Refined Large Deviation Asymptotics for the Classical Occupancy Problem

In this paper refined large deviation asymptotics are derived for the classical occupancy problem. The asymptotics are established for a sequential filling experiment and an occupancy experiment. In the first case the random variable of interest is the number of balls required to fill a given fraction of the urns, while in the second a fixed number of balls are thrown and the random variable is the fraction of nonempty urns.      

Estimation of the Location of the Maximum of a Regression Function Using Extreme Order Statistics

In this paper, we consider the problem of approximating the location,x₀∈C, of a maximum of a regresion function,θ(x), under certain weak assumptions onθ. HereCis a bounded interval inR. A specific algorithm considered in this paper is as follows. Taking a random sampleX₁, …, X_nfrom a distribution overC, we have (X_i, Y_i), whereY_iis the outcome of noisy measurement ofθ(X_i). Arrange theY_i's in nondecreasing order and take the average of ther X_i's which are associated with therlargest order statistics ofY_i. This average,x₀, will then be used as an estimate ofx₀. The utility of such an algorithm with fixed r is evaluated in this paper. To be specific, the convergence rates ofx₀tox₀are derived. Those rates will depend on the right tail of the noise distribution and the shape ofθ(·) nearx₀.     

On the convergence rate of maximal deviation distribution for kernel regression estimates

Let (X, Y), X R^p, Y R¹ have the regression function r(x) = E(Y¦X = x). We consider the kernel nonparametric estimate r_n(x) of r(x) and obtain a sequence of distribution functions approximating the distribution of the maximal deviation with power rate. It is shown that the distribution of the maximal deviation tends to double exponent (which is a conventional form of such theorems) with logarithmic rate and this rate cannot be improved.     

Sequential design in an Urn model

Finitely many urns are filled with balls each having one of the labels 1,.Nin known proportions. Balls can be drawn with replacement from any urn at the cost of c>0 monetary units per drawing. Let X ₄be the number of labels that show up in the first kdrawings. We solve the problem to find a selection rule for the urns to be used and a stopping rule such that E(X _τ–eτ) is maximized. A second problem of this kind is also treated.     

L1-optimal estimates for a regression type function in R

Let X₁, X₂, …, X_n be random vectors that take values in a compact set in R^d, d ≥ 1. Let Y₁, Y₂, …, Y_n be random variables (“the responses”) which conditionally on X₁ = x₁, …, X_n = x_n are independent with densities f(y | x_i, θ(x_i)), i = 1, …, n. Assuming that θ lives in a sup-norm compact space Θ_q,d of real valued functions, an optimal L₁-consistent estimator of θ is constructed via empirical measures. The rate of convergence of the estimator to the true parameter θ depends on Kolmogorov's entropy of Θ_q,d.     

Stochastic Comparisons and Dependence among Concomitants of Order Statistics

Let (X_i, Y_i) i=1, 2, …, n be n independent and identically distributed random variables from some continuous bivariate distribution. If X_(r) denotes the rth ordered X-variate then the Y-variate, Y_[r], paired with X_(r) is called the concomitant of the rth order statistic. In this paper we obtain new general results on stochastic comparisons and dependence among concomitants of order statistics under different types of dependence between the parent random variables X and Y. The results obtained apply to any distribution with monotone dependence between X and Y. In particular, when X and Y are likelihood ratio dependent, it is shown that the successive concomitants of order statistics are increasing according to likelihood ratio ordering and they are TP₂ dependent in pairs. If we assume that the conditional hazard rate of Y given X=x is decreasing in x, then the concomitants are increasing according to hazard rate ordering and are dependent according to the right corner set increasing property. Finally, it is proved that if Y is stochastically increasing in X, then the concomitants of order statistics are stochastically increasing and are associated. Analogous results are obtained when the variables X and Y are negatively dependent. We also prove that if the hazard rate of the conditional distribution of Y given X=x is decreasing in x and y, then the concomitants have DFR (decreasing failure rate) distributions and are ordered according to dispersive ordering.     

Two-point concentration in random geometric graphs

A random geometric graph G _n is constructed by taking vertices X ₁,…,X _n ∈ℝ ^d at random (i.i.d. according to some probability distribution ν with a bounded density function) and including an edge between X _i and X _j if ‖X _i -X _j ‖ < r where r = r(n) > 0. We prove a conjecture of Penrose ([14]) stating that when r=r(n) is chosen such that nr ^d = o(lnn) then the probability distribution of the clique number ω(G _n ) becomes concentrated on two consecutive integers and we show that the same holds for a number of other graph parameters including the chromatic number χ(G _n ). The author was partially supported by EPSRC, the Department of Statistics, Bekkerla-Bastide fonds, Dr. Hendrik Muller’s Vaderlandsch fonds, and Prins Bernhard Cultuurfonds.     

Central limit theorem for integrated square error of kernel estimators of spherical density

LetX ₁,…,X _n be iid observations of a random variableX with probability density functionf(x) on the q-dimensional unit sphere Ω_q in R^q+1,q ⩾ 1. Let be a kernel estimator off(x). In this paper we establish a central limit theorem for integrated square error off _n under some mild conditions.     

Exponential bounds of mean error for the kernel estimates of regression functions

Let (X, Y), (X₁, Y₁), …, (X_n, Y_n) be i.d.d. R^r × R-valued random vectors with E|Y| < ∞, and let Q_n(x) be a kernel estimate of the regression function Q(x) = E(Y|X = x). In this paper, we establish an exponential bound of the mean deviation between Q_n(x) and Q(x) given the training sample Zⁿ = (X₁, Y₁, …, X_n, Y_n), under conditions as weak as possible.     

Almost sure central limit theorem for partial sums and maxima

Let X, X₁, X₂, … be i.i.d. random variables with nondegenerate common distribution function F, satisfying EX = 0, EX² = 1. Let X_i and M_n = max{X_i, 1 ≤ i ≤ n }. Suppose there exists constants a_n > 0, b_n ∈ R and a nondegenrate distribution G (y) such that Then, we have almost surely, where f (x, y) denotes the bounded Lipschitz 1 function and Φ(x) is the standard normal distribution function (© 2009 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

A certain estimate for the rate of convergence in the central limit theorem with respect to balls in the finite-dimensional space

+ Let X₁, X₂, ... be independent, identically distributed random variables (r.v.) with values in the space R^k. One assumes that these r.v. have zero mean and covariance operator equal to the identity. We denote by P the distribution of the r.v. X₁, by P_n the distribution of the r.v. (X₁+ ...+X_n)n^–1/2, and by the standard normal law. One investigates the problem of the estimation of the quantity where S_r(a)=S_r = are balls in R^k.Translated from Veroyatnostnye Raspredeleniya i Matematicheskaya Statistika, pp. 421–435, 1986.In conclusion, I use this opportunity to express my gratitude to V. M. Zolotarev for his constant interest in this paper.     

Potential Theoretic Approach to Rendezvous Numbers

We analyze relations between various forms of energies (reciprocal capacities), the transfinite diameter, various Chebyshev constants and the so-called rendezvous or average number. The latter is originally defined for compact connected metric spaces (X,d) as the (in this case unique) nonnegative real number r with the property that for arbitrary finite point systems {x ₁, …, x _n } ⊂ X, there exists some point x ∈ X with the average of the distances d(x,x _j ) being exactly r. Existence of such a miraculous number has fascinated many people; its normalized version was even named “the magic number” of the metric space. Exploring related notions of general potential theory, as set up, e.g., in the fundamental works of Fuglede and Ohtsuka, we present an alternative, potential theoretic approach to rendezvous numbers.     

Potential Theoretic Approach to Rendezvous Numbers

We analyze relations between various forms of energies (reciprocal capacities), the transfinite diameter, various Chebyshev constants and the so-called rendezvous or average number. The latter is originally defined for compact connected metric spaces (X,d) as the (in this case unique) nonnegative real number r with the property that for arbitrary finite point systems {x ₁, …, x _n } ⊂ X, there exists some point x ∈ X with the average of the distances d(x,x _j ) being exactly r. Existence of such a miraculous number has fascinated many people; its normalized version was even named “the magic number” of the metric space. Exploring related notions of general potential theory, as set up, e.g., in the fundamental works of Fuglede and Ohtsuka, we present an alternative, potential theoretic approach to rendezvous numbers.     

Positive linear operators generated by analytic functions

Let φ be a power series with positive Taylor coefficients {a _k } _k=0^∞ and non-zero radius of convergence r ≤ ∞. Let ξ _x , 0 ≤ x < r be a random variable whose values α _k , k = 0, 1, …, are independent of x and taken with probabilities a _k x ^k /φ(x), k = 0, 1, …. The positive linear operator (A _φ f)(x):= E[f(ξ _x )] is studied. It is proved that if E(ξ _x ) = x, E(ξ _x ²) = qx ² + bx + c, q, b, c ∈ R, q > 0, then A _φ reduces to the Szász-Mirakyan operator in the case q = 1, to the limit q-Bernstein operator in the case 0 < q < 1, and to a modification of the Lupaş operator in the case q > 1.     

Measure concentration for a class of random processes

Summary.   Let X={X _i } _{i =−∞} ^∞ be a stationary random process with a countable alphabet and distribution q. Let q ^∞(·|x _{− k} ⁰) denote the conditional distribution of X ^∞=(X ₁,X ₂,…,X _n ,…) given the k-length past:

Compound Poisson Approximation for Multiple Runs in a Markov Chain

We consider a sequence X ₁, ..., X _n of r.v.'s generated by a stationary Markov chain with state space A = {0, 1, ..., r}, r 1. We study the overlapping appearances of runs of k _i consecutive i's, for all i = 1, ..., r, in the sequence X ₁,..., X _n. We prove that the number of overlapping appearances of the above multiple runs can be approximated by a Compound Poisson r.v. with compounding distribution a mixture of geometric distributions. As an application of the previous result, we introduce a specific Multiple-failure mode reliability system with Markov dependent components, and provide lower and upper bounds for the reliability of the system.     

The generic division rings

LetA=k (X ₁, X₂..., X_m) be the division ring generated by genericn×n matrices over a fieldk; thenA is not a crossed product in the following cases: (i) there exists a primeq such thatq ³ℛn;(ii)[k:Q]=m, whereQ is the field of rationals, then if eitherq ³ℛn for someq for whichq-1ℛm, orq ²/nn for some other prime; (iii)k=Z _p ^r a finite field ofp ^r elements and eitherq ³ℛn for sameqℛp ^r-1 orq ²ℛn for some other primes. Other cases are also considered.     

On the ratio of the expected maximum of a martingale and theL p-norm of its last term

For eachp>1, the supremum,S, of the absolute value of a martingale terminating at a random variableX inL _p, satisfiesES≦(Γ(q))^1/q‖X‖_p (q=p(p-1)^-1).The maximum,M, of a mean-zero martingale which starts at zero and terminates atX, satisfiesES≦(Γ(q))^1/q‖X‖_p (q=p(p-1)^-1), whereσ _q is the unique solution of the equationt = ‖Z −t ‖ _q for an exponentially distributed random variableZ with mean 1.σ _p has other characterizations and satisfies lim _p‖ q ^{− 1} σ _q =c withc determined byce ^c+1 = 1. Equalities in (1) and (2) are attainable by appropriate martingales which can be realized as stopped segments of Brownian motion. A presumably new property of the exponential distribution is obtained en route to inequality (2).     

The Asymptotic Distribution of Singular-Values with Applications to Canonical Correlations and Correspondence Analysis

Let X_n, n = 1, 2, ... be a sequence of p × q random matrices, p ≥ q. Assume that for a fixed p × q matrix B and a sequence of constants b_n → ∞, the random matrix b_n(X_n − B) converges in distribution to Z. Let ψ(X_n) denote the q-vector of singular values of X_n. Under these assumptions, the limiting distribution of b_n (ψ(X_n) − ψ(B)) is characterized as a function of B and of the limit matrix Z. Applications to canonical correlations and to correspondence analysis are given.     

Updating of moments

Consider a statistical model, given by the distribution of the observation X, conditional on the parameter θ, and the prior distribution of the parameter θ. Let H_x denote the function that maps the prior mean and the prior covariance matrix into the posterior mean and the posterior covariance matrix, when X = x is observed. We prove that if the conditional distribution of X belongs to an exponential family, then the function H_x characterizes the distribution of Xθ.