool prototype designed to integrate graphical and text-based aviation weather data to provide clear situational awareness in the context of a specific pilot, flight and equipment profile. The initial implementation of AWARE was as a web-based preflight planning tool, specifically for general aviation pilots, who do not have access to support such as the dispatchers available for commercial airlines. Initial usability tests showed that for VFR (Visual Flight Rules) pilots, AWARE provided faster and more effective weather evaluation. In a subsequent formal usability test for IFR (Instrument Flight Rules) pilots, all users finished the AWARE tests faster than the parallel DUAT tests, and all subjects graded AWARE higher for effectiveness, efficiency, and usability. The decision analysis basis of AWARE differentiates it from other aviation safety programs, providing analysis of context-sensitive data in a personalized graphical format to aid pilots/dispatchers in their complex flight requirements.

Busquets, Anthony M.↗
NASA NTRS1977

MACSYMA's symbolic ordinary differential equation solver

The MACSYMA's symbolic ordinary differential equation solver ODE2 is described. The code for this routine is delineated, which is of interest because it is written in top-level MACSYMA language, and may serve as a good example of programming in that language. Other symbolic ordinary differential equation solvers are mentioned.

Golden, J. P.↗
NASA NTRS1992

Modeling of the WSTF frictional heating apparatus in high pressure systems

In order to develop a computer program able to model the frictional heating of metals in high pressure oxygen or nitrogen a number of additions have been made to the frictional heating model originally developed for tests in low pressure helium. These additions include: (1) a physical property package for the gases to account for departures from the ideal gas state; (2) two methods for spatial discretization (finite differences with quadratic interpolation or orthogonal collocation on finite elements) which substantially reduce the computer time required to solve the transient heat balance; (3) more efficient programs for the integration of the ordinary differential equations resulting from the discretization of the partial differential equations; and (4) two methods for determining the best-fit parameters via minimization of the mean square error (either a direct search multivariable simplex method or a modified Levenburg-Marquardt algorithm). The resulting computer program has been shown to be accurate, efficient and robust for determining the heat flux or friction coefficient vs. time at the interface of the stationary and rotating samples.

Skowlund, Christopher T.↗
NASA NTRS1975

Studies of the differential absorption rocket experiment

Investigations of the ionosphere, in the rocket program of the Aeronomy Laboratory, include a propagation experiment, the data from which may be analyzed in several modes. This report considers in detail the differential absorption experiment. The sources of error and limitations of sensitivity are discussed. Methods of enhancing the performance of the experiment are described. Some changes have been made in the system and the improvement demonstrated. Suggestions are made for further development of the experiment.

Ginther, J. C.↗
NASA NTRS1982

Programming for energy monitoring/display system in multicolor lidar system research

The Z80 microprocessor based computer program that directs and controls the operation of the six channel energy monitoring/display system that is a part of the NASA Multipurpose Airborne Differential Absorption Lidar (DIAL) system is described. The program is written in the Z80 assembly language and is located on EPROM memories. All source and assembled listings of the main program, five subroutines, and two service routines along with flow charts and memory maps are included. A combinational block diagram shows the interfacing (including port addresses) between the six power sensors, displays, front panel controls, the main general purpose minicomputer, and this dedicated microcomputer system.

Alvarado, R. C., Jr.↗
DOE OSTI2025

McCormick envelopes in mixed-integer PDE-constrained optimization

McCormick envelopes are a standard tool for deriving convex relaxations of optimization problems that involve polynomial terms. Such McCormick relaxations provide lower bounds, for example, in branch-and-bound procedures for mixed-integer nonlinear programs but have not gained much attention in PDE-constrained optimization so far. This lack of attention may be due to the distributed nature of such problems, which on the one hand leads to infinitely many linear constraints (generally state constraints that may be difficult to handle) in addition to the state equation for a pointwise formulation of the McCormick envelopes and renders bound-tightening procedures that successively improve the resulting convex relaxations computationally intractable. We analyze McCormick envelopes for a model problem class that is governed by a semilinear PDE involving a bilinearity and integrality constraints. We approximate the nonlinearity and in turn the McCormick envelopes by averaging the involved terms over the cells of a partition of the computational domain on which the PDE is defined. This yields convex relaxations that underestimate the original problem up to an a priori error estimate that depends on the mesh size of the discretization. These approximate McCormick relaxations can be improved by means of an optimization-based bound-tightening procedure. We show that their minimizers converge to minimizers to a limit problem with a pointwise formulation of the McCormick envelopes when driving the mesh size to zero. We provide a computational example, for which we certify all of our imposed assumptions. The results point to both the potential of the methodology and the gaps in the research that need to be closed. Our methodology provides a framework first for obtaining pointwise underestimators for nonconvexities and second for approximating them with finitely many linear inequalities in an infinite-dimensional setting.

Approximations and Expansions↗
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)]↗