Search NASA⌕ Search

SEARCH · Search NASA

Results for “convolutional”

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 397 records · Page 22
All sourcesNASA NTRSDOE OSTI
NASA NTRS1990

Performance comparison of combined ECC/RLL codes

In this paper, we present a performance comparison of several combined error correcting/run-lenth limited (ECC/RLL) codes created by concatenating a convolutional code with a run-length limited code. In each case, encoding and decoding are accomplished using a single trellis based on the combined code. Half of the codes under investigation use conventionally (d,k) run-length limited codes, where d is the minimum and k is the maximum allowable run of 0's between 1's. The other half of the combined codes use a special class of (d,k) codes known as distance preserving codes. These codes have the property that pairwise Hamming distances out of the (d,k) encoder are at least as large as the corresponding distances into the encoder (i.e., the codes preserve distance). Thus a combined code, created using a convolutional code concatenated with a distance preserving (d,k) code, will have a free distance (dfree) no smaller than the free distance of the original convolutional code. It should be noted that this does not hold if the (d,k) code was not distance preserving. A computer simulation is used to compare the performance of these two types of codes over the binary symmetric channel for various (d,k) constraints, rates, free distances, and numbers of states. Of particular interest for magnetic recording applications are codes with run-length constraints (1,3), (1,7), and (2,7).

French, C.↗
NASA NTRS1993

Enhanced decoding for the Galileo S-band mission

A coding system under consideration for the Galileo S-band low-gain antenna mission is a concatenated system using a variable redundancy Reed-Solomon outer code and a (14,1/4) convolutional inner code. The 8-bit Reed-Solomon symbols are interleaved to depth 8, and the eight 255-symbol codewords in each interleaved block have redundancies 64, 20, 20, 20, 64, 20, 20, and 20, respectively (or equivalently, the codewords have 191, 235, 235, 235, 191, 235, 235, and 235 8-bit information symbols, respectively). This concatenated code is to be decoded by an enhanced decoder that utilizes a maximum likelihood (Viterbi) convolutional decoder; a Reed Solomon decoder capable of processing erasures; an algorithm for declaring erasures in undecoded codewords based on known erroneous symbols in neighboring decodable words; a second Viterbi decoding operation (redecoding) constrained to follow only paths consistent with the known symbols from previously decodable Reed-Solomon codewords; and a second Reed-Solomon decoding operation using the output from the Viterbi redecoder and additional erasure declarations to the extent possible. It is estimated that this code and decoder can achieve a decoded bit error rate of 1 x 10(exp 7) at a concatenated code signal-to-noise ratio of 0.76 dB. By comparison, a threshold of 1.17 dB is required for a baseline coding system consisting of the same (14,1/4) convolutional code, a (255,223) Reed-Solomon code with constant redundancy 32 also interleaved to depth 8, a one-pass Viterbi decoder, and a Reed Solomon decoder incapable of declaring or utilizing erasures. The relative gain of the enhanced system is thus 0.41 dB. It is predicted from analysis based on an assumption of infinite interleaving that the coding gain could be further improved by approximately 0.2 dB if four stages of Viterbi decoding and four levels of Reed-Solomon redundancy are permitted. Confirmation of this effect and specification of the optimum four-level redundancy profile for depth-8 interleaving is currently being done.

Dolinar, S.↗
NASA NTRS1993

FD-TD modeling of 2-D dielectric waveguides for propagation and scattering of femtosecond optical solitons

Experimentalists have produced all-optical switches capable of 100-fs responses. To adequately model such switches, nonlinear effects in optical materials (both instantaneous and dispersive) must be included. In principle, the behavior of electromagnetic fields in nonlinear dielectrics can be determined by solving Maxwell's equations subject to the assumption that the electric polarization has a nonlinear relation to the electric field. However, until our previous work, the resulting nonlinear Maxwell's equations have not been solved directly. Rather, approximations have been made that result in a class of generalized nonlinear Schrodinger equations (GNLSE) that solve only for the envelope of the optical pulses. In this paper, we present first-time calculations from the vector nonlinear Maxwell's equations of femtosecond soliton propagation and scattering, including carrier waves, in two-dimensional systems of dielectric waveguides exhibiting the Kerr and Raman quantum effects. We use the finite-difference time-domain (FD-TD) method in an extension of our 1-D work. There, in a fundamental innovation, we treated the linear and nonlinear convolutions for the electric polarization as new dependent variables. By differentiating these convolutions in the time domain, we derived an equivalent system of coupled, nonlinear second-order ODE's. These equations together with Maxwell's equations form the system that is solved to determine the electromagnetic fields in inhomogeneous nonlinear dispersive media. Backstorage in time is limited to only that needed by the time-integration algorithm for the ODE's, rather than that needed to store the time-history of the kernel functions of the convolutions (1000-10,000 time steps). Thus, a 2-D nonlinear optics model from Maxwell's equations is now feasible.

Joseph, Rose↗
NASA NTRS1994

Error control techniques for satellite and space communications

The unequal error protection capabilities of convolutional and trellis codes are studied. In certain environments, a discrepancy in the amount of error protection placed on different information bits is desirable. Examples of environments which have data of varying importance are a number of speech coding algorithms, packet switched networks, multi-user systems, embedded coding systems, and high definition television. Encoders which provide more than one level of error protection to information bits are called unequal error protection (UEP) codes. In this work, the effective free distance vector, d, is defined as an alternative to the free distance as a primary performance parameter for UEP convolutional and trellis encoders. For a given (n, k), convolutional encoder, G, the effective free distance vector is defined as the k-dimensional vector d = (d(sub 0), d(sub 1), ..., d(sub k-1)), where d(sub j), the j(exp th) effective free distance, is the lowest Hamming weight among all code sequences that are generated by input sequences with at least one '1' in the j(exp th) position. It is shown that, although the free distance for a code is unique to the code and independent of the encoder realization, the effective distance vector is dependent on the encoder realization.

Costello, Daniel J., Jr.↗
NASA NTRS1999

Investigation of Near Shannon Limit Coding Schemes

Turbo codes can deliver performance that is very close to the Shannon limit. This report investigates algorithms for convolutional turbo codes and block turbo codes. Both coding schemes can achieve performance near Shannon limit. The performance of the schemes is obtained using computer simulations. There are three sections in this report. First section is the introduction. The fundamental knowledge about coding, block coding and convolutional coding is discussed. In the second section, the basic concepts of convolutional turbo codes are introduced and the performance of turbo codes, especially high rate turbo codes, is provided from the simulation results. After introducing all the parameters that help turbo codes achieve such a good performance, it is concluded that output weight distribution should be the main consideration in designing turbo codes. Based on the output weight distribution, the performance bounds for turbo codes are given. Then, the relationships between the output weight distribution and the factors like generator polynomial, interleaver and puncturing pattern are examined. The criterion for the best selection of system components is provided. The puncturing pattern algorithm is discussed in detail. Different puncturing patterns are compared for each high rate. For most of the high rate codes, the puncturing pattern does not show any significant effect on the code performance if pseudo - random interleaver is used in the system. For some special rate codes with poor performance, an alternative puncturing algorithm is designed which restores their performance close to the Shannon limit. Finally, in section three, for iterative decoding of block codes, the method of building trellis for block codes, the structure of the iterative decoding system and the calculation of extrinsic values are discussed.

Kwatra, S. C.↗
NASA NTRS2000

Tera-Ops Processing for ATR

A three-dimensional microelectronic device (3DANN-R) capable of performing general image convolution at the speed of 1012 operations/second (ops) in a volume of less than 1.5 cubic centimeter has been successfully built under the BMDO/JPL VIGILANTE program. 3DANN-R was developed in partnership with Irvine Sensors Corp., Costa Mesa, California. 3DANN-R is a sugar-cube-sized, low power image convolution engine that in its core computation circuitry is capable of performing 64 image convolutions with large (64x64) windows at video frame rates. This paper explores potential applications of 3DANN-R such as target recognition, SAR and hyperspectral data processing, and general machine vision using real data and discuss technical challenges for providing deployable systems for BMDO surveillance and interceptor programs.

Udomkesmalee, Suraphol↗
NASA NTRS2003

An Approximate Dissipation Function for Large Strain Rubber Thermo-Mechanical Analyses

Mechanically induced viscoelastic dissipation is difficult to compute. When the constitutive model is defined by history integrals, the formula for dissipation is a double convolution integral. Since double convolution integrals are difficult to approximate, coupled thermo-mechanical analyses of highly viscous rubber-like materials cannot be made with most commercial finite element software. In this study, we present a method to approximate the dissipation for history integral constitutive models that represent Maxwell-like materials without approximating the double convolution integral. The method requires that the total stress can be separated into elastic and viscous components, and that the relaxation form of the constitutive law is defined with a Prony series. Numerical data is provided to demonstrate the limitations of this approximate method for determining dissipation. Rubber cylinders with imbedded steel disks and with an imbedded steel ball are dynamically loaded, and the nonuniform heating within the cylinders is computed.

Johnson, Arthur R.↗
NASA NTRS2003

Evaluation of Genetic Algorithm Concepts Using Model Problems: Multi-Objective Optimization - Part 2

A genetic algorithm approach suitable for solving multi-objective optimization problems is described and evaluated using a series of simple model problems. Several new features including a binning selection algorithm and a gene-space transformation procedure are included. The genetic algorithm is suitable for finding pareto optimal solutions in search spaces that are defined by any number of genes and that contain any number of local extrema. Results indicate that the genetic algorithm optimization approach is flexible in application and extremely reliable, providing optimal results for all optimization problems attempted. The binning algorithm generally provides pareto front quality enhancements and moderate convergence efficiency improvements for most of the model problems. The gene-space transformation procedure provides a large convergence efficiency enhancement for problems with non-convoluted pareto fronts and a degradation in efficiency for problems with convoluted pareto fronts. The most difficult problems --multi-mode search spaces with a large number of genes and convoluted pareto fronts-- require a large number of function evaluations for GA convergence, but always converge.

Holst, Terry L.↗
NASA NTRS2004

An Evaluation of Quantitative Methods of Determining the Degree of Melting Experienced by a Chondrule

Many workers have considered the degree to which partial melting occurred in chondrules they have studied, and this has led to attempts to find reliable methods of determining the degree of melting. At least two quantitative methods have been used in the literature: a convolution index (CVI), which is a ratio of the perimeter of the chondrule as seen in thin section divided by the perimeter of a circle with the same area as the chondrule, and nominal grain size (NGS), which is the inverse square root of the number density of olivines and pyroxenes in a chondrule (again, as seen in thin section). We have evaluated both nominal grain size and convolution index as melting indicators. Nominal grain size was measured on the results of a set of dynamic crystallization experiments previously described, where aliquots of LEW97008(L3.4) were heated to peak temperatures of 1250, 1350, 1370, and 1450 C, representing varying degrees of partial melting of the starting material. Nominal grain size numbers should correlate with peak temperature (and therefore degree of partial melting) if it is a good melting indicator. The convolution index is not directly testable with these experiments because the experiments do not actually create chondrules (and therefore they have no outline on which to measure a CVI). Thus we had no means to directly test how well the CVI predicted different degrees of melting. Therefore, we discuss the use of the CVI measurement and support the discussion with X-ray Computed Tomography (CT) data.

Nettles, J. W.↗
NASA NTRS1998

Attachment of Free Filament Thermocouples for Temperature Measurements on Ceramic Matrix Composites

At the NASA Lewis Research Center, a new installation technique utilizing convoluted wire thermocouples (TC's) was developed and proven to produce very good adhesion on CMC's, even in a burner rig environment. Because of their unique convoluted design, such TC's of various types and sizes adhere to flat or curved CMC specimens with no sign of delamination, open circuits, or interactions-even after testing in a Mach 0.3 burner rig to 1200 C (2200 F) for several thermal cycles and at several hours at high temperatures. Large differences in thermal expansion between metal thermocouples and low-expansion materials, such as CMC's, normally generate large stresses in the wires. These stresses cause straight wires to detach, but convoluted wires that are bonded with strips of coating allow bending in the unbonded portion to relieve these expansion stresses.

Lei, Jih-Fen↗
NASA NTRS2011

Optimized FPGA Implementation of Multi-Rate FIR Filters Through Thread Decomposition

Multi-rate finite impulse response (MRFIR) filters are among the essential signal-processing components in spaceborne instruments where finite impulse response filters are often used to minimize nonlinear group delay and finite precision effects. Cascaded (multistage) designs of MRFIR filters are further used for large rate change ratio in order to lower the required throughput, while simultaneously achieving comparable or better performance than single-stage designs. Traditional representation and implementation of MRFIR employ polyphase decomposition of the original filter structure, whose main purpose is to compute only the needed output at the lowest possible sampling rate. In this innovation, an alternative representation and implementation technique called TD-MRFIR (Thread Decomposition MRFIR) is presented. The basic idea is to decompose MRFIR into output computational threads, in contrast to a structural decomposition of the original filter as done in the polyphase decomposition. A naive implementation of a decimation filter consisting of a full FIR followed by a downsampling stage is very inefficient, as most of the computations performed by the FIR state are discarded through downsampling. In fact, only 1/M of the total computations are useful (M being the decimation factor). Polyphase decomposition provides an alternative view of decimation filters, where the downsampling occurs before the FIR stage, and the outputs are viewed as the sum of M sub-filters with length of N/M taps. Although this approach leads to more efficient filter designs, in general the implementation is not straightforward if the numbers of multipliers need to be minimized. In TD-MRFIR, each thread represents an instance of the finite convolution required to produce a single output of the MRFIR. The filter is thus viewed as a finite collection of concurrent threads. Each of the threads completes when a convolution result (filter output value) is computed, and activated when the first input of the convolution becomes available. Thus, the new threads get spawned at exactly the rate of N/M, where N is the total number of taps, and M is the decimation factor. Existing threads retire at the same rate of N/M. The implementation of an MRFIR is thus transformed into a problem to statically schedule the minimum number of multipliers such that all threads can be completed on time. Solving the static scheduling problem is rather straightforward if one examines the Thread Decomposition Diagram, which is a table-like diagram that has rows representing computation threads and columns representing time. The control logic of the MRFIR can be implemented using simple counters. Instead of decomposing MRFIRs into subfilters as suggested by polyphase decomposition, the thread decomposition diagrams transform the problem into a familiar one of static scheduling, which can be easily solved as the input rate is constant.

Kobayashi, Kayla N.↗
NASA NTRSDate not supplied

Dynamic Bellows for a Pulse Tube Cryocooler Application

Electrified aircraft propulsion systems will require powertrain components that are both lightweight and highly efficient. NASA’s 1.4 MW High Efficiency Megawatt Motor (HEMM) has been designed to run at 6800 RPM with >98% efficiency and 16 kW/kg, achieving low losses by incorporating a cryocooler within its rotor shaft to cool superconducting DC windings. A pulse-tube cryocooler typically contains an oscillating piston that drives acoustic waves; however, leakage around the piston reduces the efficiency of the cryocooler. A metal bellows was designed to replace the piston to allow for little to no leakage. The requirements for this bellows were extreme in terms of displacement amplitude, oscillating pressure amplitude, temperature, and frequency. It was determined that an off-the-shelf edge-welded bellows would not work for this application, so a custom bellows was designed to include full convolutions welded only on their outer diameters where the stress is lower. Several high-fatigue-strength materials were considered, including 17-4PH H900 and Aermet 100. The final bellows design includes 74 convolutions with a variable inner diameter that helps distribute and reduce the dynamic stress throughout the length rather than concentrating the high stress in one location. To test the concept, a short 10-convolution bellows with constant inner diameter was designed for testing. One short bellows test article was fabricated with the outer diameter weld, and one by 3D printing. Both will be fatigue-tested with the goal of achieving 10^7 cycles.

Bellows↗
DOE OSTI2025

Revisiting single inclusive jet production: timelike factorization and reciprocity

Factorization theorems for single inclusive jet production play a crucial role in the study of jets and their substructure. In the case of small radius jets, the dynamics of the jet clustering can be factorized from both the hard production dynamics, and the dynamics of the low scale jet substructure measurement, and is described by a matching coefficient that can be computed in perturbative Quantum Chromodynamics (QCD). A proposed factorization formula describing this process has been previously presented in the literature, and is referred to as the semi-inclusive, or fragmenting jets formalism. By performing an explicit two-loop calculation, we show the inconsistency of this factorization formula, in agreement with another recent result in the literature. Building on recent progress in the factorization of single logarithmic observables, and the understanding of reciprocity, we then derive a new all-order factorization theorem for inclusive jet production. The use of a jet algorithm, being only a modification of the infrared structure of the measurement, modifies the structure of convolutions in the factorization theorem, as compared to inclusive fragmentation, but maintains the universality of the inclusive hard function and its associated Dokshitzer-Gribov-Lipatov-Altarelli-Parisi (DGLAP) evolution, which are ultraviolet properties. However, the non-trivial structure of convolutions in the factorization theorem implies that the jet functions exhibit a modified evolution. We perform an explicit two-loop calculation of the jet function in both N = 4 super Yang-Mills (SYM), and for all color channels in QCD, finding exact agreement with the structure derived from our renormalization group equations. In addition, we derive several new results, including an extension of our factorization formula to jet substructure observables, a jet algorithm definition of a generating function for the energy correlators, and new results for exclusive jet functions. Our results are a key ingredient for achieving precision jet substructure at colliders.

Effective Field Theories↗
DOE OSTI2025

The effective number of parameters in kernel density estimation

We devise a new formula for measuring the effective degrees of freedom (EDoF) in kernel density estimation (KDE). Starting from the orthogonal polynomial sequence (OPS) expansion for the ratio of the empirical to the oracle density, we show how convolution with the kernel leads to a new OPS with respect to which one may express the resulting KDE. The expansion coefficients of the two OPS systems can then be related via a kernel sensitivity matrix, which leads to a natural oracle definition of EDoF through the trace operator. Asymptotic properties of the (empirical) plug-in EDoF are worked out through influence functions, and connections with other empirical EDoFs are established. Minimization of Kullback-Leibler divergence is investigated as an alternative to integrated squared error based bandwidth selection rules, yielding a new normal scale rule. The methodology, which arises from a proper oracle formulation and is not restricted to convolution kernels, suggests the possibility of a new bandwidth selection rule based on an information criterion such as AIC.

bandwidth selection↗
DOE OSTI2025

Scalable training of trustworthy and energy-efficient predictive graph foundation models for atomistic materials modeling: a case study with HydraGNN

We present our work on developing and training scalable, trustworthy, and energy-efficient predictive graph foundation models (GFMs) using HydraGNN, a multi-headed graph convolutional neural network architecture. HydraGNN expands the boundaries of graph neural network (GNN) computations in both training scale and data diversity. It abstracts over message passing algorithms, allowing both reproduction of and comparison across algorithmic innovations that define nearest-neighbor convolution in GNNs. This work discusses a series of optimizations that have allowed scaling up the GFMs training to tens of thousands of GPUs on datasets consisting of hundreds of millions of graphs. Our GFMs use multitask learning (MTL) to simultaneously learn graph-level and node-level properties of atomistic structures, such as energy and atomic forces. Using over 154 million atomistic structures for training, we illustrate the performance of our approach along with the lessons learned on two state-of-the-art US Department of Energy (US-DOE) supercomputers, namely the Perlmutter petascale system at the National Energy Research Scientific Computing Center and the Frontier exascale system at Oak Ridge Leadership Computing Facility. The HydraGNN architecture enables the GFM to achieve near-linear strong scaling performance using more than 2000 GPUs on Perlmutter and 16,000 GPUs on Frontier.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2025

Online thermal profile prediction for large format additive manufacturing: A hybrid CNN-LSTM based approach

Large format additive manufacturing (LFAM) is an advanced 3D printing technique that efficiently fabricates large-scale components through a layer-by-layer extrusion and deposition process. Accurate surface layer temperature monitoring is essential to prevent manufacturing failures and ensure final product quality. Traditional physics-based offline approaches for simulating thermal behavior are often inefficient and complex, posing challenges on real-time, in-situ monitoring. Here, to address this, we propose a data-driven hybrid CNN-LSTM model to predict sequential thermal images of arbitrary length using real-time infrared thermal imaging. In this approach, a Convolutional Neural Networks (CNN) is trained offline to capture spatial features, reduce dimensional complexity, and enhance time efficiency, while a stacked Long Short-Term Memory (LSTM) is applied online to capture temporal information for improved prediction of future thermal behavior in subsequent printing layers. Model performance is evaluated using MSE, SSIM, and PSNR metrics and is benchmarked against stacked LSTM and convolutional LSTM models, demonstrating superior accuracy and applicability. Additionally, to mitigate noise from moving extruders and gantry backgrounds in thermal images, a fine-tuned semantic segmentation model is implemented offline to extract printing geometry, enabling precise temperature tracking along the tool path for further thermal analysis. The frameworks developed in this study significantly advance temperature monitoring, thermal analysis, and in-situ manufacturing control for LFAM, bridging the gap between theoretical modeling and practical application.

Geometry extraction↗
DOE OSTI2025

A Green’s function fast multipole method for computation of micromechanical fields in heterogeneous materials

Computation of micromechanical fields in heterogeneous materials is usually performed using either the finite element method or the Green’s function method based on FFTs. The finite element method allows for accurate discretization and for non-periodic boundary conditions but is computationally expensive. On the other hand, the FFT-based method is computationally efficient but requires discretization on a regular grid of hexahedral voxels. In this paper, a Green’s function method allowing for accurate discretization using tetrahedral elements and for non-periodic boundary conditions is proposed. The convolution is computed using the fast multipole method, which provides good accuracy even for low-order expansion due to the fast decay of interactions between elements. The proposed Green’s function fast multipole method is verified by comparison with analytical and FFT-based solutions. Furthermore, the computational time is analyzed and compared to the FFT-based method for non-periodic convolution. Finally, effective properties of an elastic polycrystalline microstructure containing thin intergranular cracks are computed and analyzed.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Analyzing inference workloads for spatiotemporal modeling

Ensuring power grid resiliency, forecasting climate conditions, and optimization of transportation infrastructure are some of the many application areas where data is collected in both space and time. Spatiotemporal modeling is about modeling those patterns for forecasting future trends and carrying out critical decision-making by leveraging machine learning/deep learning. Once trained offline, field deployment of trained models for near real-time inference could be challenging because performance can vary significantly depending on the environment, available compute resources and tolerance to ambiguity in results. Users deploying spatiotemporal models for solving complex problems can benefit from analytical studies considering a plethora of system adaptations to understand the associated performance-quality trade-offs. To facilitate the co-design of next-generation hardware architectures for field deployment of trained models, it is critical to characterize the workloads of these deep learning (DL) applications during inference and assess their computational patterns at different levels of the execution stack. In this paper, we develop several variants of deep learning applications that use spatiotemporal data from dynamical systems. We study the associated computational patterns for inference workloads at different levels, considering relevant models (Long short-term Memory, Convolutional Neural Network and Spatio-Temporal Graph Convolution Network), DL frameworks (Tensorflow and PyTorch), precision (FP16, FP32, AMP, INT16 and INT8), inference runtime (ONNX and AI Template), post-training quantization (TensorRT) and platforms (Nvidia DGX A100 and Sambanova SN10 RDU). Overall, our findings indicate that although there is potential in mixed-precision models and post-training quantization for spatiotemporal modeling, extracting efficiency from contemporary GPU systems might be challenging. Instead, co-designing custom accelerators by leveraging optimized High Level Synthesis frameworks (such as SODA High-Level Synthesizer for customized FPGA/ASIC targets) can make workload-specific adjustments to enhance the efficiency.

97 MATHEMATICS AND COMPUTING↗