Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “probability generating function”

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 199 records · Page 11

DNA Break Clustering as a Predictor of Cell Death across Various Radiation Qualities: Influence of Cell Size, Cell Asymmetry, and Beam Orientation

Cosmic radiation, composed of high charge and energy (HZE) particles, causes cellular DNA damage that can result in cell death or mutation that can evolve into cancer. In this work, a cell death model is applied to several cell lines exposed to HZE ions spanning a broad range of linear energy transfer (LET) values. We hypothesize that chromatin movement leads to the clustering of multiple double strand breaks (DSB) within one radiation-induced foci (RIF). The survival probability of a cell population is determined by averaging the survival probabilities of individual cells, which is function of the number of pairwise DSB interactions within RIF. The simulation code RITCARD was used to compute DSB. Two clustering approaches were applied to determine the number of RIF per cell. RITCARD outputs were combined with experimental data from four normal human cell lines to derive the model parameters and expand its predictions in response to ions with LET ranging from ∼0.2keV/μmto∼3000keV/μm. Spherical and ellipsoidal nuclear shapes and two ion beam orientations were modeled to assess the impact of geometrical properties on cell death. The calculated average number of RIF per cell reproduces the saturation trend for high doses and high-LET values that is usually experimentally observed. The cell survival model generates the recognizable bell shape of LET dependence for the relative biological effectiveness (RBE). At low LET, smaller nuclei have lower survival due to increased DNA density and DSB clustering. At high LET, nuclei with a smaller irradiation area either because of a smaller size or a change in beam orientation have a higher survival rate due to a change in the distribution of DSB/RIF per cell. If confirmed experimentally, the geometric characteristics of cells would become a significant factor in predicting radiation-induced biological effects.

cell survival↗

Statistical analysis of mesoscale rainfall: Dependence of a random cascade generator on large-scale forcing

Under the theory of independent and identically distributed random cascades, the probability distribution of the cascade generator determines the spatial and the ensemble properties of spatial rainfall. Three sets of radar-derived rainfall data in space and time are analyzed to estimate the probability distribution of the generator. A detailed comparison between instantaneous scans of spatial rainfall and simulated cascades using the scaling properties of the marginal moments is carried out. This comparison highlights important similarities and differences between the data and the random cascade theory. Differences are quantified and measured for the three datasets. Evidence is presented to show that the scaling properties of the rainfall can be captured to the first order by a random cascade with a single parameter. The dependence of this parameter on forcing by the large-scale meteorological conditions, as measured by the large-scale spatial average rain rate, is investigated for these three datasets. The data show that this dependence can be captured by a one-to-one function. Since the large-scale average rain rate can be diagnosed from the large-scale dynamics, this relationship demonstrates an important linkage between the large-scale atmospheric dynamics and the statistical cascade theory of mesoscale rainfall. Potential application of this research to parameterization of runoff from the land surface and regional flood frequency analysis is briefly discussed, and open problems for further research are presented.

Over, Thomas, M.↗

On the Contribution of Turbulent Boundary Layers to the Noise Inside a Fuselage

The following report deals in preliminary fashion with the transmission through a fuselage of random noise generated on the fuselage skin by a turbulent boundary layer. The concept of attenuation is abandoned and instead the problem is formulated as a sequence of two linear couplings: the turbulent boundary layer fluctuations excite the fuselage skin in lateral vibrations and the skin vibrations induce sound inside the fuselage. The techniques used are those required to determine the response of linear systems to random forcing functions of several variables. A certain degree of idealization has been resorted to. Thus the boundary layer is assumed locally homogeneous, the fuselage skin is assumed flat, unlined and free from axial loads and the 'cabin' air is bounded only by the vibrating plate so that only outgoing waves are considered. Some of the details of the statistical description have been simplified in order to reveal the basic features of the problem. The results, strictly applicable only to the limiting case of thin boundary layers, show that the sound pressure intensity is proportional to the square of the free stream density, the square of cabin air density and inversely proportional to the first power of the damping constant and to the second power of the plate density. The dependence on free stream velocity and boundary layer thickness cannot be given in general without a detailed knowledge of the characteristics of the pressure fluctuations in the boundary layer (in particular the frequency spectrum). For a flat spectrum the noise intensity depends on the fifth power of the velocity and the first power of the boundary layer thickness. This suggests that boundary layer removal is probably not an economical means for decreasing cabin noise. In general, the analysis presented here only reduces the determination of cabin noise intensity to the measurement of the effect of any one of our variables (free stream velocity, boundary layer thickness, plate thickness or the characteristic velocity of propagation in the plate). The plate generates noise by vibrating in resonance over a wide range of frequencies and increasing the damping constant is consequently an effective method of decreasing noise generation. One of the main features of the results is that the relevant quantities upon which noise intensity depends are non-dimensional numbers in which boundary layer and plate properties enter as ratios. This is taken as an indication that in testing models of structures for boundary layer noise it is not sufficient to duplicate in the model the structural characteristics of the fuselage. One must match properly the characteristics of the exicitng pressure fluctuations to that of the structure.

Corcos, G. M.↗

Further efforts to limit lunar internal temperatures from electrical conductivity determinations

A representative set of 18 electrical conductivity profiles consistent with lunar transfer function data in the 0.00001 to 0.001 Hz range is generated by means of a Monte Carlo procedure. Laboratory electrical conductivity vs temperature data extrapolated to lower and higher temperatures are used to convert the 18 profiles into temperature profiles for a series of radially homogeneous olivine-pyroxene mixtures ranging from 100% olivine to 100% aluminous orthopyroxene. The former yields selenotherms approaching the Ringwood-Essene solidus at depths near 500 km, while additions of the latter in concentrations greater than 15-30 vol% leads to cooler selenotherms which approach the solidus only at depths greater than about 1000 km. On the basis of independent geophysical constraints, the profile envelopes for compositions containing more than 15-30 vol% aluminous orthopyroxene are considered to be the most probable.

Hood, L. L.↗

Evaluation of Refractivity Profiles From Champ and SAC-C GPS Radio Occultation

The CHAMP and SAC-C missions are the first missions to carry a second-generation 'Blackjack' GPS receiver. One of the new features of this receiver is its ability to sense the lower troposphere closer to the surface than the proof-of-concept GPS/MET 1995 experiment. Since their launch, CHAMP and SAC-C have collected thousands of GPS radio occultations, representing a wealth of measurements available for data assimilation in Numerical Weather Prediction (NWP) models. In order to evaluate the refractivity data derived by JPL from raw radio occultation measurements, we use Data Assimilation Office (DAO) shout forecasts as an independent state of the atmosphere. We compare CHAMP and SAC-C refractivity (processed by JPL) with refractivity calculated from the DAO global fields of temperature, water vapor content and humidity. We will show statistics of the differences as well as Probability Density Functions (PDFs) of the differences. Depending upon availability of AIRS data, we plan to show individual profile comparisons between GPS radio occultation and AIRS retrievals.

Poli, Paul↗

Evaluation of Refractivity Profiles from CHAMP and SAC-C GPS Radio Occultation

The CHAMP and SAC-C missions are the first missions to carry a second-generation 'Blackjack' GPS receiver. One of the new features of this receiver is its ability to sense the lower troposphere closer to the surface than the proof-of-concept GPS/MET 1995 experiment. Since their launch, CHAMP and SAC-C have collected thousands of GPS radio occultations, representing a wealth of measurements available for data assimilation in Numerical Weather Prediction (NWP) models. In order to evaluate the refractivity data derived by JPL from raw radio occultation measurements, we use Data Assimilation Office (DAO) 6-hour forecasts as an independent state of the atmosphere. We compare CHAMP and SAC-C refractivity (processed by JPL) with refractivity calculated from the DAO global fields of temperature, water vapor content and humidity. We will show statistics of the differences as well as Probability Density Functions (PDFs) of the differences. Depending upon availability of AIRS data, we plan to show individual profile comparisons between GPS radio occultation and AIRS retrievals.

Poli, Paul↗

Ballistic Performance Model of Crater Formation in Monolithic, Porous Thermal Protection Systems

Porous monolithic ablative systems insulate atmospheric reentry vehicles from reentry plasmas generated by atmospheric braking from orbital and exo-orbital velocities. Due to the necessity that these materials create a temperature gradient up to several thousand Kelvin over their thickness, it is important that these materials are near their pristine state prior to reentry. These materials may also be on exposed surfaces to space environment threats like orbital debris and meteoroids leaving a probability that these exposed surfaces will be below their prescribed values. Owing to the typical small size of impact craters in these materials, the local flow fields over these craters and the ablative process afford some margin in thermal protection designs for these locally reduced performance values. In this work, tests to develop ballistic performance models for thermal protection materials typical of those being used on Orion are discussed. A density profile as a function of depth of a typical monolithic ablator and substructure system is shown in Figure 1a.

Miller, J. E.↗

Pristine Igneous Rocks and the Early Differentiation of Planetary Materials

Our studies are highly interdisciplinary, but are focused on the processes and products of early planetary and asteroidal differentiation, especially the genesis of the ancient lunar crust. Most of the accessible lunar crust consists of materials hybridized by impact-mixing. Rare pristine (unmixed) samples reflect the original genetic diversity of the early crust. We studied the relative importance of internally generated melt (including the putative magma ocean) versus large impact melts in early lunar magmatism, through both sample analysis and physical modeling. Other topics under investigation included: lunar and SNC (martian?) meteorites; igneous meteorites in general; impact breccias, especially metal-rich Apollo samples and polymict eucrites; effects of regolith/megaregolith insulation on thermal evolution and geochronology; and planetary bulk compositions and origins. We investigated the theoretical petrology of impact melts, especially those formed in large masses, such as the unejected parts of the melts of the largest lunar and terrestrial impact basins. We developed constraints on several key effects that variations in melting/displacement ratio (a strong function of both crater size and planetary g) have on impact melt petrology. Modeling results indicate that the impact melt-derived rock in the sampled, megaregolith part of the Moon is probably material that was ejected from deeper average levels than the non-impact-melted material (fragmental breccias and unbrecciated pristine rocks). In the largest lunar impacts, most of the impact melt is of mantle origin and avoids ejection from the crater, while most of the crust, and virtually all of the impact-melted crust, in the area of the crater is ejected. We investigated numerous extraordinary meteorites and Apollo rocks, emphasizing pristine rocks, siderophile and volatile trace elements, and the identification of primary partial melts, as opposed to partial cumulates. Apollo 15 sample 15434,28 is an extraodinarily large glass spherule, nearly if not entirely free of meteoritic contamination, and provides insight into the diversity of mare basalts in the Hadley-Apennine region. Apollo 14 sample 14434 is in many respects a new rock type, intermediate between nonmare gabbronorites and mare basalts. We helped to both plan and implement a consortium to study the Yamato-793605 SNC/martian meteorite.

Warren, Paul H.↗

Extending Newtonian Dynamics to Include Stochastic Processes

A paper presents further results of continuing research reported in several previous NASA Tech Briefs articles, the two most recent being Stochastic Representations of Chaos Using Terminal Attractors (NPO-41519), [Vol. 30, No. 5 (May 2006), page 57] and Physical Principle for Generation of Randomness (NPO-43822) [Vol. 33, No. 5 (May 2009), page 56]. This research focuses upon a mathematical formalism for describing post-instability motions of a dynamical system characterized by exponential divergences of trajectories leading to chaos (including turbulence as a form of chaos). The formalism involves fictitious control forces that couple the equations of motion of the system with a Liouville equation that describes the evolution of the probability density of errors in initial conditions. These stabilizing forces create a powerful terminal attractor in probability space that corresponds to occurrence of a target trajectory with probability one. The effect in configuration space (ordinary three-dimensional space as commonly perceived) is to suppress exponential divergences of neighboring trajectories without affecting the target trajectory. As a result, the post-instability motion is represented by a set of functions describing the evolution of such statistical quantities as expectations and higher moments, and this representation is stable.

Zak, Michail↗

Bayesian Estimation of Earth’s Undiscovered Mineralogical Diversity Using Noninformative Priors

Recently, statistical distributions have been explored to provide estimates of the mineralogical diversity of Earth, and Earth-like planets. In this paper, a Bayesian approach is introduced to estimate Earth’s undiscovered mineralogical diversity. Samples are generated from a posterior distribution of the model parameters using Markov chain Monte Carlo simulations such that estimates and inference are directly obtained. It was previously shown that the mineral species frequency distribution conforms to a generalized inverse Gauss–Poisson (GIGP) large number of rare events model. Even though the model fit was good, the population size estimate obtained by using this model was found to be unreasonably low by mineralogists. In this paper, several zero-truncated, mixed Poisson distributions are fitted and compared, where the Poisson-lognormal distribution is found to provide the best fit. Subsequently, the population size estimates obtained by Bayesian methods are compared to the empirical Bayes estimates. Species accumulation curves are constructed and employed to estimate the population size as a function of sampling size. Finally, the relative abundances, and hence the occurrence probabilities of species in a random sample, are calculated numerically for all mineral species in Earth’s crust using the Poisson-lognormal distribution. These calculations are connected and compared to the calculations obtained in a previous paper using the GIGP model for which mineralogical criteria of an Earth-like planet were given.

Bayesian statistics↗

Higher-Order Statistical Moments of Predicted Sonic Boom Waveforms Through Turbulence

Sonic booms generated by supersonic aircraft are affected by turbulence in the atmospheric boundary layer through which they propagate. Turbulence effects lead to random variability of the sonic boom waveforms measured on the ground, complicating the prediction of such waveforms. As an initial effort to predict the waveform variability, the solution of the coherent or mean sonic boom waveform has previously been formulated by the author. The current paper extends the formulation to derive an expression of the second-order statistical moment necessary to calculate the variance of the spectral amplitudes. Since the derivation uses a full wave equation, results using the derived expression will be compared with those obtained using a parabolic approximation. In order to fully quantify the uncertainty of our predictions, the higher order statistical moments are also formulated and are shown to be approximately zero. Consequently, the probability density function (pdf) of the spectral amplitudes is predicted to be Gaussian. The formulation is further extended to determine the pdf of the loudness of sonic booms and quantify the uncertainties associated with the loudness prediction. Results from the formulation are compared with available flight test data.

sonic booms↗

Voyager Observations of the Color of Saturn's Ring

Previously unreduced high resolution Voyager 2 images of Saturn's main rings are used to generate reflectivity (I/F) profiles as a function of radius. Ratios of profiles taken from green, violet, orange, and UV filter images are then produced. The I/F ratios are diagnostic of composition, and provide us with information on the rings' present state of compositional evolution. The rings are extremely reddish in color which suggests that they could not be pure water ice. The most likely candidates for the non-icy components are silicates and organics. The sources of these pollutants are of extreme importance in determining the compositional history of the rings. The radial profiles of ring color ratio exhibit several very interesting properties: (a) broad-scale, fairly smooth, color variations which are only weakly correlated with underlying ring structure (optical depth variations) across the outer C ring and inner B ring as well as the Cassini division region. These variations are probably consistent with ballistic transport; (b) fine-scale, noise-Like (but unquestionably real) color variations across at least the outer two-thirds of the B ring. Not only the "redness" but the spectral shape varies. These variations are currently unexplained. Groundbased spectroscopic observations should be pursued to study the implied compositional heterogeneities on at least the larger scales. This data set will be used for modeling of the color and composition of the main rings using ballistic transport and radiative transfer theories.

Estrada, Paul R.↗

A reexamination of two-fluid solar wind models

The two-fluid solar-wind equations have been solved by a method which is approximately 50 times faster than any previously developed, through the use of asymptotic expansions which are self-consistently iterated upon to find a solution that passes through the critical point. The energy assumptions in two-fluid solar-wind models are reexamined, and the conclusions are as follows: (1) proton thermal conduction may not be neglected, (2) the Coulomb logarithm must be calculated as a function of radius, and (3) the electron and proton temperatures at the base need not be equal, even when the time scale for energy exchange between the species is an order of magnitude smaller than the expansion time at the base. It is possible to reproduce reasonable quiet-time solar-wind parameters at 1 AU, but only if the proton temperature is approximately twice the electron temperature at 1 solar radius. This may indicate that extended proton heating is important in the outer solar corona. Winds with velocities at 1 AU of 450 km/s are generated without nonthermal energy deposition but require high proton temperatures as well as very low densities at the base. Higher-velocity solutions are not possible in a spherically symmetric geometry for reasonable particle fluxes at 1 AU, and it is suggested that these higher-velocity states probably require additional heating, acceleration mechanisms, or nonradial flow.

Nerney, S.↗

A flight-test and simulation evaluation of the longitudinal final approach and landing performance of an automatic system for a light wing loading STOL aircraft equipped with wing spoilers

As part of a comprehensive flight-test investigation of short takeoff and landing (STOL) operating systems for the terminal systems for the terminal area, an automatic landing system has been developed and evaluated for a light wing-loading turboprop-powered aircraft. An advanced digital avionics system performed display, navigation, guidance, and control functions for the test aircraft. Control signals were generated in order to command powered actuators for all conventional controls and for a set of symmetrically driven wing spoilers. This report describes effects of the spoiler control on longitudinal autoland (automatic landing) performance. Flight-test results, with and without spoiler control, are presented and compared with available (basically, conventional takeoff and landing) performance criteria. These comparisons are augmented by results from a comprehensive simulation of the controlled aircraft that included representations of the microwave landing system navigation errors that were encountered in flight as well as expected variations in atmospheric turbulence and wind shear. Flight-test results show that the addition of spoiler control improves the touchdown performance of the automatic landing system. Spoilers improve longitudinal touchdown and landing pitch-attitude performance, particularly in tailwind conditions. Furthermore, simulation results indicate that performance would probably be satisfactory for a wider range of atmospheric disturbances than those encountered in flight. Flight results also indicate that the addition of spoiler control during the final approach does not result in any measurable change in glidepath track performance, and results in a very small deterioration in airspeed tracking. This difference contrasts with simulations results, which indicate some improvement in glidepath tracking and no appreciable change in airspeed tracking. The modeling problem in the simulation that contributed to this discrepancy with flight was not resolved.

Brown, S. C.↗

Adding GPU Support to the Markov Chain Monte Carlo Code Catmip

In geophysics, we are confronted with many under-determined inverse problems. For example, all of our observations of earthquakes are made at the Earth’s surface. So, when we try to infer how slip during an earthquake evolves in space and time, we find that there are many potential slip histories that are consistent with our limited observations and our understanding of earthquake physics. One way to approach these problems is with Bayesian analysis which allows us to infer the ensemble of all potential slip models that satisfy the observations and our prior knowledge of earthquake physics. In Bayesian analysis, our prior knowledge is known as the prior probability density function or prior PDF, the fit to the data is known as the data likelihood, and the target PDF that satisfies both the prior PDF and data likelihood is known as the posterior PDF. However, simulating the posterior PDF typically requires using Markov Chain Monte Carlo (MCMC) to draw tens of billions of random realizations of earthquake slip models, which may not be computationally feasible. To make this and similar geophysical inversions computationally tractable, we developed the Cascading Adaptive Transitional Metropolis In Parallel (CATMIP) algorithm. CATMIP is an efficient parallel Markov Chain Monte Carlo (MCMC) sampler that is used for model fitting and uncertainty quantification in geophysics. Example use cases are earthquake rupture modeling, determining mineral composition on Mars, reconstructing the history of ocean salinity, and historical earthquake relocation. CATMIP employs many parallel instances of the Metropolis algorithm for sampling in a transitioning framework. Transitioning is a process in which a set of random samples at equilibrium with a known probability density function (PDF) are used as seeds for the Markov chains to sample successive target PDFs that incrementally move the distribution from the starting seeds to the final desired PDF that describes the relative plausibility of potential values for the model parameters. The algorithm is implemented as a Master-Worker model employing MPI for communication. The worker processes are loosely coupled with global parameters periodically optimized by the master process. This provides a very high amount of parallelism with little communication between updates. During the presentation we will discuss the history of the algorithm and elaborate the earthquake rupture modeling use case for the CATMIP package. Our first step toward GPU optimization was to optimize the code for the CPU. CPU profiling revealed that most of the compute time is spent in calls to level 2 BLAS routines and calls to GSL random number generators. We revised the algorithm to employ level 3 BLAS routines instead. In our presentation we will describe how this was accomplished. Adding GPU support to CATMIP consisted mostly of replacing the calls to GSL with calls to GPU vendor-provided library routines. A small number of loops were directly implemented in CUDA. In the presentation will provide implementation details. Finally, we will discuss methods for profiling and opportunities for further optimizing GPU execution. By creating a code with the flexibility to run on either a CPU or GPU architecture, CATMIP can be used on systems ranging from large CPU-based HPC environments to single servers with GPU acceleration and everything in between.

HECC↗

Heat transfer measurements for Stirling machine cylinders

The primary purpose of this study was to measure the effects of inflow-produced heat turbulence on heat transfer in Stirling machine cylinders. A secondary purpose was to provide new experimental information on heat transfer in gas springs without inflow. The apparatus for the experiment consisted of a varying-volume piston-cylinder space connected to a fixed volume space by an orifice. The orifice size could be varied to adjust the level of inflow-produced turbulence, or the orifice plate could be removed completely so as to merge the two spaces into a single gas spring space. Speed, cycle mean pressure, overall volume ratio, and varying volume space clearance ratio could also be adjusted. Volume, pressure in both spaces, and local heat flux at two locations were measured. The pressure and volume measurements were used to calculate area averaged heat flux, heat transfer hysteresis loss, and other heat transfer-related effects. Experiments in the one space arrangement extended the range of previous gas spring tests to lower volume ratio and higher nondimensional speed. The tests corroborated previous results and showed that analytic models for heat transfer and loss based on volume ratio approaching 1 were valid for volume ratios ranging from 1 to 2, a range covering most gas springs in Stirling machines. Data from experiments in the two space arrangement were first analyzed based on lumping the two spaces together and examining total loss and averaged heat transfer as a function of overall nondimensional parameter. Heat transfer and loss were found to be significantly increased by inflow-produced turbulence. These increases could be modeled by appropriate adjustment of empirical coefficients in an existing semi-analytic model. An attempt was made to use an inverse, parameter optimization procedure to find the heat transfer in each of the two spaces. This procedure was successful in retrieving this information from simulated pressure-volume data with artificially generated noise, but it failed with the actual experimental data. This is evidence that the models used in the parameter optimization procedure (and to generate the simulated data) were not correct. Data from the surface heat flux sensors indicated that the primary shortcoming of these models was that they assumed turbulence levels to be constant over the cycle. Sensor data in the varying volume space showed a large increase in heat flux, probably due to turbulence, during the expansion stroke.

Kornhauser, Alan A.↗

Realistic Covariance Generation for the GPM Spacecraft

A covariance realism process for NASA's Global Precipitation Measurement (GPM) spacecraft is detailed. The GPM spacecraft is in a low earth orbit, and performs collision avoidance maneuvers few times a year. Currently GPM is below the International Space Station (ISS). So, in addition to cataloged debris objects, GPM must contend with smallsat/cubesat objects that are deployed from the ISS. Both operational scenarios require complete knowledge of the expected GPM prediction errors as a function of time. In this study, we present a method for generating realistic predicted covariance that uses linear propagation of the covariance with the addition of process noise. Further analyses are presented for the process noise ''tuning'' that generates an inflation factor based on the observed error statistics of the predictive satellite trajectories when compared to the definitive ones. Different tuning strategies are considered and compared via a Goodness-of-Fit testing for the Gaussian properties of the scaled covariance. SpaceNav's realistic covariance generation approach takes into account the contribution of predicted maneuver errors in the increased propagation uncertainty. Corresponding maneuver uncertainty is injected into the state uncertainty, and is used within the collision avoidance process to determine the collision risk for close approach events that follow a maneuver. This is a critical step in the maneuver planning process that provides the satellite operator with an accurate quantification of the collision probability for planned maneuvers. Using this information, an informed decision can be made to proceed with a maneuver if the collision risk is acceptable. This approach is validated by Monte-Carlo simulations and results are presented.

spacecraft uncertainty propagation↗

A Communication Channel Density Estimating Generative Adversarial Network

Autoencoder-based communication systems use neural network channel models to backwardly propagate message reconstruction error gradients across an approximation of the physical communication channel. In this work, we develop and test a new generative adversarial network (GAN) architecture for the purpose of training a stochastic channel approximating neural network. In previous research, investigators have focused on additive white Gaussian noise (AWGN) channels and/or simplified Rayleigh fading channels, both of which are linear and have well defined analytic solutions. Given that training a neural network is computationally expensive, channel approximation networks— and more generally the autoencoder systems—should be evaluated in communication environments that are traditionally difficult. To that end, our investigation focuses on channels that contain a combination of non-linear amplifier distortion, pulse shape filtering, intersymbol interference, frequency-dependent group delay, multipath, and non-Gaussian statistics. Each of our models are trained without any prior knowledge of the channel. We show that the trained models have learned to generalize over an arbitrary amplifier drive level and constellation alphabet. We demonstrate the versatility of our GAN architecture by comparing the marginal probability density function of several channel simulations with that of their corresponding neural network approximations

Smith, Aaron↗