Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “linear extrapolation method”

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 109 records · Page 6
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Collaborative Research: Properties and Dynamics of the Shallow Crust (Final Report)

Ground motions recorded at one location are often extrapolated to nearby regions within a given radius. Here, we explore the appropriateness of spatial extrapolation using data from seven small aperture seismic network deployments in southern California. Six of these deployments are linear arrays of 4-13 stations, and one is a 2D array of 13 stations at Pinyon Flats Observatory. The spatial footprint array diameters are 3 km or less, and each array was operational for a year or more. From our base catalog (M2.5+ earthquakes; 4038 events; September 2010 - June 2023), automated methods remove temporally overprinted waveforms from nearby events (< 5 km) in quick succession (< 5 min) and data with nonviable waveforms. These 200 samples per second data are filtered at 0.5-25 Hz and must have signal-tonoise ratios (SNRs) of 2.5+. Peak ground acceleration (PGA) and peak ground velocity (PGV) are derived individually from the maximum absolute values of each of the 3-component waveforms (vertical, northsouth, and east-west). Five of the seven arrays traverse the San Jacinto fault, and two do not. Ground motion observations are compared with theoretical estimates from Abrahamson et al., 2014. On average, arrays deployed within and across fault zones consistently record ground motions above theoretical expectations, whereas off-fault arrays record ground motions at or slightly below theoretical expectations. We attribute these differences to site conditions because these trends prevail for the full data suites. For each network and each individual channel, the coefficient of variation indicates that the standard deviations are ~30±6% of the mean. Exploring relative ground motion contributions from all three channels (ternary plots), as expected, most data show that vertical ground motions are attenuated compared to horizontal ones. However, this is not always the case for RA array data, where vertical motions can be ~2-3 times larger than horizontal motions for select events near Cahuilla, CA. These anomalously high vertical motions are focal mechanism-related. These results suggest that ternary plots created using only a small amount of data can be used as a data quality metric and a tool to find anomalous features in three-component data.

58 GEOSCIENCES↗
NASA NTRS2001

Monte Carlo predictions of DNA fragment-size distributions for large sizes after HZE particle irradiation

DSBs (double-strand breaks) produced by densely ionizing space radiation are not located randomly in the genome: recent data indicate DSB clustering along chromosomes. DSB clustering at large scales, from >100 Mbp down to approximately 2 kbp, is modeled using a Monte-Carlo algorithm. A random-walk model of chromatin is combined with a track model, that predicts the radial distribution of energy from an ion, and the RLC (randomly-located-clusters) formalism, in software called DNAbreak. This model generalizes the random-breakage model, whose broken-stick fragment-size distribution is applicable to low-LET radiation. DSB induction due to track interaction with the DNA volume depends on the radiation quality parameter Q. This dose-independent parameter depends only weakly on LET. Multi-track, high-dose effects depend on the cluster intensity parameter lambda, proportional to fluence as defined by the RLC formalism. After lambda is determined by a numerical experiment, the model reduces to one adjustable parameter Q. The best numerical fits to the experimental data, determining Q, are obtained. The knowledge of lambda and Q allows us to give biophysically based extrapolations of high-dose DNA fragment-size data to low doses or to high LETs.

NASA Center JSC↗
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

Calculation of non-stationary aerodynamic forces in the near sonic range

The development of a mathematical method for calculating nonstationary supersonic flow in the near-sonic range is described. A perturbation formula is derived based on the exact stationary values; it is applicable to the equation for potential. The problem can thus be divided into stationary and nonstationary fields. The pressure distribution in an oscillating profile is determined, based on hyperbolic differential equations. It is shown that there are important corollaries concerning the application of linear theory. With suitable extrapolations, linear theory can be used up to about Mach 0.8. Linear theory is not applicable, however, when determining the moment coefficients; for this case, a special technique is described.

Teipel, I.↗
NASA NTRS2001

Deriving the Coronal Magnetic Field Using Parametric Transformation Analysis

When plasma-beta greater than 1 then the gas pressure dominates over the magnetic pressure. This ratio as a function along the coronal magnetic field lines varies from beta greater than 1 in the photosphere at the base of the field lines, to beta much less than 1 in the mid-corona, to beta greater than 1 in the upper corona. Almost all magnetic field extrapolations do not or cannot take into account the full range of beta. They essentially assume beta much less than 1, since the full boundary conditions do not exist in the beta greater than 1 regions. We use a basic parametric representation of the magnetic field lines such that the field lines can be manipulated to match linear features in the EUV and SXR coronal images in a least squares sense. This research employs free-form deformation mathematics to generate the associated coronal magnetic field. In our research program, the complex magnetic field topology uses Parametric Transformation Analysis (PTA) which is a new and innovative method to describe the coronal fields that we are developing. In this technique the field lines can be viewed as being embedded in a plastic medium, the frozen-in-field-line concept. As the medium is deformed the field lines are similarly deformed. However the advantage of the PTA method is that the field line movement represents a transformation of one magnetic field solution into another magnetic field solution. When fully implemented, this method will allow the resulting magnetic field solution to fully match the magnetic field lines with EUV/SXR coronal loops by minimizing the differences in direction and dispersion of a collection of PTA magnetic field lines and observed field lines. The derived magnetic field will then allow beta greater than 1 regions to be included, the electric currents to be calculated, and the Lorentz force to be determined. The advantage of this technique is that the solution is: (1) independent of the upper and side boundary conditions, (2) allows non-vanishing magnetic forces, and (3) provides a global magnetic field solution, which contains high- and low-beta regimes and maximizes the similarity between the field lines structure and all the coronal images of the region. The coronal image analysis is crucial to the investigation and for the first time these images can be exploited to derive the coronal magnetic field in a well-posed mathematical formulation. This program is an outgrowth of an investigation in which an extrapolated potential field was required to be "inflated" in order to have the field lines match the Yohkoh/SXT images. The field lines were radially stretched resulting in a better match to the coronal loops of an active region. The PTA method of radial and non-radial deformations of field lines to provide a match to the EUV/SXR images will be presented.

Gary, G. Allen↗
NASA NTRS1983

Convergence and stability properties of minimal polynomial and reduced rank extrapolation algorithms

The minimal polynomial and reduced rank extrapolation algorithms are two acceleration of convergence methods for sequences of vectors. In a recent survey these methods were tested and compared with the scalar, vector, topological epsilon algorithms, and were observed to be more efficient than the latter. It was also observed that the two methods have similar convergence properties. The convergence and stability properties of these methods are analyzed and the performance of the acceleration methods when applied to a class of vector sequences that includes those sequences obtained from systems of linear equations by using matrix iterative methods is discussed.

Sidi, A.↗
NASA NTRS2010

Decline in Aerobic Fitness After Long-Term Stays on the International Space Station

U.S. and non-Russian International Partner astronauts who participate in long-term International Space Station (ISS) expeditions perform submaximal cycle exercise tests before, during, and after space flight. The heart rate (HR) and oxygen uptake (VO2) responses to exercise are used to estimate peak VO2 (EVO2pk). Purpose: To determine if the following factors are associated with the preflight-to-post flight change in EVO2pk: gender, age, body weight (BW), number of aerobic exercise sessions/wk- during flight, length of flight, EVO2pk measured before and late during the flight, ISS Expedition number and time between landing and the first post flight test. Methods: Records of 37 ISS astronauts (30 male, BW=81.6 plus or minus 8.6 kg; 7 female BW=66.1 plus or minus 4.9 kg [mean plus or minus SD]), age 46 plus or minus 4 years, were retrospectively examined. Peak HR and VO2 were measured approximately 9 months before flight to establish the test protocol. The submaximal cycle test consisted of three 5-minute stages designed to elicit 25, 50, and 75% of VO2pk. EVO2pk was calculated using linear least-squares extrapolation of average HR and VO2 during the last minute of each stage to predict VO2 at maximal HR. VO2 was not measured during flight and was assumed to not be different from preflight. Testing was performed 45 days before launch, late during flight, and during the week after landing. A random-intercept multivariate model was used to determine which characteristics significantly contributed to post flight EVO2pk. Results: In-flight aerobic exercise averaged 5.4 plus or minus 1.2 sessions/wk. ISS flight duration averaged 163 plus or minus 39 d. Mean EVO2pk values were 3.41 plus or minus 0.64 L (raised dot) per minute before flight, 3.09 plus or minus 0.57 L (raised dot) per minute late in flight, and 3.02 plus or minus 0.65 L (raised dot) per minute after flight. Late- and after-flight values were lower (p less than 0.05) than preflight values and did not differ from each other. Time between landing and post flight testing was 4.5 plus or minus 1.6 days. The only factor significantly associated with the post flight EVO2pk value was the late-flight EVO2pk score. Conclusion: Testing performed late during a mission provides a prediction of EVO2pk after landing. This approach may be implemented during longer missions.

Lynn, Peggy A.↗
DOE OSTI2024

Quantum error mitigation by layerwise Richardson extrapolation

A widely used method for mitigating errors in noisy quantum computers is Richardson extrapolation, a technique in which the overall effect of noise on the estimation of quantum expectation values is captured by a single parameter that, after being scaled to larger values, is eventually extrapolated to the zero-noise limit. We generalize this approach by introducing layerwise Richardson extrapolation (LRE), an error mitigation protocol in which the noise of different individual layers (or larger chunks of the circuit) is amplified and the associated expectation values are linearly combined to estimate the zero-noise limit. The coefficients of the linear combination are analytically obtained from the theory of multivariate Lagrange interpolation. LRE leverages the flexible configurational space of layerwise unitary folding, allowing for a more nuanced mitigation of errors by treating the noise level of each layer of the quantum circuit as an independent variable. Furthermore, we provide numerical simulations demonstrating scenarios where LRE achieves superior performance compared to traditional (single-variable) Richardson extrapolation.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2023

Temperature Uncertainty Modeling with Proxy Structural Data as Geostatistical Constraints for Well Siting: An Example Applied to Granite Springs Valley, NV, USA

Utilizing existing temperature and structural information around Granite Springs Valley, Nevada, we build 3D stochastic temperature models with the aim of evaluating the 3D uncertainty of temperature and choosing between candidate exploration well locations . The data used to support the modeling are measured temperatures and structural proxies from 3D geologic modeling, the latter considered "secondary" data. Two stochastic geostatistical techniques are explored for incorporating the structural proxies: cosimulation and local varying mean. With both the cosimulation and local varying mean methods, many equally likely temperature models (i.e., realizations) are produced, from which temperature probability profiles are calculated at candidate well locations. To aid in choosing between the candidate locations, two quantities summarize the temperature probabilities: Vprior and entropy. Vprior quantifies the likelihood for economic temperatures at each candidate location, whereas entropy identifies where new information has the most potential to reduce uncertainty. In general, the cosimulation realizations have smoother spatial structure, and extrapolate high temperatures at candidate locations that are located along the direction of the longest spatial correlation, which are down dip from existing temperature logs. The smooth realizations result in tight temperature probability profiles that are easier to interpret, but they have unrealistic temperature reversals in some locations because the cosimulation technique does not enforce a conductive geothermal gradient as a baseline (i.e., linearly increasing temperature with depth). The local varying mean results produce realizations with more realistic geothermal gradients, with temperatures increasing downward since a depth-temperature relationship is included. However, because they have much noisier spatial nature compared to cosimulation, it is harder to interpret the temperature probability profiles. The different local varying mean results allow the geologist to determine which proxy (e.g., dilation versus distance to fault termination) should be used given the specific geothermal system. In general, Vprior from local varying mean results identify locations that are close to high values for the structural proxies: areas with highe r probabilities for higher temperatures. The entropy results identify where uncertainty is greatest and therefore new drilling information could be most useful. Though these techniques provide useful information, even when applied to areas of sparse data, our comp arison of these two techniques demonstrates the need for new geothermal geostatistics techniques that combine the advantages of these two methods and that are tailored to the spatial uncertainty issues inherent in geothermal exploration.

3D temperature modeling↗
NASA NTRS1978

Interpolation and extrapolation of creep rupture data by the minimum commitment method. III - Analysis of multiheats

An outline is presented of approaches for treating multiheats on the basis of the focal point convergence method. The method has been employed in the case of two highly characterized multiheats, including a 304 stainless steel studied in Japan and a low alloy carbon steel studied in England. The method makes use of the same functional form for all materials. Only the constants are varied for each multiheat. Completely computerized procedures are employed for the determination of the constants. Once the basic analysis has been performed, the representation of various members in the same system is achieved by adding linear expressions of log stress, changing only two constants in the equations to represent a selected heat.

Manson, S. S.↗
DOE OSTI2023

Temperature uncertainty modelling with proxy structural data as geostatistical constraints for well siting: an example applied to Granite Springs Valley, NV, USA

Utilizing existing temperature and structural geology information around Granite Springs Valley, Nevada, we build 3D stochastic temperature models with the aims of evaluating the 3D uncertainty of temperature and choosing between candidate exploration well locations. The data used to support the modelling are measured temperatures and structural proxies from 3D geologic modelling (distance to fault, distance to fault intersections and terminations, Coulomb stress change and dilation tendency), the latter considered ‘secondary’ data. Two stochastic geostatistical techniques are explored for incorporating the structural proxies: cosimulation and local varying mean. With both the cosimulation and local varying mean methods, many equally-likely temperature models (i.e. realizations) are produced, from which temperature probability profiles are calculated at candidate well locations. To aid in choosing between the candidate locations, two quantities summarize the temperature probabilities: V prior and entropy. V prior quantifies the likelihood for economic temperatures at each candidate location, whereas entropy identifies where new information has the most potential to reduce uncertainty. In general, the cosimulation realizations have smoother spatial structure, and extrapolate high temperatures at candidate locations that are located along the direction of the longest spatial correlation, which are down dip from existing temperature logs. The smooth realizations result in tight temperature probability profiles that are easier to interpret, but they have unrealistic temperature reversals in some locations because of the dipping ellipsoid shape created and that the cosimulation technique does not enforce a conductive geothermal gradient as a baseline (i.e. linearly increasing temperature with depth). The local varying mean results produce realizations with more realistic geothermal gradients, with temperatures increasing downward since a depth-temperature relationship is included. However, because they have much noisier spatial nature compared to cosimulation, it is harder to interpret the temperature probability profiles. The different local varying mean results allow the geologist to determine which proxy (e.g. dilation v. distance to fault termination) should be used given the specific geothermal system. In general, V prior from local varying mean results identify locations that are close to high values for the structural proxies: areas with higher probabilities for higher temperatures. The entropy results identify where uncertainty is greatest and therefore new drilling information could be most useful. Though these techniques provide useful information, even when applied to areas of sparse data, our comparison of these two techniques demonstrates the need for new geothermal geostatistics techniques that combine the advantages of these two methods and that are tailored to the spatial uncertainty issues inherent in geothermal exploration.

15 GEOTHERMAL ENERGY↗
NASA NTRS1992

Rational approximations from power series of vector-valued meromorphic functions

Let F(z) be a vector-valued function, F: C yields C(sup N), which is analytic at z = 0 and meromorphic in a neighborhood of z = 0, and let its Maclaurin series be given. In this work we developed vector-valued rational approximation procedures for F(z) by applying vector extrapolation methods to the sequence of partial sums of its Maclaurin series. We analyzed some of the algebraic and analytic properties of the rational approximations thus obtained, and showed that they were akin to Pade approximations. In particular, we proved a Koenig type theorem concerning their poles and a de Montessus type theorem concerning their uniform convergence. We showed how optical approximations to multiple poles and to Laurent expansions about these poles can be constructed. Extensions of the procedures above and the accompanying theoretical results to functions defined in arbitrary linear spaces was also considered. One of the most interesting and immediate applications of the results of this work is to the matrix eigenvalue problem. In a forthcoming paper we exploited the developments of the present work to devise bona fide generalizations of the classical power method that are especially suitable for very large and sparse matrices. These generalizations can be used to approximate simultaneously several of the largest distinct eigenvalues and corresponding eigenvectors and invariant subspaces of arbitrary matrices which may or may not be diagonalizable, and are very closely related with known Krylov subspace methods.

Sidi, Avram↗
DOE OSTI2024

Unveiling the effect of composition on nuclear waste immobilization glasses’ durability by nonparametric machine learning

Abstract Ensuring the long-term chemical durability of glasses is critical for nuclear waste immobilization operations. Durable glasses usually undergo qualification for disposal based on their response to standardized tests such as the product consistency test or the vapor hydration test (VHT). The VHT uses elevated temperature and water vapor to accelerate glass alteration and the formation of secondary phases. Understanding the relationship between glass composition and VHT response is of fundamental and practical interest. However, this relationship is complex, non-linear, and sometimes fairly variable, posing challenges in identifying the distinct effect of individual oxides on VHT response. Here, we leverage a dataset comprising 654 Hanford low-activity waste (LAW) glasses across a wide compositional envelope and employ various machine learning techniques to explore this relationship. We find that Gaussian process regression (GPR), a nonparametric regression method, yields the highest predictive accuracy. By utilizing the trained model, we discern the influence of each oxide on the glasses’ VHT response. Moreover, we discuss the trade-off between underfitting and overfitting for extrapolating the material performance in the context of sparse and heterogeneous datasets.

36 MATERIALS SCIENCE↗
NASA NTRS1977

Linear magnetization feature associated with Rima Sirsalis

Measurements of the magnetic fields by the electron reflection method in the neighborhood of the long structural rille Rima Sirsalis show that a magnetic field with a strength of at least 100 nT (100 gammas) is present over a region of the order of 10 km in width and at least 300 km long. The center of the magnetized region closely parallels and is centered on the rille. The linear magnetization feature extends at least to latitude 8 deg S, 60 km beyond the place where the rille disappears at the edge of Oceanus Procellarum. This extension is coincident with the extrapolation of the rille based on photographs. However, the magnetization is much weaker or entirely absent at 5 deg S and has vanished at 0 deg latitude. These results suggest that the rille is indeed a structural feature and has associated with it magnetization, either in the form of intrusive magnetized rock or of a gap in an otherwise more or less uniformly magnetized layer of rock of large extent in two dimensions. Furthermore, the rille structure evidently is present for some distance beneath the lava flows of the Oceanus Procellarum basin.

Anderson, K. A.↗
NASA NTRS1993

Numerical Study of Base Pressure Characteristic Curve for a Four-Engine Clustered Nozzle Configuration

Excessive base heating has been a problem for many launch vehicles. For certain designs such as the direct dump of turbine exhaust in the nozzle section and at the nozzle lip of the Space Transportation Systems Engine (STME), the potential burning of the turbine exhaust in the base region has caused tremendous concern. Two conventional approaches have been considered for predicting the base environment: (1) empirical approach, and (2) experimental approach. The empirical approach uses a combination of data correlations and semi-theoretical calculations. It works best for linear problems, simple physics and geometry. However, it is highly suspicious when complex geometry and flow physics are involved, especially when the subject is out of historical database. The experimental approach is often used to establish database for engineering analysis. However, it is qualitative at best for base flow problems. Other criticisms include the inability to simulate forebody boundary layer correctly, the interference effect from tunnel walls, and the inability to scale all pertinent parameters. Furthermore, there is a contention that the information extrapolated from subscale tests with combustion is not conservative. One potential alternative to the conventional methods is computational fluid dynamics (CFD), which has none of the above restrictions and is becoming more feasible due to maturing algorithms and advancing computer technology. It provides more details of the flowfield and is only limited by computer resources. However, it has its share of criticisms as a predictive tool for base environment. One major concern is that CFD has not been extensively tested for base flow problems. It is therefore imperative that CFD be assessed and benchmarked satisfactorily for base flows. In this study, the turbulent base flowfield of a experimental investigation for a four-engine clustered nozzle is numerically benchmarked using a pressure based CFD method. Since the cold air was the medium, accurate prediction of the base pressure distributions at high altitudes is the primary goal. Other factors which may influence the numerical results such as the effects of grid density, turbulence model, differencing scheme, and boundary conditions are also being addressed. Preliminary results of the computed base pressure agreed reasonably well with that of the measurement. Basic base flow features such as the reverse jet, wall jet, recompression shock, and static pressure field in plane of impingement have been captured.

Wang, Ten-See↗
NASA NTRS1984

3-D inelastic analysis methods for hot section components (base program)

A 3-D inelastic analysis methods program consists of a series of computer codes embodying a progression of mathematical models (mechanics of materials, special finite element, boundary element) for streamlined analysis of combustor liners, turbine blades, and turbine vanes. These models address the effects of high temperatures and thermal/mechanical loadings on the local (stress/strain) and global (dynamics, buckling) structural behavior of the three selected components. These models are used to solve 3-D inelastic problems using linear approximations in the sense that stresses/strains and temperatures in generic modeling regions are linear functions of the spatial coordinates, and solution increments for load, temperature and/or time are extrapolated linearly from previous information. Three linear formulation computer codes, referred to as MOMM (Mechanics of Materials Model), MHOST (MARC-Hot Section Technology), and BEST (Boundary Element Stress Technology), were developed and are described.

Wilson, R. B.↗
NASA NTRS1993

An automatic multigrid method for the solution of sparse linear systems

An automatic version of the multigrid method for the solution of linear systems arising from the discretization of elliptic PDE's is presented. This version is based on the structure of the algebraic system solely, and does not use the original partial differential operator. Numerical experiments show that for the Poisson equation the rate of convergence of our method is equal to that of classical multigrid methods. Moreover, the method is robust in the sense that its high rate of convergence is conserved for other classes of problems: non-symmetric, hyperbolic (even with closed characteristics) and problems on non-uniform grids. No double discretization or special treatment of sub-domains (e.g. boundaries) is needed. When supplemented with a vector extrapolation method, high rates of convergence are achieved also for anisotropic and discontinuous problems and also for indefinite Helmholtz equations. A new double discretization strategy is proposed for finite and spectral element schemes and is found better than known strategies.

Shapira, Yair↗
DOE OSTI2026

Local reduced-order modeling for electrostatic plasmas by physics-informed solution manifold decomposition

Despite advancements in high-performance computing and modern numerical algorithms, computational cost remains prohibitive for multi-query kinetic plasma simulations. Here, in this work, we develop data-driven reduced-order models (ROMs) for collisionless electrostatic plasma dynamics, based on the kinetic Vlasov-Poisson equation. Our ROM approach projects the equation onto a linear subspace defined by the proper orthogonal decomposition (POD) modes. We introduce an efficient tensorial method to update the nonlinear term using a precomputed third-order tensor. We capture multiscale behavior with a minimal number of POD modes by decomposing the solution manifold into multiple time windows and creating temporally local ROMs. We consider two strategies for decomposition: one based on the physical time and the other based on the electric field energy. Applied to the 1D1V Vlasov–Poisson simulations, that is, prescribed E-field, Landau damping, and two-stream instability, we demonstrate that our ROMs accurately capture the total energy of the system both for parametric and time extrapolation cases. The temporally local ROMs are more efficient and accurate than the single ROM. In addition, in the two-stream instability case, we show that the energy-windowing reduced-order model (EW-ROM) is more efficient and accurate than the time-windowing reduced-order model (TW-ROM). With the tensorial approach, EW-ROM solves the equation approximately 90 times faster than Eulerian simulations while maintaining a maximum relative error of 7.5% for the training data and 11% for the testing data.

Electrostatic plasmas↗