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 91 records · Page 5
All sourcesNASA NTRSDOE OSTI
NASA NTRS2011

Evolutionary Computational Methods for Identifying Emergent Behavior in Autonomous Systems

A technique based on Evolutionary Computational Methods (ECMs) was developed that allows for the automated optimization of complex computationally modeled systems, such as autonomous systems. The primary technology, which enables the ECM to find optimal solutions in complex search spaces, derives from evolutionary algorithms such as the genetic algorithm and differential evolution. These methods are based on biological processes, particularly genetics, and define an iterative process that evolves parameter sets into an optimum. Evolutionary computation is a method that operates on a population of existing computational-based engineering models (or simulators) and competes them using biologically inspired genetic operators on large parallel cluster computers. The result is the ability to automatically find design optimizations and trades, and thereby greatly amplify the role of the system engineer.

Terrile, Richard J.↗
NASA NTRS2019

Differences in the Vertical and Microphysical Evolution of Volcanic and Pyrocumulonimbus Stratospheric Aerosol Plumes as Observed by CALIOP and CATS Satellite Lidar

For some time, volcanic eruptions have been thought to be the only significant direct injectors of aerosols in the stratosphere. However, recent fire seasons have featured fire events resulting in large volcanic-sized pyrocumulonimbus plumes of smoke aerosols reaching many kilometers into the lower stratosphere. To understand and model the effects of these pyrocumulonimbus events on stratospheric composition and climate, a natural analogy lies with better studied volcanic events; however, differences in plume composition may limit this comparison. Using satellite lidar from both CATS and CALIOP, we show that the stratospheric aerosol plumes from the record-setting Pacific Northwest pyrocumulonimbus event of 2017 and the Calbuco volcanic eruption of 2015 evolve differently both vertically and microphysically. Specifically, depolarization ratios indicate that this pyrocumulonimbus event’s aerosol particles became more irregularly shaped over time in contrast to volcanic aerosols which become more spherical over time. Accounting for these changes in aerosol properties may be significant in assessing the effects of pyrocumulonimbus events on the Earth’s radiative balance and aid in refining stratospheric aerosol typing algorithms to differentiate volcanic from pyrocumulonimbus plumes.

Christian, Kenneth↗
NASA NTRS2019

Evaluation of a Multizone Impedance Eduction Method

A computational study is used to evaluate the PyCHE impedance eduction method developed at the NASA Langley Research Center. This method combines an aeroacoustic duct propagation code based on numerical solution to the convected Helmholtz equation with a global optimizer that uses the Differential Evolution algorithm. The efficacy of this method is evaluated with acoustic pressure data simulated to represent that measured with one-zone, two-zone, and three-zone liners mounted in the NASA Langley Grazing Flow Impedance Tube. The PyCHE method has a normalized impedance error of approximately 0.2 for (uniform) one-zone liners with a length of at least 5”, and produces quite reasonable results for liners as short as 2”. Whereas the impedance of the liner has an effect on eduction accuracy, the amount of attenuation is shown to be the dominant parameter. Similar results are observed for two-zone liners, for which the impedance of each zone is unique. The two-zone results also indicate it is more difficult to accurately educe resistance than reactance, and a zone length of at least 6” (slightly longer than for uniform liners) is needed to limit the normalized error to 0.2. The PyCHE method is also demonstrated to successfully educe the impedances for each zone of a three-zone liner. These results are sufficiently encouraging to warrant the continued usage of the PyCHE impedance eduction method for single and multizone liners.

Jones, M. G.↗
NASA NTRS2021

Development and Analysis of a Thick Cloud Layers Database for Lightning Launch Commit Criteria Improvement

Lightning can pose a potential threat to space launch vehicles. In response to this, rules were created called the Lightning Launch Commit Criteria (LLCC) that help weather personnel evaluate the potential for natural and rocket-triggered lightning. One of the ten LLCC with the least research is called the Thick Cloud Layers rule. To further understand electrification of thick cloud layers and potentially improve the Thick Cloud Layers rule, a database of thick cloud layers that occurred over the Eastern Range was created. This database is then used to create an algorithm for identifying and differentiating thick cloud layers from other cloud types based on radar characteristics, temperature levels in reference to cloud height, and the surface electric field. By analyzing and identifying thick cloud events, this project could help narrow down when thick clouds are occurring and potentially minimize unnecessary launch delays. Events that caused LLCC violations involving the Thick Cloud Layers rule were analyzed by hand using Level-2 NEXRAD radar data from the National Weather Service WSR-88D radar in Melbourne with the program GR2Analyst. Cases that were found to be isolated and not involved with convection were recorded (date, start/end time, location) in a database. Radar data associated with these cases was collected and gridded using Python radar packages. Once gridded, I calculated and recorded for each radar scan the following radar reflectivity driven parameters within an 11x11 km bin centered on each 1 square km grid point: the mean reflectivity colder than 0 degrees Celsius, Maximum Radar Reflectivity (MRR) colder than 0 degrees Celsius, Volume Averaged Height Integrated Radar Reflectivity (VAHIRR), Hydrometeor Identification (HID), the difference between the maximum and mean reflectivity, the cloud depth colder than 0 degrees Celsius, the overall cloud depth, the cloud top, and the cloud bottom. Soundings for each event were used to determine cloud temperature levels, and where the cloud is in relation to the freezing level. Electric field mill data collected over the Eastern Range was used to determine surface electric fields below each cloud. All parameters were analyzed in depth for several thick cloud cases to gain an understanding of typical thick cloud characteristics. Cases of thick clouds and other isolated cloud types were also recorded for training purposes to see if enough differences exist between cloud types to differentiate them with an algorithm. Each case along with its corresponding characteristics was recorded in a database, and this database was used to compare differing cloud types, as well as train the algorithm to detect thick clouds.

Lightning↗
NASA NTRS2004

The coverage of elliptical orbits using Ergodic theory

One of the key performance metrics for satellite constellations is the statistics of the visibility periods between the satellites and poins on the ground. Associated with this are other desirable communications statistis such as data through-put, link qualities, etc. Typically, the computation of coverage statistics requires the propagation of the trajectories. In this paper a new algorithm using differential geometry enables us to extend the Ergodic theory to eliptical orbits.

Lo, Martin W.↗
NASA NTRS2022

Tropospheric Ozone Retrieval By A Combination of TROPOMI/S5P Measurements With BASCOE Assimilated Data

We present a new tropospheric ozone dataset based on TROPOspheric Monitoring Instrument (TROPOMI)/Sentinel-5 Precursor (S5P) total ozone measurements combined with stratospheric ozone data from the Belgian Assimilation System for Chemical ObsErvations (BASCOE) constrained by assimilating ozone observations from the Microwave Limb Sounder (MLS). The BASCOE stratospheric data are interpolated to the S5P observations and subtracted from the TROPOMI total ozone data. The difference is equal to the tropospheric ozone residual column from the surface up to the tropopause. The tropospheric ozone columns are retrieved at the full spatial resolution of the TROPOMI sensor (5.5×3.5 km 2 ) with daily global coverage. Compared to the Ozone Mapping and Profiler Suite Modern-Era Retrospective analysis for Research and Applications 2 (OMPS-MERRA-2) data, a global mean positive bias of 3.3 DU is found for the analysed period April 2018 to June 2020. A small negative bias of about −0.91 DU is observed in the tropics relative to the operational TROPOMI tropical tropospheric data based on the convective cloud differential (CCD) algorithm throughout the same period. The new tropospheric ozone data (S5P-BASCOE) are compared to a set of globally distributed ozonesonde data integrated up to the tropopause level. We found 2254 comparisons with cloud-free TROPOMI observations within 25 km of the stations. In the global mean, S5P-BASCOE deviates by 2.6 DU from the integrated ozonesondes. Depending on the latitude the S5P-BASCOE deviate from the sondes and between −4.8 and 7.9 DU, indicating a good agreement. However, some exceptional larger positive deviations up to 12 DU are found, especially in the northern polar regions (north of 70∘). The monthly mean tropospheric column and time series for selected areas showed the expected spatial and temporal pattern, such as the wave one structure in the tropics or the seasonal cycle, including a summer maximum, in the mid-latitudes.

Satellite Retrieval↗
NASA NTRS2023

Bandelier Ecological Conservation: Mapping Invasive Species Along the Rio Grande Corridor in Bandelier National Monument

The Southwest U.S. has experienced a growth of invasive riparian species, specifically Elaeagnus angustifolia (Russian olive), Tamarix ramosissima (saltcedar), and Ulmus pumila (Siberian elm), which alter local soil chemistry and outcompete native species. Locating these exotic species is critical for ecological conservation; however, field identification can be resource intensive. NASA DEVELOP partnered with the National Park Service (NPS) at Bandelier National Monument (BAND) to assess the feasibility of using Earth observation data to map invasive species along the Rio Grande corridor of the park. The team used Landsat 8 OLI, Sentinel-2 MSI, and ISS DESIS imagery to compute principal components based on spectral bands, vegetation indices, and terrain indices. Using the first five principal components, the team created classification maps using both a k-means classification algorithm and a random forest algorithm to differentiate between native and non-native species. The team derived maps for the three invasive riparian species in the region for the last five years. The team found that invasive species covered 33% of the park's river corridor in 2023, and the invasive species extent has increased by 5.7% from 2019 to 2023. The methods will serve as a guide for aiding historic and present invasive species identification in riparian regions, and the NPS staff at BAND will use the results to inform local mitigation practices and advocate for invasive species removal.

Evan Barrett↗
NASA NTRS1983

Parallel architectures for iterative methods on adaptive, block structured grids

A parallel computer architecture well suited to the solution of partial differential equations in complicated geometries is proposed. Algorithms for partial differential equations contain a great deal of parallelism. But this parallelism can be difficult to exploit, particularly on complex problems. One approach to extraction of this parallelism is the use of special purpose architectures tuned to a given problem class. The architecture proposed here is tuned to boundary value problems on complex domains. An adaptive elliptic algorithm which maps effectively onto the proposed architecture is considered in detail. Two levels of parallelism are exploited by the proposed architecture. First, by making use of the freedom one has in grid generation, one can construct grids which are locally regular, permitting a one to one mapping of grids to systolic style processor arrays, at least over small regions. All local parallelism can be extracted by this approach. Second, though there may be a regular global structure to the grids constructed, there will be parallelism at this level. One approach to finding and exploiting this parallelism is to use an architecture having a number of processor clusters connected by a switching network. The use of such a network creates a highly flexible architecture which automatically configures to the problem being solved.

Gannon, D.↗
NASA NTRS1992

Numerical Asymptotic Solutions Of Differential Equations

Numerical algorithms derived and compared with classical analytical methods. In method, expansions replaced with integrals evaluated numerically. Resulting numerical solutions retain linear independence, main advantage of asymptotic solutions.

Thurston, Gaylen A.↗
NASA NTRS2003

Aerodynamic Shape Optimization Using Hybridized Differential Evolution

An aerodynamic shape optimization method that uses an evolutionary algorithm known at Differential Evolution (DE) in conjunction with various hybridization strategies is described. DE is a simple and robust evolutionary strategy that has been proven effective in determining the global optimum for several difficult optimization problems. Various hybridization strategies for DE are explored, including the use of neural networks as well as traditional local search methods. A Navier-Stokes solver is used to evaluate the various intermediate designs and provide inputs to the hybrid DE optimizer. The method is implemented on distributed parallel computers so that new designs can be obtained within reasonable turnaround times. Results are presented for the inverse design of a turbine airfoil from a modern jet engine. (The final paper will include at least one other aerodynamic design application). The capability of the method to search large design spaces and obtain the optimal airfoils in an automatic fashion is demonstrated.

Madavan, Nateri K.↗
NASA NTRS1994

Efficient Autonomous Learning for Statistical Pattern Recognition

We describe a neural network learning algorithm that implements differential learning in a generalized backpropagation framework. The algorithm regulates model complexity during the learning procedure, generating the best low-complexity approximation to the Bayer-optimal classifier allowed by the training sample.

Pattern↗
DOE OSTI2024

Differentiable vertex fitting for jet flavor tagging

This work explores the use of differentiable programming to integrate domain knowledge, in the form of domain specific software, into neural networks to develop scientific machine learning systems. We propose a differentiable vertex fitting algorithm that estimates the crossing point of multiple curves. In the high energy physics setting, these curves are defined by particle equations of motion and the crossing point represents the origin of particle production. This differentiable vertex fitting algorithm can be seamlessly integrated into neural networks, and we show its utility and efficacy in the high energy physics application of the classification of jets, i.e., collimated streams of particles in particle detectors whose originating parent particle we aim to classify. We demonstrate how differentiable vertex fitting can be integrated into larger transformer-based models for jet flavor tagging and show improvements in heavy flavor jet classification when compared to baseline models. Published by the American Physical Society 2024

Smith, Rachel E. C. (ORCID:0000000335851262)↗
NASA NTRS1981

A diagonal form of an implicit approximate-factorization algorithm

A modification of an implicit approximate-factorization finite-difference algorithm applied to partial differential equations is presented. This algorithm is applied to the two- and three-dimensional Euler equations in general curvilinear coordinates. The modification transforms the coupled system of equations into an uncoupled diagonal form that requires less computational work. For steady-state applications, the resulting diagonal algorithm retains the stability and accuracy characteristics of the original algorithm. The diagonal algorithm reduces the storage requirement of the implicit solution process and therefore has an important effect on the application of implicit finite-difference schemes to vector processors. Results are presented for realistic two-dimensional transonic flow fields about airfoils. Computation costs are reduced to 24-34%.

Pulliam, T. H.↗
NASA NTRS1989

Algorithms for computing the multivariable stability margin

Stability margin for multiloop flight control systems has become a critical issue, especially in highly maneuverable aircraft designs where there are inherent strong cross-couplings between the various feedback control loops. To cope with this issue, we have developed computer algorithms based on non-differentiable optimization theory. These algorithms have been developed for computing the Multivariable Stability Margin (MSM). The MSM of a dynamical system is the size of the smallest structured perturbation in component dynamics that will destabilize the system. These algorithms have been coded and appear to be reliable. As illustrated by examples, they provide the basis for evaluating the robustness and performance of flight control systems.

Tekawy, Jonathan A.↗
NASA NTRS1998

Trellises and Trellis-Based Decoding Algorithms for Linear Block Codes

A code trellis is a graphical representation of a code, block or convolutional, in which every path represents a codeword (or a code sequence for a convolutional code). This representation makes it possible to implement Maximum Likelihood Decoding (MLD) of a code with reduced decoding complexity. The most well known trellis-based MLD algorithm is the Viterbi algorithm. The trellis representation was first introduced and used for convolutional codes [23]. This representation, together with the Viterbi decoding algorithm, has resulted in a wide range of applications of convolutional codes for error control in digital communications over the last two decades. There are two major reasons for this inactive period of research in this area. First, most coding theorists at that time believed that block codes did not have simple trellis structure like convolutional codes and maximum likelihood decoding of linear block codes using the Viterbi algorithm was practically impossible, except for very short block codes. Second, since almost all of the linear block codes are constructed algebraically or based on finite geometries, it was the belief of many coding theorists that algebraic decoding was the only way to decode these codes. These two reasons seriously hindered the development of efficient soft-decision decoding methods for linear block codes and their applications to error control in digital communications. This led to a general belief that block codes are inferior to convolutional codes and hence, that they were not useful. Chapter 2 gives a brief review of linear block codes. The goal is to provide the essential background material for the development of trellis structure and trellis-based decoding algorithms for linear block codes in the later chapters. Chapters 3 through 6 present the fundamental concepts, finite-state machine model, state space formulation, basic structural properties, state labeling, construction procedures, complexity, minimality, and sectionalization of trellises. Chapter 7 discusses trellis decomposition and subtrellises for low-weight codewords. Chapter 8 first presents well known methods for constructing long powerful codes from short component codes or component codes of smaller dimensions, and then provides methods for constructing their trellises which include Shannon and Cartesian product techniques. Chapter 9 deals with convolutional codes, puncturing, zero-tail termination and tail-biting.Chapters 10 through 13 present various trellis-based decoding algorithms, old and new. Chapter 10 first discusses the application of the well known Viterbi decoding algorithm to linear block codes, optimum sectionalization of a code trellis to minimize computation complexity, and design issues for IC (integrated circuit) implementation of a Viterbi decoder. Then it presents a new decoding algorithm for convolutional codes, named Differential Trellis Decoding (DTD) algorithm. Chapter 12 presents a suboptimum reliability-based iterative decoding algorithm with a low-weight trellis search for the most likely codeword. This decoding algorithm provides a good trade-off between error performance and decoding complexity. All the decoding algorithms presented in Chapters 10 through 12 are devised to minimize word error probability. Chapter 13 presents decoding algorithms that minimize bit error probability and provide the corresponding soft (reliability) information at the output of the decoder. Decoding algorithms presented are the MAP (maximum a posteriori probability) decoding algorithm and the Soft-Output Viterbi Algorithm (SOVA) algorithm. Finally, the minimization of bit error probability in trellis-based MLD is discussed.

Lin, Shu↗
NASA NTRS1981

A Diagonal Form of an Implicit Approximate-Factorization Algorithm

A modification of an implicit approximate-factorization finite-difference algorithm applied to partial differential equations is presented. This algorithm is applied to the two- and three-dimensional Euler equations in general curvilinear coordinates. The modification transforms the coupled system of equations into an uncoupled diagonal form that requires less computational work. For steady-state applications, the resulting diagonal algorithm retains the stability and accuracy characteristics of the original algorithm. The diagonal algorithm reduces the storage requirement of the implicit solution process and therefore has an important effect on the application of implicit finite-difference schemes to vector processors. Results are presented for realistic two-dimensional transonic flow fields about airfoils. Computation costs are reduced 24-34%.

Pulliam, T. H.↗
NASA NTRS2002

Turbomachinery Airfoil Design Optimization Using Differential Evolution

An aerodynamic design optimization procedure that is based on a evolutionary algorithm known at Differential Evolution is described. Differential Evolution is a simple, fast, and robust evolutionary strategy that has been proven effective in determining the global optimum for several difficult optimization problems, including highly nonlinear systems with discontinuities and multiple local optima. The method is combined with a Navier-Stokes solver that evaluates the various intermediate designs and provides inputs to the optimization procedure. An efficient constraint handling mechanism is also incorporated. Results are presented for the inverse design of a turbine airfoil from a modern jet engine. The capability of the method to search large design spaces and obtain the optimal airfoils in an automatic fashion is demonstrated. Substantial reductions in the overall computing time requirements are achieved by using the algorithm in conjunction with neural networks.

Madavan, Nateri K.↗