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 1,549 records · Page 86

On the Convergence of Overlapping Schwarz Decomposition for Nonlinear Optimal Control

Here, we study the convergence properties of an overlapping Schwarz decomposition algorithm for solving nonlinear optimal control problems (OCPs). The algorithm decomposes the time domain into a set of overlapping subdomains, and solves all subproblems defined over subdomains in parallel. The convergence is attained by updating primal-dual information at the boundaries of overlapping subdomains. We show that the algorithm exhibits local linear convergence, and that the convergence rate improves exponentially with the overlap size. We also establish global convergence results for a general quadratic programming, which enables the application of the Schwarz scheme inside second-order optimization algorithms (e.g., sequential quadratic programming). The theoretical foundation of our convergence analysis is a sensitivity result of nonlinear OCPs, which we call "exponential decay of sensitivity" (EDS). Intuitively, EDS states that the impact of perturbations at domain boundaries (i.e., initial and terminal time) on the solution decays exponentially as one moves into the domain. Here, we expand a previous analysis available in the literature by showing that EDS holds for both primal and dual solutions of nonlinear OCPs, under uniform second-order sufficient condition, controllability condition, and boundedness condition. We conduct experiments with a quadrotor motion planning problem and a partial differential equations (PDE) control problem to validate our theory, and show that the approach is significantly more efficient than alternating direction method of multipliers and as efficient as the centralized interior-point solver.

42 ENGINEERING↗

A gyrokinetic collision operator for magnetized Lorentz plasmas

A gyrocenter collision operator for magnetized Lorentz plasmas is derived using the Fokker-Plank method. The gyrocenter collision operator consists of drift and diffusion terms in the gyrocenter coordinates, including the diffusion of the gyrocenter, which does not exist for the collision operator in the particle phase space coordinates. The gyrocenter collision operator also depends on the transverse electric field explicitly, which is crucial for the correct treatment of collisional effects and transport in the gyrocenter coordinates. The gyrocenter collision operator derived is applied to calculate the particle and heat transport fluxes in a magnetized Lorentz plasma with an electric field. The particle and heat transport fluxes calculated from our gyrocenter collision operator agree exactly with the classical Braginskii's result [S. I. Braginskii, Reviews of Plasma Physics (Consultants Bureau, New York, 1965), Vol. 1, p. 205: P. Helander and D. J. Sigmar, Collisional Transport in Magnetized Plasmas (Cambridge University, Cambridge, 2002), p. 65], which validates the correctness of our collision operator. To calculate the transport fluxes correctly, it is necessary to apply the pullback transformation associated with gyrocenter coordinate transformation in the presence of collisions, which also serves as a practical algorithm for evaluating collisional particle and heat transport fluxes in the gyrocenter coordinates.

42 ENGINEERING↗

Nuclear Power Plant Ontology in OWL Format

This document describes an ontology for Nuclear Power Plant sites for use in software developed by Peng Yue of George Mason University, the intelligent Geospatial Feature Discovery System (iGFDS). The ontology was developed from unclassified and open sources available on the Web, primarily from Wikipedia. Figures 1 and 2 are screenshots of the Protégé application showing the classes of the Nuclear Power Plant Ontology which was developed in the OWL format. The ontology contains supplemental information that is not shown in these screenshots and that appears as “comments” in the Annotations window. For example, “SteamGenerator” and “SecondaryCoolant” only apply to “PWR” reactors. The ontology is based on general professional knowledge of Nuclear Power Plant operation and contains a number of components necessary for understanding and modeling the functionality of a Nuclear Power Plant. The content of the ontology is not limited to equipment and buildings that can be typically observed in the imagery. The things that can be seen are often the higher level items (CoolingTower, TurbineBuilding, ContainmentBuilding, etc.). It should be pointed out that PWR plants will be much easier to identify, generally, compared to BWR plants, due to the characteristic containment structure. The information on the TransformerYard was obtained from other sources not associated specifically with Nuclear Power Plants This ontology has to be viewed as a starting point of development ontologies for assisting with identification of Nuclear Power Plant sites with the GMU software. It can be a model for ontologies for other industrial sites. The classes and relations in the ontology have to be rearranged in a way that is most suitable for feature detection. Also the classes pertaining to visible components or signs of operation of a nuclear power plant will have to be linked to classes and properties of the objects in imagery that can be detected by image processing algorithms. Note that the current ontology does not contain statements about anything that is not present at a Nuclear Power Plant site, for example, “a Nuclear Power Plant DOES NOT HAVE a large pile of coal or other raw materials”, which would help logically to differentiate it from other types of power plants.

Ward, Richard C.↗

Machine Learning-Based Classification of Lignocellulosic Biomass from Pyrolysis-Molecular Beam Mass Spectrometry Data

High-throughput analysis of biomass is necessary to ensure consistent and uniform feedstocks for agricultural and bioenergy applications and is needed to inform genomics and systems biology models. Pyrolysis followed by mass spectrometry such as molecular beam mass spectrometry (py-MBMS) analyses are becoming increasingly popular for the rapid analysis of biomass cell wall composition and typically require the use of different data analysis tools depending on the need and application. Here, the authors report the py-MBMS analysis of several types of lignocellulosic biomass to gain an understanding of spectral patterns and variation with associated biomass composition and use machine learning approaches to classify, differentiate, and predict biomass types on the basis of py-MBMS spectra. Py-MBMS spectra were also corrected for instrumental variance using generalized linear modeling (GLM) based on the use of select ions relative abundances as spike-in controls. Machine learning classification algorithms e.g., random forest, k-nearest neighbor, decision tree, Gaussian Naïve Bayes, gradient boosting, and multilayer perceptron classifiers were used. The k-nearest neighbors (k-NN) classifier generally performed the best for classifications using raw spectral data, and the decision tree classifier performed the worst. After normalization of spectra to account for instrumental variance, all the classifiers had comparable and generally acceptable performance for predicting the biomass types, although the k-NN and decision tree classifiers were not as accurate for prediction of specific sample types. Gaussian Naïve Bayes (GNB) and extreme gradient boosting (XGB) classifiers performed better than the k-NN and the decision tree classifiers for the prediction of biomass mixtures. The data analysis workflow reported here could be applied and extended for comparison of biomass samples of varying types, species, phenotypes, and/or genotypes or subjected to different treatments, environments, etc. to further elucidate the sources of spectral variance, patterns, and to infer compositional information based on spectral analysis, particularly for analysis of data without a priori knowledge of the feedstock composition or identity.

59 BASIC BIOLOGICAL SCIENCES↗

Linear change of variable in normal systems.

Linear transformations of variable in differential correction processes cannot always be accomplished in the equations of condition, yet may be desirable to reduce high correlations. The means by which this may be accomplished in the system of normal equations is derived and the process reduced to a very simple computational algorithm.

Mulholland, J. D.↗

Transient wave propagation in composite plates due to impact

Stress-wave propagation in a composite plate due to an impact has been examined using anisotropic elasticity theory. The plate is modelled as a number of identical anisotropic layers whose elastic moduli are given. Mindlin's (1959) approximate theory of plates is then applied to each layer to obtain a set of difference-differential equations of motion by using the interlaminar stresses and displacements as explicit variables. Dispersion relationships for harmonic waves are found when traction-free boundary conditions are applied to both surfaces of the plate. The difference-differential equations are reduced to difference equations via integral transforms. With given impact boundary conditions these equations are solved for an arbitrary number of layers in the plate, and the transient propagation of a stress wave is calculated by means of a fast Fourier transform algorithm.

Sung Kim, B.↗

Incompressible Navier-Stokes and parabolized Navier-Stokes formulations and computational techniques

The differential formulations and computational techniques currently used for the incompressible Navier-Stokes (NS) and parabolic Navier-Stokes (PNS) equations are reviewed. In particular, attention is given to problems associated with the choice of difference equations, the method of solution and the choice of algorithm, the coupling of dependent variables and discretized equations, the application of boundary conditions, and grid generation. A new composite velocity NS and PNS formulation in (u,v,p) variables is presented, and the applicability of a 'forward' difference global pressure iteration for the (u,v,p) PNS system is demonstrated.

Rubin, S. G.↗

Applications of an exponential finite difference technique

An exponential finite difference scheme first presented by Bhattacharya for one dimensional unsteady heat conduction problems in Cartesian coordinates was extended. The finite difference algorithm developed was used to solve the unsteady diffusion equation in one dimensional cylindrical coordinates and was applied to two and three dimensional conduction problems in Cartesian coordinates. Heat conduction involving variable thermal conductivity was also investigated. The method was used to solve nonlinear partial differential equations in one and two dimensional Cartesian coordinates. Predicted results are compared to exact solutions where available or to results obtained by other numerical methods.

Handschuh, Robert F.↗

Observations on computational methodologies for use in large-scale, gradient-based, multidisciplinary design

Various computational methodologies relevant to large-scale multidisciplinary gradient-based optimization for engineering systems design problems are examined with emphasis on the situation where one or more discipline responses required by the optimized design procedure involve the solution of a system of nonlinear partial differential equations. Such situations occur when advanced CFD codes are applied in a multidisciplinary procedure for optimizing an aerospace vehicle design. A technique for satisfying the multidisciplinary design requirements for gradient information is presented. The technique is shown to permit some leeway in the CFD algorithms which can be used, an expansion to 3D problems, and straightforward use of other computational methodologies.

Newman, P. A.↗

Testing the Preliminary X-33 Navigation System

The X-33 Reusable Launch Vehicle (RLV) must meet the demanding requirements of landing autonomously on a narrow landing strip following a flight that reaches an altitude of up to 200,000 feet and a speed in excess of Mach 9 with significant in-flight energy bleed-off maneuvers. To execute this flight regimen a highly reliable avionics system has been designed that includes three LN-100G Inertial Navigation System/Global Positioning System (INS/GPS) units as the primary navigation system for the X-33. NASA's Marshall Space Flight Center (MSFC) tested an INS/GPS system in real-time simulations to determine the ability of this navigation suite to meet the in flight and autonomous landing requirements of the X-33 RLV. A total of sixty-one open loop tests were performed to characterize the navigation accuracy of the LN-100G. Twenty-seven closed-loop tests were also performed to evaluate the performance of the X-33 Guidance, Navigation and Control (GN&C) algorithms with the real navigation hardware. These closed-loop tests were also designed to expose any integration or operational issues with the real-time X-33 vehicle simulation. Dynamic road tests of the INS/GPS were conducted by Litton to assess the performance of differential and nondifferential INS/GPS hybrid navigation solutions. The results of the simulations and road testing demonstrate that this novel solution is capable of meeting the demanding requirements of take-off, in-flight navigation, and autonomous landing of the X-33 RLV. This paper describes the test environment developed to stimulate the LN-100G and discusses the results of this test effort. This paper also presents recommendations for a navigation system suitable to an operational RLV system.

Lomas, James J.↗

The MERiT Onboard the CeREs: A Novel Instrument to Study Energetic Particles in the Earth's Radiation Belts

The Miniaturized Electron pRoton Telescope, MERiT, is a low‐mass, low‐power, compact instrument using an innovative combination of particle detectors, sensor electronics, and onboard processing. MERiT is flying on the Compact Radiation belt Explorer, CeREs, a 3U CubeSat launched into a low earth orbit of 500‐km altitude and inclination of 85° on 16 December 2018. The primary and secondary science goals of CeREs are to investigate electron microbursts and to study solar particles. MERiT comprises a stack of solid state detectors (SSD) behind space facing avalanche photo diodes (APDs) surrounded by W‐Al shielding to reduce side‐penetrating particle background. The APD‐SSD combination enables measurement of electrons from 5 to 200 keV and 1 to 8 MeV; protons from 200–400 keV and 7–100 MeV in differential channels with energy resolution ΔE/E≈30% for both electrons and protons. MERiT measures microbursts with a high time resolution ranging from 4 to 16 ms and solar particles with a cadence of 1 s. MERiT energy channels and cadences are software configurable via algorithms and lookup tables residing on a field‐programmable gate array. The lookup tables can be changed via ground commands. MERiT geometry factor is 31 sq.cm‐sr and optimized to measure microbursts with the instrument viewing the local zenith in orbit. MERiT enables investigation of dynamical processes of radiation belt electron energization and loss, solar electron and proton transport, and their access to the Earth's polar caps. We describe the MERiT sensor design, calibration, operational modes, data products, and science goals.

Kanekal, S. G.↗

Efficient Floating-Point Arithmetic on Fault-Tolerant Quantum Computers

We propose a novel floating-point encoding scheme that builds on prior work involving fixed-point encodings. We encode floating-point numbers using Two's Complement fixed-point mantissas and Two's Complement integral exponents. We used our proposed approach to develop quantum algorithms for fundamental arithmetic operations, such as bit-shifting, reciprocation, multiplication, and addition. We prototyped and investigated the performance of the floating-point encoding scheme on quantum computer simulations by performing reciprocation on randomly drawn inputs and by solving first-order ordinary differential equations, while varying the number of qubits in the encoding. We observed rapid convergence to the exact solutions as we increased the number of qubits and a significant reduction in the number of ancilla qubits required for reciprocation when compared with similar approaches.

Serrallés, José Cruz [Weill Cornell Med. Coll.]↗

EXPANDED VERY LARGE ARRAY OBSERVATIONS OF GALACTIC SUPERNOVA REMNANTS: WIDE-FIELD CONTINUUM AND SPECTRAL-INDEX IMAGING

The radio continuum emission from the Galaxy has a rich mix of thermal and non-thermal emission. This very richness makes its interpretation challenging since the low radio opacity means that a radio image represents the sum of all emission regions along the line of sight. These challenges make the existing narrowband radio surveys of the Galactic plane difficult to interpret: e.g., a small region of emission might be a supernova remnant (SNR) or an H II region, or a complex combination of both. Instantaneous wide bandwidth radio observations in combination with the capability for high-resolution spectral-index mapping can be directly used to disentangle these effects. Here we demonstrate simultaneous continuum and spectral-index imaging capability at the full continuum sensitivity and resolution using newly developed wide-band wide-field imaging algorithms. Observations were conducted in the L and C bands with a total bandwidth of 1 and 2 GHz, respectively. We present preliminary results in the form of a full-field continuum image covering the wide-band sensitivity pattern of the EVLA centered on a large but poorly studied SNR (G55.7 + 3.4) and relatively narrower field continuum and spectral-index maps of three fields containing SNR and diffused thermal emission. We demonstrate that spatially resolved spectral-index maps differentiate regions with emission of different physical origins (spectral-index variation across composite SNRs and separation of thermal and non-thermal emission), superimposed along the line of sight. The wide-field image centered on the SNR G55.7+3.4 also demonstrates the excellent wide-field wide-band imaging capability of the EVLA.

79 ASTRONOMY AND ASTROPHYSICS↗

Fast and Scalable FFT-Based GPU-Accelerated Algorithms for Block-Triangular Toeplitz Matrices with Application to Linear Inverse Problems Governed by Autonomous Dynamical Systems

In this work, we present an efficient and scalable algorithm for performing matrix-vector multiplications (matvecs) for block Toeplitz matrices. Such matrices, which are shift-invariant with respect to their blocks, arise in the context of solving inverse problems governed by autonomous systems, and time-invariant systems in particular. In this article, we consider inverse problems that infer unknown parameters from observational data of a linear time-invariant dynamical system given in the form of partial differential equations (PDEs). Matrix-free Newton-conjugate-gradient methods are often the gold standard for solving these inverse problems, but they require numerous actions of the Hessian on a vector. Matrix-free adjoint-based Hessian matvecs require solution of a pair of linearized forward/adjoint PDE solves per Hessian action, which may be prohibitive for large-scale inverse problems. Time invariance of the forward PDE problem leads to a block Toeplitz structure of the discretized parameter-to-observable (p2o) map defining the mapping from inputs (parameters) to outputs (observables) of the PDEs. This block Toeplitz structure enables us to exploit two key properties: (1) compact storage of the p2o map and its adjoint, and (2) efficient fast Fourier transform–based Hessian matvecs. The proposed algorithm is mapped onto large multi-GPU clusters and achieves more than 80% of peak bandwidth on NVIDIA A100 GPUs. Excellent weak scaling is shown for up to 48 A100 GPUs. For the targeted problems, the implementation executes Hessian matvecs within fractions of a second, which is orders of magnitude faster than can be achieved by conventional matrix-free Hessian matvecs via forward/adjoint PDE solves.

97 MATHEMATICS AND COMPUTING↗

Special issue: Advanced stencil-code engineering

Here, stencil codes are compute-intensive algorithms, in which data points arranged in a large grid are being recomputed repeatedly from the values of data points in a predefined neighborhood. This fixed neighborhood pattern is called a stencil. Stencils codes see wide-spread use in computing the discrete solutions of partial differential equations and systems composed of such equations.

97 MATHEMATICS AND COMPUTING↗

Online randomized interpolative decomposition with a posteriori error estimator for temporal PDE data reduction

Traditional low-rank approximation is a powerful tool for compressing large data matrices that arise in simulations of partial differential equations (PDEs), but suffers from high computational cost and requires several passes over the PDE data. The compressed data may also lack interpretability thus making it difficult to identify feature patterns from the original data. Here, to address these issues, we present an online randomized algorithm to compute the interpolative decomposition (ID) of large-scale data matrices in situ. Compared to previous randomized IDs that used the QR decomposition to determine the column basis, we adopt a streaming ridge leverage score-based column subset selection algorithm that dynamically selects proper basis columns from the data and thus avoids an extra pass over the data to compute the coefficient matrix of the ID. In particular, we adopt a single-pass error estimator based on the non-adaptive Hutch++ algorithm to provide real-time error approximation for determining the best coefficients. As a result, our approach only needs a single pass over the original data and thus is suitable for large and high-dimensional matrices stored outside of core memory or generated in PDE simulations. A strategy to improve the accuracy of the reconstructed data gradient, when desired, within the ID framework is also presented. We provide numerical experiments on turbulent channel flow and ignition simulations, and on the NSTX Gas Puff Image dataset, comparing our algorithm with the offline ID algorithm to demonstrate its utility in real-world applications.

Column subset selection↗

Limited-memory adaptive snapshot selection for proper orthogonal decomposition

Reduced order models are useful for accelerating simulations in many-query contexts, such as optimization, uncertainty quantification, and sensitivity analysis. However, offline training of reduced order models can have prohibitively expensive memory and floating-point operation costs in high-performance computing applications, where memory per core is limited. To overcome this limitation for proper orthogonal decomposition, we propose a novel adaptive selection method for snapshots in time that limits offline training costs by selecting snapshots according an error control mechanism similar to that found in adaptive time-stepping ordinary differential equation solvers. The error estimator used in this work is related to theory bounding the approximation error in time of proper orthogonal decomposition-based reduced order models, and memory usage is minimized by computing the singular value decomposition using a single-pass incremental algorithm. Results for a viscous Burgers’ test problem demonstrate convergence in the limit as the algorithm error tolerances go to zero; in this limit, the full order model is recovered to within discretization error. The resulting method can be used on supercomputers to generate proper orthogonal decomposition-based reduced order models, or as a subroutine within hyperreduction algorithms that require taking snapshots in time, or within greedy algorithms for sampling parameter space.

97 MATHEMATICS AND COMPUTING↗

A Two-Wheel Observing Mode 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). Due to the MAP project's limited mass, power, and budget, a traditional reliability concept including fully redundant components was not feasible. The MAP design employs selective hardware redundancy, along with backup software modes and algorithms, to improve the odds of mission success. This paper describes the effort to develop a backup control mode, known as Observing II, that will allow the MAP science mission to continue in the event of a failure of one of its three reaction wheel assemblies. This backup science mode requires a change from MAP's nominal zero-momentum control system to a momentum-bias system. In this system, existing thruster-based control modes are used to establish a momentum bias about the sun line sufficient to spin the spacecraft up to the desired scan rate. Natural spacecraft dynamics exhibits spin and nutation similar to the nominal MAP science mode with different relative rotation rates, so the two reaction wheels are used to establish and maintain the desired nutation angle from the sun line. Detailed descriptions of the ObservingII control algorithm and simulation results will be presented, along with the operational considerations of performing the rest of MAP's necessary functions with only two wheels.

Starin, Scott R.↗