Improved concentration bounds for sums of independent sub-exponential random variables

Iosif Pinelis

DOI: https://doi.org/10.1016/j.spl.2022.109666

2022-12-01

Abstract:Known Bernstein-type upper bounds on the tail probabilities for sums of independent zero-mean sub-exponential random variables are improved in several ways at once. The new upper bounds have a certain optimality property.

statistics & probability

What problem does this paper attempt to address?

Approximation of the Tail Probability of Randomly Weighted Sums and Applications

Yi Zhang,Xinmei Shen,Chengguo Weng

DOI: https://doi.org/10.1016/j.spa.2008.03.004

IF: 1.43

2009-01-01

Stochastic Processes and their Applications

Abstract:Consider the problem of approximating the tail probability of randomly weighted sums ∑i=1nΘiXi and their maxima, where {Xi,i≥1} is a sequence of identically distributed but not necessarily independent random variables from the extended regular variation class, and {Θi,i≥1} is a sequence of nonnegative random variables, independent of {Xi,i≥1} and satisfying certain moment conditions. Under the assumption that {Xi,i≥1} has no bivariate upper tail dependence along with some other mild conditions, this paper establishes the following asymptotic relations: Pr(max1≤k≤n∑i=1kΘiXi>x)∼Pr(∑i=1nΘiXi>x)∼∑i=1nPr(ΘiXi>x), and Pr(max1≤k<∞∑i=1kΘiXi>x)∼Pr(∑i=1∞ΘiXi+>x)∼∑i=1∞Pr(ΘiXi>x), as x→∞. In doing so, no assumption is made on the dependence structure of the sequence {Θi,i≥1}.
Improved nonuniform Berry--Esseen-type bounds

Iosif Pinelis

DOI: https://doi.org/10.48550/arXiv.1109.0741

2011-09-05

Abstract:New nonuniform Berry--Esseen-type bounds for sums of independent random variables are obtained, motivated by recent studies concerning such bounds for nonlinear statistics. The proofs are based on the Chen--Shao concentration techniques within the framework of the Stein method.

Probability,Statistics Theory
On Increments of Sums of Independent Non-Identically Distributed Random-Variables

ZY LIN

1988-01-01

Abstract:The paper gives some theorems about the increments of partial sums of independent, but not necessarily independent identically distributed random variables. The main results are ideal and corresponding to those for i. i. d. random variables. Therefore these sharp results of Hanson and Russo (1983) answer some of their questions.
The Convergence of the Sums of Independent Random Variables under the Sub-Linear Expectations

Li Xin Zhang

DOI: https://doi.org/10.1007/s10114-020-8508-0

2020-01-01

Acta Mathematica Sinica English Series

Abstract:Let {Xn;n ≥ 1} be a sequence of independent random variables on a probability space (Ω,F,P) and $${S_n} = \sum\nolimits_{k = 1}^n {{X_k}} $$. It is well-known that the almost sure convergence, the convergence in probability and the convergence in distribution of Sn are equivalent. In this paper, we prove similar results for the independent random variables under the sub-linear expectations, and give a group of sufficient and necessary conditions for these convergence. For proving the results, the Levy and Kolmogorov maximal inequalities for independent random variables under the sub-linear expectation are established. As an application of the maximal inequalities, the sufficient and necessary conditions for the central limit theorem of independent and identically distributed random variables are also obtained.
On Sums of Conditionally Independent Subexponential Random Variables

Sergey Foss,Andrew Richards

DOI: https://doi.org/10.48550/arXiv.0806.0490

2009-03-18

Abstract:The asymptotic tail behaviour of sums of independent subexponential random variables is well understood, one of the main characteristics being the principle of the single big jump. We study the case of dependent subexponential random variables, for both deterministic and random sums, using a fresh approach, by considering conditional independence structures on the random variables. We seek sufficient conditions for the results of the theory with independent random variables still to hold. For a subexponential distribution, we introduce the concept of a boundary class of functions, which we hope will be a useful tool in studying many aspects of subexponential random variables. The examples we give in the paper demonstrate a variety of effects owing to the dependence, and are also interesting in their own right.

Probability
Bounds for the Concentration Functions of Random Sums under Relaxed Moment Conditions

V. Yu. Korolev,A. V. Dorofeeva

DOI: https://doi.org/10.1137/s0040585x97t988502

2018-01-01

Theory of Probability and Its Applications

Abstract:Estimates are constructed for the deviation of the concentration functions of sums of independent random variables with finite variances from the folded normal distribution function without any assumptions concerning the existence of the moments of summands of higher orders. The results obtained are extended to Poisson-binomial, binomial, and Poisson random sums. Under the same assumptions, bounds are obtained for the approximation of the concentration functions of mixed Poisson random sums by the corresponding limit distributions. In particular, bounds are put forward for the accuracy of approximation of the concentration functions of geometric, negative binomial, and Sichel random sums by exponential, folded variance gamma, and folded Student distributions. Numerical estimates of all the constants involved are written down explicitly.

statistics & probability
Strong Limit Theorems for Extended Independent Random Variables and Extended Negatively Dependent Random Variables under Sub-Linear Expectations

Zhang Li-Xin

DOI: https://doi.org/10.1007/s10473-022-0203-z

2022-01-01

Acta Mathematica Scientia

Abstract:Limit theorems for non-additive probabilities or non-linear expectations are challenging issues which have attracted a lot of interest recently. The purpose of this paper is to study the strong law of large numbers and the law of the iterated logarithm for a sequence of random variables in a sub-linear expectation space under a concept of extended independence which is much weaker and easier to verify than the independence proposed by Peng [20]. We introduce a concept of extended negative dependence which is an extension of the kind of weak independence and the extended negative independence relative to classical probability that has appeared in the recent literature. Powerful tools such as moment inequality and Kolmogorov's exponential inequality are established for these kinds of extended negatively independent random variables, and these tools improve a lot upon those of Chen, Chen and Ng [1]. The strong law of large numbers and the law of iterated logarithm are also obtained by applying these inequalities.
On Asymptotically Tight Tail Bounds for Sums of Geometric and Exponential Random Variables

Yaonan Jin,Yingkai Li,Yining Wang,Yuan Zhou

2019-01-01

Abstract:In this note we prove bounds on the upper and lower probability tails of sums of independent geometric or exponentially distributed random variables. We also prove negative results showing that our established tail bounds are asymptotically tight.
LIL and the Approximation of Rectangular Sums of B-valued Random Variables when Extreme Terms Are Excluded

Li Xin Zhang

DOI: https://doi.org/10.1007/s10114-002-0181-y

2002-01-01

Acta Mathematica Sinica English Series

Abstract:Let {X,X n ̄; n̄ ∈ ℕ d } be a field of i.i.d. random variables indexed by d-tuples of positive integers and taking values in a Banach space ${\user1{\mathcal{B}}}$ and let $X^{{{\left( r \right)}}}_{{\ifmmode\expandafter\bar\else\expandafter\=\fi{n}}} = X_{{\ifmmode\expandafter\bar\else\expandafter\=\fi{m}}} \;{\text{if}}\;{\left\| {X_{{\ifmmode\expandafter\bar\else\expandafter\=\fi{m}}} } \right\|}$ is the r-th maximum of {∥X k ̄∥ ; k̄ ≤ n̄}. Let S n ̄ =∑ k̄ ≤ n X k̄ and ${}^{{{\left( r \right)}}}S_{{\ifmmode\expandafter\bar\else\expandafter\=\fi{n}}} = S_{{\ifmmode\expandafter\bar\else\expandafter\=\fi{n}}} - {\left( {X^{{{\left( 1 \right)}}}_{{\ifmmode\expandafter\bar\else\expandafter\=\fi{n}}} + \cdots + X^{{{\left( r \right)}}}_{{\ifmmode\expandafter\bar\else\expandafter\=\fi{n}}} } \right)}$. We approximate the trimmed sums (r) S n ̄ by a Brownian sheet and obtain sufficient and necessary conditions for (r) S n to satisfy the compact and functional laws of the iterated logarithm. These results improve the previous works by Morrow (1981), Li and Wu (1989) and Ledoux and Talagrand (1990).
Improved Semi-Parametric Bounds for Tail Probability and Expected Loss: Theory and Applications

Zhaolin Li,Artem Prokhorov

DOI: https://doi.org/10.48550/arXiv.2404.02400

2024-05-03

Abstract:Many management decisions involve accumulated random realizations for which the expected value and variance are assumed to be known. We revisit the tail behavior of such quantities when individual realizations are independent, and we develop new sharper bounds on the tail probability and expected linear loss. The underlying distribution is semi-parametric in the sense that it remains unrestricted other than the assumed mean and variance. Our bounds complement well-established results in the literature, including those based on aggregation, which often fail to take full account of independence and use less elegant proofs. New insights include a proof that in the non-identical case, the distributions attaining the bounds have the equal range property, and that the impact of each random variable on the expected value of the sum can be isolated using an extension of the Korkine identity. We show that the new bounds %not only complement the extant results but also open up abundant practical applications, including improved pricing of product bundles, more precise option pricing, more efficient insurance design, and better inventory management. For example, we establish a new solution to the optimal bundling problem, yielding a 17\% uplift in per-bundle profits, and a new solution to the inventory problem, yielding a 5.6\% cost reduction for a model with 20 retailers.

Econometrics,Methodology,Other Statistics
Tail bounds for sums of geometric and exponential variables

Svante Janson

DOI: https://doi.org/10.48550/arXiv.1709.08157

2017-09-24

Abstract:We give explicit bounds for the tail probabilities for sums of independent geometric or exponential variables, possibly with different parameters.

Probability
Improved bounds on tails of convolutions of compound distributions: Application to ruin probabilities for the risk process perturbed by diffusion

Stathis Chadjiconstantinidis,Panos Xenos

DOI: https://doi.org/10.1016/j.cam.2024.115835

IF: 2.872

2024-02-17

Journal of Computational and Applied Mathematics

Abstract:In this paper we examine the convolution of a compound distribution with another given distribution. For this, new general upper and lower bounds on the tail probabilities of the convolution of a compound geometric distribution are derived. Also, such bounds are obtained under the truncated Cramér-Lundberg condition for the heavy-tailed case. Applications to the ruin probabilities of compound Poisson claims processes perturbed by diffusion as well as some new upper bounds are given. Also, several numerical examples to illustrate the effectiveness of the bounds are given.

mathematics, applied
Concentration Inequalities for Sums of Markov Dependent Random Matrices

Joe Neeman,Bobby Shi,Rachel Ward

2023-10-17

Abstract:We give Hoeffding and Bernstein-type concentration inequalities for the largest eigenvalue of sums of random matrices arising from a Markov chain. We consider time-dependent matrix-valued functions on a general state space, generalizing previous that had only considered Hoeffding-type inequalities, and only for time-independent functions on a finite state space. In particular, we study a kind of noncommutative moment generating function, give tight bounds on it, and use a method of Garg et al. to turn this into tail bounds. Our proof proceeds spectrally, bounding the norm of a certain perturbed operator. In the process we make an interesting connection to dynamical systems and Banach space theory to prove a crucial result on the limiting behavior of our moment generating function that may be of independent interest.

Probability
Combinatorial anti-concentration inequalities, with applications

JACOB FOX,MATTHEW KWAN,LISA SAUERMANN

DOI: https://doi.org/10.1017/s0305004120000183

2021-06-30

Mathematical Proceedings of the Cambridge Philosophical Society

Abstract:Abstract We prove several different anti-concentration inequalities for functions of independent Bernoulli-distributed random variables. First, motivated by a conjecture of Alon, Hefetz, Krivelevich and Tyomkyn, we prove some “Poisson-type” anti-concentration theorems that give bounds of the form 1/ e + o (1) for the point probabilities of certain polynomials. Second, we prove an anti-concentration inequality for polynomials with nonnegative coefficients which extends the classical Erdős–Littlewood–Offord theorem and improves a theorem of Meka, Nguyen and Vu for polynomials of this type. As an application, we prove some new anti-concentration bounds for subgraph counts in random graphs.

mathematics
Probability Inequalities for Sums of Independent Unbounded Random Variables

Di-xin Zhang,Zhi-cheng Wang

DOI: https://doi.org/10.1007/bf02437750

2001-01-01

Abstract:The tail probability inequalities for the sum of independent unbounded random variables on a probability space (Ω, T, P) were studied and a new method was proposed to treat the sum of independent unbounded random variables by truncating the original probability space (Ω, T, P). The probability exponential inequalities for sums of independent unbounded random variables were given. As applications of the results, some interesting examples were given. The examples show that the method proposed in the paper and the results of the paper are quite useful in the study of the large sample properties of the sums of independent unbounded random variables.
An Inequality for the Sum of Independent Bounded Random Variables

Christopher R. Dance

DOI: https://doi.org/10.48550/arXiv.1210.6484

2012-10-24

Abstract:We give a simple inequality for the sum of independent bounded random variables. This inequality improves on the celebrated result of Hoeffding in a special case. It is optimal in the limit where the sum tends to a Poisson random variable.

Probability
Nearly Tight Universal Bounds for the Binomial Tail Probabilities

Huangjun Zhu,Zihao Li,Masahito Hayashi

2022-01-01

Abstract:We derive simple but nearly tight upper and lower bounds for the binomial lower tail probability (with straightforward generalization to the upper tail probability) that apply to the whole parameter regime. These bounds are easy to compute and are tight within a constant factor of 89/44. Moreover, they are asymptotically tight in the regimes of large deviation and moderate deviation. By virtue of a surprising connection with Ramanujan's equation, we also provide strong evidences suggesting that the lower bound is tight within a factor of 1.26434. It may even be regarded as the natural lower bound, given its simplicity and appealing properties. Our bounds significantly outperform the familiar Chernoff bound and reverse Chernoff bounds known in the literature and may find applications in various research areas.
Empirical Bernstein in smooth Banach spaces

Diego Martinez-Taboada,Aaditya Ramdas

2024-09-10

Abstract:Existing concentration bounds for bounded vector-valued random variables include extensions of the scalar Hoeffding and Bernstein inequalities. While the latter is typically tighter, it requires knowing a bound on the variance of the random variables. We derive a new vector-valued empirical Bernstein inequality, which makes use of an empirical estimator of the variance instead of the true variance. The bound holds in 2-smooth separable Banach spaces, which include finite dimensional Euclidean spaces and separable Hilbert spaces. The resulting confidence sets are instantiated for both the batch setting (where the sample size is fixed) and the sequential setting (where the sample size is a stopping time). The confidence set width asymptotically exactly matches that achieved by Bernstein in the leading term. The method and supermartingale proof technique combine several tools of Pinelis (1994) and Waudby-Smith and Ramdas (2024).

Statistics Theory
Randomized and Exchangeable Improvements of Markov's, Chebyshev's and Chernoff's Inequalities

Aaditya Ramdas,Tudor Manole

2023-05-10

Abstract:We present simple randomized and exchangeable improvements of Markov's inequality, as well as Chebyshev's inequality and Chernoff bounds. Our variants are never worse and typically strictly more powerful than the original inequalities. The proofs are short and elementary, and can easily yield similarly randomized or exchangeable versions of a host of other inequalities that employ Markov's inequality as an intermediate step. We point out some simple statistical applications involving tests that combine dependent e-values. In particular, we uniformly improve the power of universal inference, and obtain tighter betting-based nonparametric confidence intervals. Simulations reveal nontrivial gains in power (and no losses) in a variety of settings.

Statistics Theory,Information Theory,Probability,Methodology
Bounding probability of small deviation on sum of independent random variables: Combination of moment approach and Berry-Esseen theorem

Jiayi Guo,Simai He,Zi Ling,Yicheng Liu

DOI: https://doi.org/10.48550/arXiv.2003.03197

2020-03-06

Abstract:In the context of bounding probability of small deviation, there are limited general tools. However, such bounds have been widely applied in graph theory and inventory management. We introduce a common approach to substantially sharpen such inequality bounds by combining the semidefinite optimization approach of moments problem and the Berry-Esseen theorem. As an application, we improve the lower bound of Feige's conjecture from 0.14 to 0.1798.

Optimization and Control,Probability

Improved concentration bounds for sums of independent sub-exponential random variables

Approximation of the Tail Probability of Randomly Weighted Sums and Applications

Improved nonuniform Berry--Esseen-type bounds

On Increments of Sums of Independent Non-Identically Distributed Random-Variables

The Convergence of the Sums of Independent Random Variables under the Sub-Linear Expectations

On Sums of Conditionally Independent Subexponential Random Variables

Bounds for the Concentration Functions of Random Sums under Relaxed Moment Conditions

Strong Limit Theorems for Extended Independent Random Variables and Extended Negatively Dependent Random Variables under Sub-Linear Expectations

On Asymptotically Tight Tail Bounds for Sums of Geometric and Exponential Random Variables

LIL and the Approximation of Rectangular Sums of B-valued Random Variables when Extreme Terms Are Excluded

Improved Semi-Parametric Bounds for Tail Probability and Expected Loss: Theory and Applications

Tail bounds for sums of geometric and exponential variables

Improved bounds on tails of convolutions of compound distributions: Application to ruin probabilities for the risk process perturbed by diffusion

Concentration Inequalities for Sums of Markov Dependent Random Matrices

Combinatorial anti-concentration inequalities, with applications

Probability Inequalities for Sums of Independent Unbounded Random Variables

An Inequality for the Sum of Independent Bounded Random Variables

Nearly Tight Universal Bounds for the Binomial Tail Probabilities

Empirical Bernstein in smooth Banach spaces

Randomized and Exchangeable Improvements of Markov's, Chebyshev's and Chernoff's Inequalities

Bounding probability of small deviation on sum of independent random variables: Combination of moment approach and Berry-Esseen theorem