Saddlepoint approximation for moments of random variables

In this paper, we introduce a saddlepoint approximation method for higher-order moments like E(S − a)₊ ^m , a>0, where the random variable S in these expectations could be a single random variable as well as the average or sum of some i.i.d random variables, and a > 0 is a constant. Numerical results are given to show the accuracy of this approximation method.     

Set partitions and moments of random variables

It is well known that the sequence of Bell numbers (Bn)_n?0 (Bn being the number of partitions of the set [n]) is the sequence of moments of a mean 1 Poisson random variable τ (a fact expressed in the Dobiński formula), and the shifted sequence (Bn+1)_n?0 is the sequence of moments of 1+τ. In this paper, we generalize these results by showing that both and (where is the number of m-partitions of [n], as they are defined in the paper) are moment sequences of certain random variables. Moreover, such sequences also are sequences of falling factorial moments of related random variables. Similar results are obtained when is replaced by the number of ordered m-partitions of [n]. In all cases, the respective random variables are constructed from sequences of independent standard Poisson processes.     

Inequalities for moments of symmetric polynomial forms in random variables

In the last few years, interest in the properties of distributions of quadratic and polynomial forms in random variables has again grown. Translated fromStatisticheskie Metody Otsenivaniya i Proverki Gipotez, pp. 159–161, Perm, 1993.     

Strong large deviations for arbitrary sequences of random variables

We establish strong large deviation results for an arbitrary sequence of random variables under some assumptions on the normalized cumulant generating function. In other words, we give asymptotic expansions for the tail probabilities of the same kind as those obtained by Bahadur and Rao (Ann. Math. Stat. 31:1015–1027, 1960) for the sample mean. We consider both the case where the random variables are absolutely continuous and the case where they are lattice-valued. Our proofs make use of arguments of Chaganty and Sethuraman (Ann. Probab. 21:1671–1690, 1993) who also obtained strong large deviation results and local limit theorems. We illustrate our results with the kernel density estimator, the sample variance, the Wilcoxon signed-rank statistic and the Kendall tau statistic.     

Minima of sequences of Gaussian random variables

For a given sequence of real numbers $a_1,\dots ,a_n$ we denote the k-th smallest one by $k\text{-}{\min }_{1?i?n}{a}_i$. We show that there exist two absolute positive constants c and C such that for every sequence of positive real numbers $x_1,\dots ,x_n$ and every $k?n$ one has

$c\underset{1?j?k}{\max }\frac{k+1?j}{{\sum }_{i=j}^{n}1/x_i}?\mathbb{E}k\text{-}\underset{1?i?n}{\min }|x_i{g}_i|?C\ln (k+1)\underset{1?j?k}{\max }\frac{k+1?j}{{\sum }_{i=j}^{n}1/x_i},$

where $g_i\in N(0,1)$, $i=1,\dots ,n$, are independent Gaussian random variables. Moreover, if $k=1$ then the left hand side estimate does not require independence of the $g_i$s. Similar estimates hold for $\mathbb{E}k\text{-}{\min }_{1?i?n}{|x_i{g}_i|}^p$ as well. To cite this article: Y. Gordon et al., C. R. Acad. Sci. Paris, Ser. I 340 (2005).     

On sums of independent random variables without power moments

In 1952 Darling proved the limit theorem for the sums of independent identically distributed random variables without power moments under the functional normalization. This paper contains an alternative proof of Darling’s theorem, using the Laplace transform. Moreover, the asymptotic behavior of probabilities of large deviations is studied in the pattern under consideration.     

A limit theorem for the moments of sums of independent random variables

Forn≧1, letS _n=ΣX _n,i (1≦i≦r _n <∞), where the summands ofS _n are independent random variables having medians bounded in absolute value by a finite number which is independent ofn. Letf be a nonnegative function on (− ∞, ∞) which vanishes and is continuous at the origin, and which satisfies, for some for allt≧1 and all values ofx. Theorem.For centering constants c _n,let S _n − c _n converge in distribution to a random variable S. (A)In order that Ef(S_n − c_n) converge to a limit L, it is necessary and sufficient that there exist a common limit (B)If L exists, then L<∞ if and only if R<∞, and when L is finite, L=Ef(S)+R. Applications are given to infinite series of independent random variables, and to normed sums of independent, identically distributed random variables.     

Moderate deviations for stationary sequences of Hilbert-valued bounded random variables

In this paper, we derive the Moderate Deviation Principle for stationary sequences of bounded random variables with values in a Hilbert space. The conditions obtained are expressed in terms of martingale-type conditions. The main tools are martingale approximations and a new Hoeffding inequality for non-adapted sequences of Hilbert-valued random variables. Applications to Cramér-Von Mises statistics, functions of linear processes and stable Markov chains are given.     

The central limit problem for mixing sequences of random variables

Summary In this paper the central limit problem is solved for sums of random variables having bounded variances and satisfying certain mixing conditions. In case of a stochastic process these mixing conditions essentially say that as time passes events concerning the future of the process are almost independent from the events in the past. It turns out that the class of limit laws for sums of mixing random variables is exactly the same as for the bounded variances case of independent random variables. We also shall give criteria for convergence to any specified law of this class of possible limit laws. Finally we shall derive the central limit theorem involving a kind of Lindeberg-Feller condition and as a corollary thereof a kind of Ljapounov theorem.     

Comparison of numbers of records in the sequences of discrete and continuous random variables

Different record achievements are fixed in many domains of human activities. This process very often happens with some rate of digitization (up to seconds, meters, or thousands of individuals) of the observed results. By the examples of exponential and geometrical distributions, it is shown how such a type of the transitions from continuous to discrete distributions may vary the numbers of the record values in the corresponding sequences of the random variables.     

Convergence of sequences of pairwise independent random variables

In spite of the fact that the tail -algebra of a sequence of pairwise independent random variables may not be trivial, we have discovered that if such a sequence converges in probability or almost everywhere, then the limit has to be a constant. This enables us to provide necessary and sufficient conditions for its convergence, in terms of its marginal distribution functions.