Binomial upper bounds on generalized moments and tail probabilities of (super)martingales with differences bounded from above

نویسنده

  • Iosif Pinelis
چکیده

Let (S0, S1, . . . ) be a supermartingale relative to a nondecreasing sequence of σ-algebras H≤0, H≤1, . . . , with S0 ≤ 0 almost surely (a.s.) and differences Xi := Si − Si−1. Suppose that Xi ≤ d and Var(Xi|H≤i−1) ≤ σ 2 i a.s. for every i = 1, 2, . . . , where d > 0 and σi > 0 are non-random constants. Let Tn := Z1 + · · · + Zn, where Z1, . . . , Zn are i.i.d. r.v.’s each taking on only two values, one of which is d, and satisfying the conditions EZi = 0 and VarZi = σ := 1 n (σ 1 + · · · + σ n). Then, based on a comparison inequality between generalized moments of Sn and Tn for a rich class of generalized moment functions, the tail comparison inequality P(Sn ≥ y) ≤ cP (Tn ≥ y + h 2 ) ∀y ∈ R is obtained, where c := e/2 = 3.694 . . . , h := d + σ/d, and the function y 7→ P(Tn ≥ y) is the least log-concave majorant of the linear interpolation of the tail function y 7→ P(Tn ≥ y) over the lattice of all points of the form nd+kh (k ∈ Z). An explicit formula for P(Tn ≥ y+ h 2 ) is given. Another, similar bound is given under somewhat different conditions. It is shown that these bounds improve significantly upon known bounds.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Exact inequalities for sums of asymmetric random variables, with applications

Let BS1, . . . ,BSn be independent identically distributed random variables each having the standardized Bernoulli distribution with parameter p ∈ (0, 1). Let m∗(p) := (1 + p+ 2 p2)/(2 √ p − p2 +4 p2) if 0 < p 6 1 2 and m∗(p) := 1 if 1 2 6 p < 1. Let m > m∗(p). Let f be such a function that f and f ′′ are nondecreasing and convex. Then it is proved that for all nonnegative numbers c1, . . . , c...

متن کامل

Optimal binomial, Poisson, and normal left-tail domination for sums of nonnegative random variables

Exact upper bounds on the generalized moments E f(Sn) of sums Sn of independent nonnegative random variables Xi for certain classes F of nonincreasing functions f are given in terms of (the sums of) the first two moments of the Xi’s. These bounds are of the form E f(η), where the random variable η is either binomial or Poisson depending on whether n is fixed or not. The classes F contain, and a...

متن کامل

Bounds on Tail Probabilities of Discrete Distributions

There has not been much work concerning bounds on tail probabilities of discrete distributions+ In the work of Johnson, Kotz, and Kemp @9# , for example, only two ~quite bad! upper bounds on Poisson tail probabilities are quoted+ This article examines several more generally applicable methods to derive lower and upper bounds on tail probabilities of discrete distributions+ Section 2 presents an...

متن کامل

Binomial moments for divisible self-dual codes

For self-dual codes with all weights divisible by an integer greater than one, the minimum distance is bounded by the Mallows-Sloane upper bounds and by their improvements due to Krasikov-Litsyn and Rains. We obtain the improved upper bounds from short relations with constant coefficients on suitable binomial moments of the codes. In this approach, the Mallows-Sloane bounds are analogues of the...

متن کامل

Estimation of Buffer Overflow Probabilities and Economies of Scale in ATM Multiplexers by Analysis of a Model of Packetized Voice Traffic

Estimation of Buffer Overflow Probabilities and Economies of Scale in ATM M ultiplexers by Analysis of a M odel of Packetized Voice Traffic by Paul J. Farrell We obtain upper bounds on the probability of buffer overflow for an ATM multiplexer of L identical packetized voice sources The multiplexer is modelled by a FCFS single server queue The arrivals at the multiplexer are a homogenous superpo...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2005