Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “matrix factorization”

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 37 records · Page 2
All sourcesNASA NTRSDOE OSTI
DOE OSTI2022

Discovering hidden geothermal signatures using non-negative matrix factorization with customized k-means clustering

Discovery of hidden geothermal resources is challenging. It requires the mining of large datasets with diverse data attributes representing subsurface hydrogeological and geothermal conditions. The commonly used play fairway analysis approach typically incorporates subject-matter expertise to analyze regional data to estimate geothermal characteristics and favorability. We demonstrate an alternative approach based on machine learning (ML) to process a geothermal dataset from southwest New Mexico (SWNM). The study region includes low- and medium-temperature hydrothermal systems. Several of these systems are not well characterized because of insufficient existing data and limited past explorative work. This study discovers hidden patterns and relations in the SWNM geothermal dataset to improve our understanding of the regional hydrothermal conditions and energy-production favorability. This understanding is obtained by applying an unsupervised ML algorithm based on non-negative matrix factorization coupled with customized k-means clustering (NMFk). NMFk can automatically identify (1) hidden signatures characterizing analyzed datasets, (2) the optimal number of these signatures, (3) the dominant data attributes associated with each signature, and (4) the spatial distribution of the extracted signatures. Here, in this study, NMFk is applied to analyze 18 geological, geophysical, hydrogeological, and geothermal attributes at 44 locations in SWNM. Using NMFk, we find data patterns and identify the spatial associations of hydrothermal signatures within two physiographic provinces (Colorado Plateau and Basin and Range) and two sub-regions of these provinces (the Mogollon-Datil volcanic field and the Rio Grande rift) in SWNM. The ML algorithm extracted five hydrothermal signatures in the SWNM datasets that differentiate between low (<90°C) and medium (90-150°C)-temperature hydrothermal systems. The algorithm also suggests that the Rio Grande rift and northern Mogollon-Datil volcanic field are the most favorable regions for future geothermal resource discovery. NMFk also identified critical attributes to identify medium-temperature hydrothermal systems in the study area. The resulting NMFk model can be applied to predict geothermal conditions and their uncertainties at new SWNM locations based on limited data from unexplored regions. The code to execute the performed analyses as well as the corresponding data can be found at https://github.com/SmartTensors/GeoThermalCloud.jl.

15 GEOTHERMAL ENERGY↗
DOE OSTI2020

Distributed non-negative matrix factorization with determination of the number of latent features

The holistic analysis and understanding of the latent (that is, not directly observable) variables and patterns buried in large datasets is crucial for data-driven science, decision making and emergency response. Such exploratory analyses require devising unsupervised learning methods for data mining and extraction of the latent features, and non-negative matrix factorization (NMF) is one of the prominent such methods. NMF is based on compute-intense non-convex constrained minimization, which, for large datasets requires fast and distributed algorithms. However, current parallel implementations of NMF fail to estimate the number of latent features. In practice, identifying these features is both difficult and significant for pattern recognition and latent feature analysis, especially for large dense matrices. Here, we introduce a distributed NMF algorithm coupled with distributed custom clustering followed by a stability analysis on dense data, which we call DnMFk, to determine the number of latent variables. The results on synthetic data and the classical Swimmer data set demonstrate the accuracy of model determination while scaling nearly linearly across multiple processors for large data. Further, we employ DnMFk to determine the number of hidden features from a terabyte matrix.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Source identification by non-negative matrix factorization combined with semi-supervised clustering

Machine-learning methods and apparatus are provided to solve blind source separation problems with an unknown number of sources and having a signal propagation model with features such as wave-like propagation, medium-dependent velocity, attenuation, diffusion, and/or advection, between sources and sensors. In exemplary embodiments, multiple trials of non-negative matrix factorization are performed for a fixed number of sources, with selection criteria applied to determine successful trials. A semi-supervised clustering procedure is applied to trial results, and the clustering results are evaluated for robustness using measures for reconstruction quality and cluster separation. The number of sources is determined by comparing these measures for different trial numbers of sources. Source locations and parameters of the signal propagation model can also be determined. Disclosed methods are applicable to a wide range of spatial problems including chemical dispersal, pressure transients, and electromagnetic signals, and also to non-spatial problems such as cancer mutation.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Source identification by non-negative matrix factorization combined with semi-supervised clustering

Machine-learning methods and apparatus are provided to solve blind source separation problems with an unknown number of sources and having a signal propagation model with features such as wave-like propagation, medium-dependent velocity, attenuation, diffusion, and/or advection, between sources and sensors. In exemplary embodiments, multiple trials of non-negative matrix factorization are performed for a fixed number of sources, with selection criteria applied to determine successful trials. A semi-supervised clustering procedure is applied to trial results, and the clustering results are evaluated for robustness using measures for reconstruction quality and cluster separation. The number of sources is determined by comparing these measures for different trial numbers of sources. Source locations and parameters of the signal propagation model can also be determined. Disclosed methods are applicable to a wide range of spatial problems including chemical dispersal, pressure transients, and electromagnetic signals, and also to non-spatial problems such as cancer mutation.

Alexandrov, Boian S.↗
DOE OSTI2020

Modeling Aerial Gamma-Ray Backgrounds Using Non-negative Matrix Factorization

Airborne gamma-ray surveys are useful for many applications, ranging from geology and mining to public health and nuclear security. In all these contexts, the ability to decompose a measured spectrum into a linear combination of background source terms can provide useful insights into the data and lead to improvements in the techniques that use spectral energy windows. Multiple methods for the linear decomposition of spectra exist but are subject to various drawbacks, such as allowing negative photon fluxes or requiring detailed Monte Carlo modeling. In this work, we propose using non-negative matrix factorization (NMF) as a data-driven approach to spectral decomposition. Using aerial surveys that include flights over water, we demonstrate that the mathematical approach of NMF finds physically relevant structure in the aerial gamma-ray background, namely, that measured spectra can be expressed as the sum of nearby terrestrial emission, distant terrestrial emission, and radon and cosmic emission. Furthermore, these NMF background components are compared with the background components obtained by noise-adjusted singular value decomposition (NASVD), which contain negative photon fluxes and, thus, do not represent the emission spectra in as straightforward a way. Finally, we comment on the potential areas of research that are enabled by NMF decompositions, such as new approaches to spectral anomaly detection and data fusion.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2022

Source apportionment of VOCs, IVOCs and SVOCs by positive matrix factorization in suburban Livermore, California

Abstract. Gas- and particle-phase molecular markers provide highly specific information about the sources and atmospheric processes that contribute to air pollution. In urban areas, major sources of pollution are changing as regulation selectively mitigates some pollution sources and climate change impacts the surrounding environment. In this study, a comprehensive thermal desorption aerosol gas chromatograph (cTAG) was used to measure volatile, intermediate-volatility and semivolatile molecular markers every other hour over a 10 d period from 11 to 21 April 2018 in suburban Livermore, California. Source apportionment via positive matrix factorization (PMF) was performed to identify major sources of pollution. The PMF analysis identified 13 components, including emissions from gasoline, consumer products, biomass burning, secondary oxidation, aged regional transport and several factors associated with single compounds or specific events with unique compositions. The gasoline factor had a distinct morning peak in concentration but lacked a corresponding evening peak, suggesting commute-related traffic emissions are dominated by cold starts in residential areas. More monoterpene and monoterpenoid mass was assigned to consumer product emissions than biogenic sources, underscoring the increasing importance of volatile chemical products to urban emissions. Daytime isoprene concentrations were controlled by biogenic sunlight- and temperature-dependent processes, mediated by strong midday mixing, but gasoline was found to be the dominant and likely only source of isoprene at night. Biomass burning markers indicated residential wood burning activity remained an important pollution source even in the springtime. This study demonstrates that specific high-time-resolution molecular marker measurements across a wide range of volatility enable more comprehensive pollution source profiles than a narrower volatility range would allow.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRS1991

Efficient multitasking of Choleski matrix factorization on CRAY supercomputers

A Choleski method is described and used to solve linear systems of equations that arise in large scale structural analysis. The method uses a novel variable-band storage scheme and is structured to exploit fast local memory caches while minimizing data access delays between main memory and vector registers. Several parallel implementations of this method are described for the CRAY-2 and CRAY Y-MP computers demonstrating the use of microtasking and autotasking directives. A portable parallel language, FORCE, is used for comparison with the microtasked and autotasked implementations. Results are presented comparing the matrix factorization times for three representative structural analysis problems from runs made in both dedicated and multi-user modes on both computers. CPU and wall clock timings are given for the parallel implementations and are compared to single processor timings of the same algorithm.

Overman, Andrea L.↗
DOE OSTI2021

Fast Grain Mapping with Sub-Nanometer Resolution Using 4D-STEM with Grain Classification by Principal Component Analysis and Non-Negative Matrix Factorization

High-throughput grain mapping with sub-nanometer spatial resolution is demonstrated using scanning nanobeam electron diffraction (also known as 4D scanning transmission electron microscopy, or 4D-STEM) combined with high-speed direct-electron detection. An electron probe size down to 0.5 nm in diameter is used and the sample investigated is a gold–palladium nanoparticle catalyst. Computational analysis of the 4D-STEM data sets is performed using a disk registration algorithm to identify the diffraction peaks followed by feature learning to map the individual grains. Two unsupervised feature learning techniques are compared: principal component analysis (PCA) and non-negative matrix factorization (NNMF). The characteristics of the PCA versus NNMF output are compared and the potential of the 4D-STEM approach for statistical analysis of grain orientations at high spatial resolution is discussed.

47 OTHER INSTRUMENTATION↗
DOE OSTI2026

Using Filter Methods to Guide Convergence for ADMM, with Applications to Nonnegative Matrix Factorization Problems

Nonconvex, nonlinear optimization problems arise naturally in parameter fitting and machine learning. While augmented Lagrangian methods have demonstrated robust convergence for classes of these problems, their convergence for block updates has been relatively unexplored outside of the context of the alternating direction method of multipliers (ADMM). ADMM has seen extensive use in these applications, but may exhibit uncertain convergence behavior in many practical nonconvex settings, and struggles with general nonlinear constraints. In contrast, filter methods have proved effective in enforcing convergence for sequential quadratic programming methods and interior point methods with feasibility criteria. We develop an ADMM-filter method for highly nonlinear and nonconvex problems. Here, we show convergence under mild assumptions for several types of coordinate descent schemes, and demonstrate our algorithm on nonnegative matrix factorization and completion problems in imaging and chemical spectrum analysis.

Nonconvex optimization↗
DOE OSTI2023

Chemical identification of new particle formation and growth precursors through positive matrix factorization of ambient ion measurements

Abstract. In the lower troposphere, rapid collisions between ions and trace gases result in the transfer of positive charge to the highest proton affinity species and negative charge to the lowest proton affinity species. Measurements of the chemical composition of ambient ions thus provide direct insight into the most acidic and basic trace gases and their ion–molecule clusters – compounds thought to be important for new particle formation and growth. We deployed an atmospheric pressure interface time-of-flight mass spectrometer (APi-ToF) to measure ambient ion chemical composition during the 2016 Holistic Interactions of Shallow Clouds, Aerosols, and Land Ecosystems (HI-SCALE) campaign at the United States Department of Energy Atmospheric Radiation Measurement facility in the Southern Great Plains (SGP), an agricultural region. Cations and anions were measured for alternating periods of ∼ 24 h over 1 month. We use binned positive matrix factorization (binPMF) and generalized Kendrick analysis (GKA) to obtain information about the chemical formulas and temporal variation in ionic composition without the need for averaging over a long timescale or a priori high-resolution peak fitting. Negative ions consist of strong acids including sulfuric and nitric acid, organosulfates, and clusters of NO3- with highly oxygenated organic molecules (HOMs) derived from monoterpene (MT) and sesquiterpene (SQT) oxidation. Organonitrates derived from SQTs account for most of the HOM signal. Combined with the diel profiles and back trajectory analysis, these results suggest that NO3 radical chemistry is active at this site. SQT oxidation products likely contribute to particle growth at the SGP site. The positive ions consist of bases including alkyl pyridines and amines and a series of high-mass species. Nearly all the positive ions contained only one nitrogen atom and in general support ammonia and amines as being the dominant bases that could participate in new particle formation. Overall, this work demonstrates how APi-ToF measurements combined with binPMF analysis can provide insight into the temporal evolution of compounds important for new particle formation and growth.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
NASA NTRS2015

Nonnegative Matrix Factorization for Efficient Hyperspectral Image Projection

Hyperspectral imaging for remote sensing has prompted development of hyperspectral image projectors that can be used to characterize hyperspectral imaging cameras and techniques in the lab. One such emerging astronomical hyperspectral imaging technique is wide-field double-Fourier interferometry. NASA's current, state-of-the-art, Wide-field Imaging Interferometry Testbed (WIIT) uses a Calibrated Hyperspectral Image Projector (CHIP) to generate test scenes and provide a more complete understanding of wide-field double-Fourier interferometry. Given enough time, the CHIP is capable of projecting scenes with astronomically realistic spatial and spectral complexity. However, this would require a very lengthy data collection process. For accurate but time-efficient projection of complicated hyperspectral images with the CHIP, the field must be decomposed both spectrally and spatially in a way that provides a favorable trade-off between accurately projecting the hyperspectral image and the time required for data collection. We apply nonnegative matrix factorization (NMF) to decompose hyperspectral astronomical datacubes into eigenspectra and eigenimages that allow time-efficient projection with the CHIP. Included is a brief analysis of NMF parameters that affect accuracy, including the number of eigenspectra and eigenimages used to approximate the hyperspectral image to be projected. For the chosen field, the normalized mean squared synthesis error is under 0.01 with just 8 eigenspectra. NMF of hyperspectral astronomical fields better utilizes the CHIP's capabilities, providing time-efficient and accurate representations of astronomical scenes to be imaged with the WIIT.

NMF↗
DOE OSTI2025

Evaluation of data driven low-rank matrix factorization for accelerated solutions of the Vlasov equation

Low-rank methods have shown success in accelerating simulations of a collisionless plasma described by the Vlasov equation, but still rely on computationally costly linear algebra every time step. We propose a data-driven factorization method using artificial neural networks, specifically with convolutional layer architecture, that trains on existing simulation data. At inference time, the model outputs a low-rank decomposition of the distribution field of the charged particles, and we demonstrate that this step is faster than the standard linear algebra technique. Numerical experiments show that the method achieves comparable reconstruction accuracy for interpolation tasks, generalizing to unseen test data in a manner beyond just memorizing training data; patterns in factorization also inherently followed the same numerical trend as those within algebraic methods (e.g., truncated singular-value decomposition). However, when training on the first 70% of a time-series data and testing on the remaining 30%, the method fails to meaningfully extrapolate. Despite this limiting result, the technique may have benefits for simulations in a statistical steady-state or otherwise showing temporal stability. These results suggest that while the model offers a computationally efficient alternative for datasets with temporal stability, its current formulation is best suited for interpolation rather than for predicting future states in time-evolving systems. This study thus lays the groundwork for further refinement of neural network-based approaches to low-rank matrix factorization in high-dimensional plasma simulations.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS1988

Recursive mass matrix factorization and inversion: An operator approach to open- and closed-chain multibody dynamics

This report advances a linear operator approach for analyzing the dynamics of systems of joint-connected rigid bodies.It is established that the mass matrix M for such a system can be factored as M=(I+H phi L)D(I+H phi L) sup T. This yields an immediate inversion M sup -1=(I-H psi L) sup T D sup -1 (I-H psi L), where H and phi are given by known link geometric parameters, and L, psi and D are obtained recursively by a spatial discrete-step Kalman filter and by the corresponding Riccati equation associated with this filter. The factors (I+H phi L) and (I-H psi L) are lower triangular matrices which are inverses of each other, and D is a diagonal matrix. This factorization and inversion of the mass matrix leads to recursive algortihms for forward dynamics based on spatially recursive filtering and smoothing. The primary motivation for advancing the operator approach is to provide a better means to formulate, analyze and understand spatial recursions in multibody dynamics. This is achieved because the linear operator notation allows manipulation of the equations of motion using a very high-level analytical framework (a spatial operator algebra) that is easy to understand and use. Detailed lower-level recursive algorithms can readily be obtained for inspection from the expressions involving spatial operators. The report consists of two main sections. In Part 1, the problem of serial chain manipulators is analyzed and solved. Extensions to a closed-chain system formed by multiple manipulators moving a common task object are contained in Part 2. To retain ease of exposition in the report, only these two types of multibody systems are considered. However, the same methods can be easily applied to arbitrary multibody systems formed by a collection of joint-connected regid bodies.

Rodriguez, G.↗