Independent Gaussian Random Variables Research Articles

Separating signal from noise

Suppose that a sequence of numbers $x_n$ (a `signal') is transmitted through a noisy channel. The receiver observes a noisy version of the signal with additive random fluctuations, $x_n + \xi_n$, where $\xi_n$ is a sequence of independent standard Gaussian random variables. Suppose further that the signal is known to come from some fixed space of possible signals. Is it possible to fully recover the transmitted signal from its noisy version? Is it possible to at least detect that a non-zero signal was transmitted? In this paper we consider the case in which signals are infinite sequences and the recovery or detection are required to hold with probability one. We provide conditions on the signal space for checking whether detection or recovery are possible. We also analyze in detail several examples including spaces of Fourier transforms of measures, spaces with fixed amplitudes and the space of almost periodic functions. Many of our examples exhibit critical phenomena, in which a sharp transition is made from a regime in which recovery is possible to a regime in which even detection is impossible.

Cramer-Rao bounds in functional form: theory and application to passive optical ranging.

A functional approach to the multivariate statistical model of a generalized incoherent passive optical ranging and imaging system with a CCD sensor is proposed. This approach implies that a large number of discrete, statistically independent, random data (pixel readouts) can be approximated by a continuous random function. Thus, the joint probability density function (PDF) takes a functional form; the statistical averages of the infinite-variate PDF and the Fisher information become functional integrals that can be treated analytically in the Gaussian approximation. The Cramer-Rao bounds on estimator-error variances are obtained for the scalar and functional deterministic parameters of the model. An approximate expression is derived for the PDF of the sum of independent Gaussian and Poisson random variables using the steepest-descent method, and the resulting PDF is shown to be asymptotically Gaussian. As an illustration, we apply the developed approach to a passive optical rangefinder with chiral wavefront coding. Numerical and experimental examples are presented.

Conditional Simulation of Flow in Heterogeneous Porous Media with the Probabilistic Collocation Method

AbstractA stochastic approach to conditional simulation of flow in randomly heterogeneous media is proposed with the combination of the Karhunen-Loeve expansion and the probabilistic collocation method (PCM). The conditional log hydraulic conductivity field is represented with the Karhunen-Loeve expansion, in terms of some deterministic functions and a set of independent Gaussian random variables. The propagation of uncertainty in the flow simulations is carried out through the PCM, which relies on the efficient polynomial chaos expansion used to represent the flow responses such as the hydraulic head. With the PCM, existing flow simulators can be employed for uncertainty quantification of flow in heterogeneous porous media when direct measurements of hydraulic conductivity are taken into consideration. With illustration of several numerical examples of groundwater flow, this study reveals that the proposed approach is able to accurately quantify uncertainty of the flow responses conditioning on hydraulic conductivity data, while the computational efforts are significantly reduced in comparison to the Monte Carlo simulations.

A Rate-Splitting Approach to Fading Channels With Imperfect Channel-State Information

As shown by M\'edard, the capacity of fading channels with imperfect channel-state information (CSI) can be lower-bounded by assuming a Gaussian channel input $X$ with power $P$ and by upper-bounding the conditional entropy $h(X|Y,\hat{H})$ by the entropy of a Gaussian random variable with variance equal to the linear minimum mean-square error in estimating $X$ from $(Y,\hat{H})$. We demonstrate that, using a rate-splitting approach, this lower bound can be sharpened: by expressing the Gaussian input $X$ as the sum of two independent Gaussian variables $X_1$ and $X_2$ and by applying M\'edard's lower bound first to bound the mutual information between $X_1$ and $Y$ while treating $X_2$ as noise, and by applying it a second time to the mutual information between $X_2$ and $Y$ while assuming $X_1$ to be known, we obtain a capacity lower bound that is strictly larger than M\'edard's lower bound. We then generalize this approach to an arbitrary number $L$ of layers, where $X$ is expressed as the sum of $L$ independent Gaussian random variables of respective variances $P_{\ell}$, $\ell = 1,\dotsc,L$ summing up to $P$. Among all such rate-splitting bounds, we determine the supremum over power allocations $P_\ell$ and total number of layers $L$. This supremum is achieved for $L\to\infty$ and gives rise to an analytically expressible capacity lower bound. For Gaussian fading, this novel bound is shown to converge to the Gaussian-input mutual information as the signal-to-noise ratio (SNR) grows, provided that the variance of the channel estimation error $H-\hat{H}$ tends to zero as the SNR tends to infinity.

Weighted quantile correlation test for the logistic family

We summarize the results of investigating the asymptotic behavior of the weighted quantile correlation tests for the location-scale family associated to the logistic distribution. Explicit representations of the limiting distribution are given in terms of integrals of weighted Brownian bridges or alternatively as infinite series of independent Gaussian random variables. The power of this test and the test for the location logistic family against some alternatives are demonstrated by numerical simulations.

History matching of statistically anisotropic fields using the Karhunen-Loeve expansion-based global parameterization technique

Traditional ensemble-based history matching method, such as the ensemble Kalman filter and iterative ensemble filters, usually update reservoir parameter fields using numerical grid-based parameterization. Although a parameter constraint term in the objective function for deriving these methods exists, it is difficult to preserve the geological continuity of the parameter field in the updating process of these methods; this is especially the case in the estimation of statistically anisotropic fields (such as a statistically anisotropic Gaussian field and facies field with elongated facies) with uncertainties about the anisotropy direction. In this work, we propose a Karhunen-Loeve expansion-based global parameterization technique that is combined with the ensemble-based history matching method for inverse modeling of statistically anisotropic fields. By using the Karhunen-Loeve expansion, a Gaussian random field can be parameterized by a group of independent Gaussian random variables. For a facies field, we combine the Karhunen-Loeve expansion and the level set technique to perform the parameterization; that is, for each facies, we use a Gaussian random field and a level set algorithm to parameterize it, and the Gaussian random field is further parameterized by the Karhunen-Loeve expansion. We treat the independent Gaussian random variables in the Karhunen-Loeve expansion as the model parameters. When the anisotropy direction of the statistically anisotropic field is uncertain, we also treat it as a model parameter for updating. After model parameterization, we use the ensemble randomized maximum likelihood filter to perform history matching. Because of the nature of the Karhunen-Loeve expansion, the geostatistical characteristics of the parameter field can be preserved in the updating process. Synthetic cases are set up to test the performance of the proposed method. Numerical results show that the proposed method is suitable for estimating statistically anisotropic fields.

CLT for the zeros of classical random trigonometric polynomials

We prove a Central Limit Theorem for the number of zeros of random trigonometric polynomials of the form $K^{-1/2}\sum_{n=1}^{K} a_n\cos(nt)$, being $(a_n)_n$ independent standard Gaussian random variables. In particular, we prove the conjecture by Farahmand, Granville & Wigman that the variance is equivalent to $V^2K$, $0

Detection performance of coprime sensor arrays

Coprime sensor arrays (CSAs) achieve the resolution of a fully populated uniform linear array (ULA) with the same aperture using fewer sensors. The CSA's reduced number of sensors diminishes its ability to attenuate white noise, and consequently also reduces its array gain. Assuming narrowband independent signal and white noise, when the signal and the noise are modeled by circular complex zero mean Gaussian distributions, the detection statistic distribution for the ULA simplifies to an exponential distribution with its parameter dependent on the input SNR. The detection statistic for the CSA is the product of the output of one subarray with the complex conjugate of the output of the second subarray. Being the product of two independent complex Gaussian random variables, the CSA detection statistic has a distribution proportional to a modified zeroth order Bessel function of the second kind [O'Donoughue and Moura, IEEE Signal Process. (2012)]. Manipulating the ULA and CSA detection statistic distributions provides analytical expressions for probabilities of detection and false alarm and also mean discriminating information. Evaluating these expressions confirms that a CSA pays a detection gain penalty of about 5 dB over a wide range of SNRs and array sizes. [Work supported by ONR.]

Recursive Identification of Wiener-Hammerstein Systems with Nonparametric Nonlinearity

AbstractA recursive scheme is proposed for identifying a single input single output (SISO) Wiener-Hammerstein system, which consists of two linear dynamic subsystems and a sandwiched nonparametric static nonlinearity. The first linear block is assumed to be a finite impulse response (FIR) filter and the second an infinite impulse response (IIR) filter. By letting the input be a sequence of mutually independent Gaussian random variables, the recursive estimates for coefficients of the two linear blocks and the value of the static nonlinear function at any fixed given point are proven to converge to the true values, with probability one as the data size tends to infinity. The static nonlinearity is identified in a nonparametric way and no structural information is directly used. A numerical example is presented that illustrates the theoretical results.

Formal Analysis of Soft Errors using Theorem Proving

Modeling and analysis of soft errors in electronic circuits has traditionally been done using computer simulations. Computer simulations cannot guarantee correctness of analysis because they utilize approximate real number representations and pseudo random numbers in the analysis and thus are not well suited for analyzing safety-critical applications. In this paper, we present a higher-order logic theorem proving based method for modeling and analysis of soft errors in electronic circuits. Our developed infrastructure includes formalized continuous random variable pairs, their Cumulative Distribution Function (CDF) properties and independent standard uniform and Gaussian random variables. We illustrate the usefulness of our approach by modeling and analyzing soft errors in commonly used dynamic random access memory sense amplifier circuits.

On the infinite sums of deflated Gaussian products

In this paper we derive the exact tail asymptotic behaviour of $S_\infty=\sum_{i=1}^\infty \lambda_i X_iY_i$, where $\lambda_i, i\ge 1,$ are non-negative square summable deflators (weights) and $X_i,Y_i, i\ge1,$ are independent standard Gaussian random variables. Further, we consider the tail asymptotics of $S_{\infty;p}=\sum_{i=1}^\infty\lambda_i X_i|Y_i|^p, p> 1$, and also discuss the influence on the asymptotic results when $\lambda_i$'s are independent random variables.

Non asymptotic minimax rates of testing in signal detection with heterogeneous variances

The aim of this paper is to establish non-asymptotic minimax rates for goodness-of-fit hypotheses testing in an heteroscedastic setting. More precisely, we deal with sequences (Yj)j∈J of independent Gaussian random variables, having mean (θj)j∈J and variance (σj)j∈J. The set J will be either finite or countable. In particular, such a model covers the inverse problem setting where few results in test theory have been obtained. The rates of testing are obtained with respect to l2 norm, without assumption on (σj)j∈J and on several functions spaces. Our point of view is entirely non-asymptotic.

Novel Simple Representations for Gaussian Class Multivariate Distributions With Generalized Correlation

Novel single-integral representations for the multivariate probability density functions and cumulative distribution functions of Gaussian class distributions (Rayleigh, Rician, Weibull, Nakagami- <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">m</i> , and generalized Rician) are derived. The solutions are expressed in terms of well-known functions which are available in common mathematical software. The marginal random variables are not necessarily identically distributed as is the case for some past solutions. A generalized correlation structure based on a special linear transformation of independent Gaussian random variables is used in this study. The advantage of the new representation is that only a single-integral computation is needed to compute an <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">N</i> -dimensional distribution.

New Representations for the Gaussian Class Multivariate Weibull Distribution with Constant Correlation and Applications

Novel single-integral representations for the multivariate probability density functions (PDFs) and cumulative distribution functions (CDFs) of the Gaussian class Weibull distribution are derived. The solutions are expressed in terms of familiar mathematical functions which are available in common mathematical software. The well known equal (constant) correlation model is considered. A special linear transformation of independent Gaussian random variables is used to generate correlated Weibull random variables. The advantage of the new representation is that only a single integral computation is needed to compute a N-dimensional distribution. The new representation of the CDF is used for the performance evaluation of a selection diversity combiner operating in equally correlated Weibull fading channels. The new PDF representation is also used for an analysis of the moments of the output signal-to-noise ratio of an equal-gain diversity combiner operating in equally correlated Weibull fading channels.

Capacity Performance of Relay Beamformings for MIMO Multirelay Networks With Imperfect ${\cal R}$-${\cal D}$ CSI at Relays

In this paper, we consider a dual-hop multiple-input-multiple-output (MIMO) wireless relay network in the presence of imperfect channel state information (CSI), in which a source-destination pair, both equipped with multiple antennas, communicates through a large number of half-duplex amplify-and-forward (AF) relay terminals. We investigate the performance of three linear beamforming schemes when the CSI of the relay-to-destination ( ℜ- <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</i> ) link is not perfect at the relay nodes. The three efficient linear beamforming schemes are based on the matched-filter (MF), zero-forcing (ZF) precoding, and regularized ZF (RZF) precoding techniques, which utilize the CSI of both the <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">S</i> -ℜ channel and the ℜ- <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</i> channel at the relay nodes. By modeling the ℜ- <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</i> CSI error at the relay nodes as independent complex Gaussian random variables, we derive the ergodic capacities of the three beamformers in terms of instantaneous signal-to-noise ratio. Using the Law of Large Number, we obtain the asymptotic capacities, upon which the optimized MF-RZF is derived. Simulation results show that the asymptotic capacities match the respective ergodic capacities very well. Analysis and simulation results demonstrate that the optimized MF-RZF outperforms MF and MF-ZF for any power of the ℜ- <i xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">D</i> CSI error.

Stein's Method and the Multivariate CLT for Traces of Powers on the Compact Classical Groups

Let $M$ be a random element of the unitary, special orthogonal, or unitary symplectic groups, distributed according to Haar measure. By a classical result of Diaconis and Shahshahani, for large matrix size $n$, the vector of traces of consecutive powers of $M$ tends to a vector of independent (real or complex) Gaussian random variables. Recently, Jason Fulman has demonstrated that for a single power $j$ (which may grow with $n$), a speed of convergence result may be obtained via Stein's method of exchangeable pairs. In this note, we extend Fulman's result to the multivariate central limit theorem for the full vector of traces of powers.

Extremes of the standardized Gaussian noise

Let { ξ k , k ∈ Z d } be a d -dimensional array of independent standard Gaussian random variables. For a finite set A ⊂ Z d define S ( A ) = ∑ k ∈ A ξ k . Let | A | be the number of elements in A . We prove that the appropriately normalized maximum of S ( A ) / | A | , where A ranges over all discrete cubes or rectangles contained in { 1 , … , n } d , converges in law to the Gumbel extreme-value distribution as n → ∞ . We also prove a continuous-time counterpart of this result.

Exact asymptotic distribution of change-point mle for change in the mean of Gaussian sequences

We derive exact computable expressions for the asymptotic distribution of the change-point mle when a change in the mean occurred at an unknown point of a sequence of time-ordered independent Gaussian random variables. The derivation, which assumes that nuisance parameters such as the amount of change and variance are known, is based on ladder heights of Gaussian random walks hitting the half-line. We then show that the exact distribution easily extends to the distribution of the change-point mle when a change occurs in the mean vector of a multivariate Gaussian process. We perform simulations to examine the accuracy of the derived distribution when nuisance parameters have to be estimated as well as robustness of the derived distribution to deviations from Gaussianity. Through simulations, we also compare it with the well-known conditional distribution of the mle, which may be interpreted as a Bayesian solution to the change-point problem. Finally, we apply the derived methodology to monthly averages of water discharges of the Nacetinsky creek, Germany.

Statistics of the Zeros of Zeta Functions in a Family of Curves over a Finite Field

Let 𝔽q be a finite field of cardinality q and l ≥ 2 be a prime number such that q ≡ 1 (mod l). Extending the work of Faifman and Rudnick [6] on hyperelliptic curves, we study the distribution of zeros of zeta functions of curves over 𝔽q varying over the moduli spaces of cyclic l-fold covers of ℙ1(𝔽q) in the limit of large genus. The zeros all lie on a circle, according to the Riemann Hypothesis for curves, and their angles are uniformly distributed. Moreover, the number of angles inside a fixed symmetric interval I is asymptotically a sum of (l − 1)/2 identical independent Gaussian random variables, each of which comes naturally from a Dirichlet character and has mean 4g|I|/(l − 1) and variance (4/π2)log(2g|I|). These results continue to hold for shrinking intervals as long as the expected number of angles 2g|I|tends to infinity.

Quasi-Static Response of Fixed Offshore Platforms to Morison-Type Wave Loadings

The stochastic response of a rigid platform is quasi-static and can be represented as a linear combination of nonlinear random wave loadings. In this study two efficient probabilistic approaches are respectively presented to estimate the higher-order statistics of mildly and highly non-Gaussian structural responses. The eigenvalue analysis of structural quasi-static response in the first approach is based on the expansion of the response in terms of independent Gaussian random variables. The second approach extends the earlier quasi-static analysis procedures to obtain the higher moments of response by including current effects and inundation effects. The interested prediction of extreme responses during a short-term storm, by mean upcrossing rates and mean extremes, can then be achieved by applying Winterstein's functional transformation. These quantities are found in good agreement with not only Monte Carlo simulation results but also the extreme values obtained using a recently proposed method by Naess et al. in 2007.