On the deviation between the distributions of sums and maximum sums of IID random vectors

Summary For a sequence of independent and identically distributed random vectors, with finite moment of order less than or equal to the second, the rate at which the deviation between the distribution functions of the vectors of partial sums and maximums of partial sums is obtained both when the sample size is fixed and when it is random, satisfying certain regularity conditions. When the second moments exist the rate is of ordern ^−1/4 (in the fixed sample size case). Two applications are given, first, we compliment some recent work of Ahmad (1979,J. Multivariate Anal.,9, 214–222) on rates of convergence for the vector of maximum sums and second, we obtain rates of convergence of the concentration functions of maximum sums for both the fixed and random sample size cases.     

Conditioned rates of convergence in the CLT for sums and maximum sums

An interesting recent result of Landers and Roggé (1977, Ann. Probability5, 595–600) is investigated further. Rates of convergence in the conditioned central limit theorem are developed for partial sums and maximum partial sums, with positive mean and zero mean separately, of sequences of independent identically distributed random variables. As corollaries we obtain a conditioned central limit theorem for maximum partial sums both for positive and zero mean cases.     

Stochastic levi sums

An analogy of Levi sums is considered for a class of stochastic partial differential equations to construct their stochastic fundamental solutions. These notions are shown to coincide with Donsker's delta functions, typical generalized Wiener functionals, which have been studied in the frame of the Malliavin calculus. © 1994 John Wiley & Sons, Inc.     

Laplace approximations for sums of independent random vectors

Let X _i , i∈N, be i.i.d. B-valued random variables, where B is a real separable Banach space. Let Φ be a mapping B→R. Under a central limit theorem assumption, an asymptotic evaluation of Z _n = E (exp (n Φ (∑ _{i =1} ⁿ X _i /n))), up to a factor (1 + o(1)), has been gotten in Bolthausen [1]. In this paper, we show that the same asymptotic evaluation can be gotten without the central limit theorem assumption. Received: 19 September 1997 / Revised version:22 April 1999     

On the moments of some first passage times for sums of dependent random variables

Let S_n,n = 1, 2, …, denote the partial sums of integrable random variables. No assumptions about independence are made. Conditions for the finiteness of the moments of the first passage times N(c) = min {n: S_n>ca(n)}, where c ≥ 0and a(y) is a positive continuous function on [0, ∞), such that a(y) = o(y)as y → ∞, are given. With the further assumption that a(y) = yP,0 ≤ p < 1, a law of large numbers and the asymptotic behaviour of the moments when c → ∞ are obtained. The corresponding stopped sums are also studied.     

Infinite paths with bounded or recurrent partial sums

We consider problems of the following type. Assign independently to each vertex of the square lattice the value +1, with probability p, or −1, with probability 1 −p. We ask whether an infinite path π exists, with the property that the partial sums of the ±1s along π are uniformly bounded, and whether there exists an infinite path π' with the property that the partial sums along π' are equal to zero infinitely often. The answers to these question depend on the type of path one allows, the value of p and the uniform bound specified. We show that phase transitions occur for these phenomena. Moreover, we make a surprising connection between the problem of finding a path to infinity (not necessarily self-avoiding, but visiting each vertex at most finitely many times) with a given bound on the partial sums, and the classical Boolean model with squares around the points of a Poisson process in the plane. For the recurrence problem, we also show that the probability of finding such a path is monotone in p, for p≥?. Received: 10 January 2000 / Revised version: 14 August 2000 / Published online: 9 March 2001     

On the almost sure central limit theorem for randomly indexed sums

Let {S_n, n ≥ 1} be partial sums of independent identically distributed random variables. The almost sure version of CLT is generalized on the case of randomly indexed sums {S_Nn, n ≥ 1}, where {N_n, n ≥ 1} is a sequence of positive integer‐valued random variables independent of {S_n, n ≥ 1}. The affects of nonrandom centering and norming are considered too (© 2009 WILEY‐VCH Verlag GmbH & Co. KGaA, Weinheim)     

Optimal non-isotropic gevrey exponents for sums of squares of vector fields

We prove sharp non-isotropic Gevrey hypoellipticiy for a class ofo partial differential operators that are sums of squares of real vector fields (and their generalizations) satisfying the Hormander bracket condition. These include the Baouendi-Goulaouic operator. Our results, which refine those of Chirst [8], are proved entirely by L ² methods and a careful study of brackets of vector fields. Applications to a recent conjecture of Traves are given.     

An arithmetical property of Rademacher sums

Let ε = {ε_i, i ≥ 1} be a Rademacher sequence, with partial sums S_n = ε₁ +… + ε_n, n ≥ 1. Let N_k be the k-th even integer such that N_k∥S_Nk². We prove that there exists a positive real s, of which the value is explicitly given, such that for any , almost surely.     

The central limit theorem for sums of trimmed variables with heavy tails

Trimming is a standard method to decrease the effect of large sample elements in statistical procedures, used, e.g., for constructing robust estimators and tests. Trimming also provides a profound insight into the partial sum behavior of i.i.d. sequences. There is a wide and nearly complete asymptotic theory of trimming, with one remarkable gap: no satisfactory criteria for the central limit theorem for modulus trimmed sums have been found, except for symmetric random variables. In this paper we investigate this problem in the case when the variables are in the domain of attraction of a stable law. Our results show that for modulus trimmed sums the validity of the central limit theorem depends sensitively on the behavior of the tail ratio P(X>t)/P(|X|>t) of the underlying variable X as t→∞ and paradoxically, increasing the number of trimmed elements does not generally improve partial sum behavior.     

A strong approximation for logarithmic averages of partial sums of random variables

LetS _n be the partial sums of -mixing stationary random variables and letf(x) be a real function. In this note we give sufficient conditions under which the logarithmic average off(S _n / _n ) converges almost surely to _– f(x)d(x). We also obtain strong approximation forH(n)= _k=1 ⁿ k ^–1 f(S _k /k)=logn _– f(x)d(x) which will imply the asymptotic normality ofH(n)/log^1/2 n. But for partial sums of i.i.d. random variables our results will be proved under weaker moment condition than assumed for -mixing random variables.     

Non-Fourier-Lebesgue trigonometric series with nonnegative partial sums

It is proved that a trigonometric cosine series of the form _n ^Emphasis>=0/ a _n cos(nx) with nonnegative coefficients can be constructed in such a way that all of its partial sums are positive on the real axis. It converges to zero almost everywhere and is not a Fourier-Lebesgue series. Some other properties of trigonometric series with nonnegative partial sums are also studied.Translated fromMatematicheskie Zametki, Vol. 59, No. 1, pp. 24–41, January, 1996.     

A note on rates of convergence in the multidimensional CLT for maximum partial sums

In this note we obtain rates of convergence in the central limit theorem for certain maximum of coordinate partial sums of independent identically distributed random vectors having positive mean vector and a nonsingular correlation matrix. The results obtained are in terms of rates of convergence in the multidimensional central limit theorem. Thus under the conditions of Sazonov (1968, Sankhya, Series A30 181–204, Theorem 2), we have the same rate of convergence for the vector of coordinate maximums. Other conditions for the multidimensional CLT are also discussed, c.f., Bhattachaya (1977, Ann. Probability 5 1–27). As an application of one of the results we obtain a multivariate extension of a theorem of Rogozin (1966, Theor. Probability Appl. 11 438–441).     

Bernoulli numbers and sums of powers

There have been many studies of Bernoulli numbers since Jakob Bernoulli first used the numbers to compute sums of powers, 1 ^p + 2 ^p + 3 ^p + ··· + n^p , where n is any natural number and p is any non-negative integer. By examining patterns of these sums for the first few powers and the relation between their coefficients and Bernoulli numbers, the author hypothesizes and proves a new recursive algorithm for computing Bernoulli numbers, sums of powers, as well as m-ford sums of powers, which enrich the existing literatures of Bernoulli numbers.     

Equidistribution of Gauss sums and Kloosterman sums

Let m be a positive integer. Fix a nontrivial additive character for each finite field F_q. To state the first result of this paper, we also fix r distinct multiplicative characters ₁,...,_r for each finite field F_q with more than r elements. We shall prove that when varies over multiplicative characters of F_q other than the m-th roots of the r-tuples of angles of Gauss sums are asymptotically equidistributed on the r-dimensional torus (S¹)^r as q goes to infinity.The n-dimensional Kloosterman sum over F_q at a F_q^× is One can define the angle (q,a) of Kl_n(q,a) in a suitable way. We shall prove that when a varies over nonzero elements of F_q, the q–1 angles (q,a^m) of Kloosterman sums are asymptotically equidistributed as q goes to infinity.Mathematics Subject Classification (2000) 11L05, 14F20     

The influence of sample extremes on the maximum of partial sums of independent random variables

Many papers exist dealing with the distribution of the maximum of partial sums of independent random variables and studying the relationship between the sums and the sample extremes (see [L. Takacs,Combinator Methods in the Theory of Stochastic Processes, Wiley, New York (1967)], [J. Calambos,The Asymptotic Theory of Extreme Order Statistics, Wiley, New York (1978)]). The present paper deals with the effect the sample extremes have on the distribution of the maximum of sums of independent random variables. Proceedings of the Seminar on Stability Problems for Stochastic Models, Moscow, 1993     

Exponential sums on An. III

We give two applications of our earlier work [4]. We compute the p-adic cohomology of certain exponential sums on A ⁿ involving a polynomial whose homogeneous component of highest degree defines a projective hypersurface with at worst weighted homogeneous isolated singularities. This study was motivated by recent work of García [9]. We also compute the p-adic cohomology of certain exponential sums on A ⁿ whose degree is divisible by the characteristic. Received: 12 October 1999     

Critical fluctuations of sums of weakly dependent random vectors

Summary LetS _n be sums of iid random vectors taking values in a Banach space andF be a smooth function. We study the fluctuations ofS _n under the transformed measureP _n given byd P _n/d P=exp (nF(S _n/n))/Z _n. If degeneracy occurs then the projection ofS _n onto the degenerate subspace, properly centered and scaled, converges to a non-Gaussian probability measure with the degenerate subspace as its support. The projection ofS _n onto the non-degenerate subspace, scaled with the usual order converges to a Gaussian probability measure with the non-degenerate subspace as its support. The two projective limits are in general dependent. We apply this theory to the critical mean field Heisenberg model and prove a central limit type theorem for the empirical measure of this model.Supported by a grant from the Swiss National Science Foundation (21–29833.90)     

Limiting distribution of sums of nonnegative stationary random variables

Summary Let {X _n,j,−∞<j<∞∼,n≧1, be a sequence of stationary sequences on some probability space, with nonnegative random variables. Under appropriate mixing conditions, it is shown thatS _n=X_n,1+…+X _n,n has a limiting distribution of a general infinitely divisible form. The result is applied to sequences of functions {f _n(x)∼ defined on a stationary sequence {X _j∼, whereX _n.f=f_n(X_j). The results are illustrated by applications to Gaussian processes, Markov processes and some autoregressive processes of a general type. This paper represents results obtained at the Courant Institute of Mathematical Sciences, New York University, under the sponsorship of the National Sciences Foundation, Grant MCS 82-01119.     

Uniformly distributed sums of bernoulli variables

Sums of independent, identically distributed (iid) binomial variates have binomial distributions; yet it is possible to construct a sequence of binomial distributions over {0, 1} for variatesX ₁,X ₂, ... such that all partial sumsY _i =X ₁ + ... +X _i have uniform distributions. The price to pay is to give up the iid condition. Requiring the property of only one sum does not alleviate the situation much.It is also possible to generate on a computerm × n-matrices, of 0–1 binomial variates with uniformly distributed row and column sums of all major submatrices, but only for smallm andn. Even a three-dimensional 2 × 2 × 2 array can have a similar property.Other target distributions than the rectangular are possible, but cumbersome. An example with smaller variance is given.The results were needed for simulating the performance of some Operations Research algorithms.Dedicated to Peter Naur on the occasion of his 60th birthday