Some notes on moment inequalities for heavy-tailed distributions

Paul Buterus,Holger Sambale

2024-10-29

Abstract:We investigate the relation between moments and tails of heavy-tailed (in particular, Pareto-type) distributions. We also discuss the sharpness of our results in a number of examples under certain regularity conditions like log-convexity. Moreover, we derive concentration bounds for polynomial chaos of any order $d$.

Probability

What problem does this paper attempt to address?

On Log-Concave-Tailed Chaos and the Restricted Isometry Property

Guozheng Dai,Zhonggen Su,Vladimir Ulyanov,Hanchao Wang

2024-01-01

Abstract:We develop a novel decoupling inequality for nonhomogeneous order 2 non-Gaussian chaoses. Having this decoupling inequality, we show two types of the p-th moment bounds for the suprema of a log-concave-tailed chaos process based on some chaining techniques and the generalized majorizing measure theorems for canonical processes. As applications, we show the restricted isometry property of partial random circulant matrices and time-frequency structured random matrices induced by standard α-subexponential vectors, 1≤α≤ 2, which extends the previously known results for the sub-Gaussian case.
Concentration and moment inequalities for heavy-tailed random matrices

Moritz Jirak,Stanislav Minsker,Yiqiu Shen,Martin Wahl

2024-07-18

Abstract:We prove Fuk-Nagaev and Rosenthal-type inequalities for the sums of independent random matrices, focusing on the situation when the norms of the matrices possess finite moments of only low orders. Our bounds depend on the "intrinsic" dimensional characteristics such as the effective rank, as opposed to the dimension of the ambient space. We illustrate the advantages of such results in several applications, including new moment inequalities for sample covariance matrices and the corresponding eigenvectors of heavy-tailed random vectors. Moreover, we demonstrate that our techniques yield sharpened versions of the moment inequalities for empirical processes.

Probability,Statistics Theory
Concentration Inequalities for General Functions of Heavy-Tailed Random Variables

Shaojie Li,Yong Liu

2024-01-01

Abstract:Concentration inequalities play an essential role in the study of machine learning and high dimensional statistics. In this paper, we obtain unbounded analogues of the popular bounded difference inequality for functions of independent random variables with heavy-tailed distributions. The main results provide a general framework applicable to all heavy-tailed distributions with finite variance. To illustrate the strength of our results, we present applications to sub-exponential tails, sub-Weibull tails, and heavier polynomially decaying tails. Applied to some standard problems in statistical learning theory (vector valued concentration, Rademacher complexity, and algorithmic stability), we show that these inequalities allow an extension of existing results to heavy-tailed distributions up to finite variance.
The generalized lognormal distribution and the Stieltjes moment problem

Christian Kleiber

DOI: https://doi.org/10.1007/s10959-013-0477-0

2013-01-08

Abstract:This paper studies a Stieltjes-type moment problem defined by the generalized lognormal distribution, a heavy-tailed distribution with applications in economics, finance and related fields. It arises as the distribution of the exponential of a random variable following a generalized error distribution, and hence figures prominently in the EGARCH model of asset price volatility. Compared to the classical lognormal distribution it has an additional shape parameter. It emerges that moment (in)determinacy depends on the value of this parameter: for some values, the distribution does not have finite moments of all orders, hence the moment problem is not of interest in these cases. For other values, the distribution has moments of all orders, yet it is moment-indeterminate. Finally, a limiting case is supported on a bounded interval, and hence determined by its moments. For those generalized lognormal distributions that are moment-indeterminate Stieltjes classes of moment-equivalent distributions are presented.

Probability,Other Statistics
Self-normalized partial sums of heavy-tailed time series

Muneya Matsui,Thomas Mikosch,Olivier Wintenberger

DOI: https://doi.org/10.48550/arXiv.2303.17221

2023-03-30

Abstract:We study the joint limit behavior of sums, maxima and $\ell^p$-type moduli for samples taken from an $\mathbb{R}^d$-valued regularly varying stationary sequence with infinite variance. As a consequence, we can determine the distributional limits for ratios of sums and maxima, studentized sums, and other self-normalized quantities in terms of hybrid characteristic functions and Laplace transforms. These transforms enable one to calculate moments of the limits and to characterize the differences between the iid and stationary cases in terms of indices which describe effects of extremal clustering on functionals acting on the dependent sequence.

Probability,Statistics Theory
Probability and moment inequalities for additive functionals of geometrically ergodic Markov chains

Alain Durmus,Eric Moulines,Alexey Naumov,Sergey Samsonov

2023-06-15

Abstract:In this paper, we establish moment and Bernstein-type inequalities for additive functionals of geometrically ergodic Markov chains. These inequalities extend the corresponding inequalities for independent random variables. Our conditions cover Markov chains converging geometrically to the stationary distribution either in $V$-norms or in weighted Wasserstein distances. Our inequalities apply to unbounded functions and depend explicitly on constants appearing in the conditions that we consider.

Probability
On Log-Concave-Tailed Chaoses and the Restricted Isometry Property

Guozheng Dai,Zhonggen Su,Vladimir Ulyanov,Hanchao Wang

2024-05-12

Abstract:In this paper, we obtain a $p$-th moment bound for the suprema of a log-concave-tailed nonhomogeneous chaos process, which is optimal in some special cases. A crucial ingredient of the proof is a novel decoupling inequality, which may be of independent interest. With this $p$-th moment bound, we show two uniform Hanson-Wright type deviation inequalities for $\alpha$-subexponential entries ($1\le \alpha\le 2$), which recover some known results. As applications, we prove the restricted isometry property of partial random circulant matrices and time-frequency structured random matrices induced by standard $\alpha$-subexponential vectors ($1\le \alpha\le 2$), which extends the previously known results for the subgaussian case.

Probability
A new approach in two-dimensional heavy-tailed distributions

Dimitrios G. Konstantinides,Charalampos D. Passalidis

2024-08-01

Abstract:We consider a new approach in the definition of two-dimensional heavy-tailed distributions. Namely, we introduce the classes of two-dimensional long-tailed, of twodimensional dominatedly varying and of two-dimensional consistently varying distributions. Next, we define the closure property with respect to two-dimensional convolution and to joint max-sum equivalence in order to study if they are satisfied by these classes. Further we examine the joint behavior of two random sums, under generalized tail asymptotic independence. Afterward we study the closure property under scalar product and two dimensional product convolution and by these results we extended our main result in the case of jointly randomly weighted sums. Our results contained some applications where we establish the asymptotic expression of the ruin probability in a two-dimensional discrete-time risk model.

Probability
From the Central Limit Theorem to Heavy-Tailed Distributions

JW Chen

DOI: https://doi.org/10.1239/jap/1059060906

2003-01-01

Abstract:It has been observed that in many practical situations randomly stopped products of random variables have power law distributions. In this note we show that, in order for such a product to have a power law distribution, the only random indices are the exponentially distributed ones. We also consider a more general problem, which is closely related to problems concerning transformation from the central limit theorem to heavy-tailed distributions.
Some informational inequalities involving generalized trigonometric functions and a new class of generalized moments

David Puertas-Centeno,Steeve Zozor

2024-11-01

Abstract:In this work, we define a family of probability densities involving the generalized trigonometric functions defined by Drábek and Manásevich, we name Generalized Trigonometric Densities. We show their relationship with the generalized stretched Gaussians and other types of laws such as logistic, hyperbolic secant, and raised cosine probability densities. We prove that, for a fixed generalized Fisher information, this family of densities is of minimal Rényi entropy. Moreover, we introduce generalized moments via the mean of the power of a deformed cumulative distribution. The latter is defined as a cumulative of the power of the probability density function, this second parameter tuning the tail weight of the deformed cumulative distribution. These generalized moments coincide with the usual moments of a deformed probability distribution with a regularized tail. We show that, for any bounded probability density, there exists a critical value for this second parameter below which the whole subfamily of generalized moments is finite for any positive value of the first parameter (power of the moment). In addition, we show that such generalized moments satisfy remarkable properties like order relation w.r.t. the first parameter, or adequate scaling behavior. Finally, we highlight that, if we constrain such a generalized moment, both the Rényi entropy and generalized Fisher information achieve respectively their maximum and minimum for the generalized trigonometric densities.

Mathematical Physics
Probability Inequalities for Weighted Distributions

Ying Li,Li Yu,Taizhong Hu

DOI: https://doi.org/10.1016/j.jspi.2011.11.023

IF: 1.095

2012-01-01

Journal of Statistical Planning and Inference

Abstract:In this paper, properties of weighted distributions for general weight functions are investigated. We establish the subadditivity [superadditivity] property of weighted distributions for log-concave [log-convex] weight functions in the sense of the usual stochastic order. The main result generalizes Lemma 2.3 in Brown (2006). Several interesting moment inequalities are presented.
Hybrid Statistics of a Random Model of Zeta over Intervals of Varying Length

Christine Chang

2024-04-13

Abstract:Arguin, Dubach & Hartung recently conjectured that an intermediate regime exists between IID and log-correlated statistics for extreme values of a random model of the Riemann zeta function. For the same model, we prove a matching upper and lower tail for the distribution of its maximum. This tail interpolates between that of the two aforementioned regimes. We apply the result to yield a new sharp estimate on moments over short intervals, generalizing a result by Harper. In particular, we observe a hybrid regime for moments with a distinctive transition to the IID regime for intervals of length larger than $\exp(\sqrt{\log \log T})$.

Probability,Number Theory
Heavy-Tailed Distributions and Their Applications

C Su,QH Tang

DOI: https://doi.org/10.1142/9789812702715_0014

2004-01-01

Abstract:Based on some recent investigations by the authors, this paper gives a short review on the concept of heavy-tailedness. Some commonly used and newly introduced classes of heavy-tailed distributions are examined. Applications to precise large deviations and to finite time ruin probabilities are proposed.
On asymptotic properties of high moments of compound Poisson distribution

O. Khorunzhiy

2024-11-07

Abstract:We study asymptotic behavior of the moments $M_k(\lambda)$ of the sum $X_1+\dots+X_{N_\lambda}$, where $N_\lambda$ follows the Poisson probability distribution with mean value $\lambda$ and $\{X_j\}$ is a family of i.i.d. random variables also independent from $N_\lambda$. We obtain an explicit expression for the leading term of $M_k(\lambda)$ as $k\to\infty$ and study it in dependence of the asymptotic behavior of $\lambda= \lambda_k$. In application, we establish a concentration property of maximal vertex degree of large weighted random graphs. Another application is related with a variable that arises in the studies of high moments of large random matrices. Finally, regarding three particular cases of probability distribution of $X_j$, we comment on the asymptotic behavior of certain combinatorial polynomials, including the Bell polynomials of even partitions.

Probability,Combinatorics
A note on generalization bounds for losses with finite moments

Borja Rodríguez-Gálvez,Omar Rivasplata,Ragnar Thobaben,Mikael Skoglund

2024-03-25

Abstract:This paper studies the truncation method from Alquier [1] to derive high-probability PAC-Bayes bounds for unbounded losses with heavy tails. Assuming that the $p$-th moment is bounded, the resulting bounds interpolate between a slow rate $1 / \sqrt{n}$ when $p=2$, and a fast rate $1 / n$ when $p \to \infty$ and the loss is essentially bounded. Moreover, the paper derives a high-probability PAC-Bayes bound for losses with a bounded variance. This bound has an exponentially better dependence on the confidence parameter and the dependency measure than previous bounds in the literature. Finally, the paper extends all results to guarantees in expectation and single-draw PAC-Bayes. In order to so, it obtains analogues of the PAC-Bayes fast rate bound for bounded losses from [2] in these settings.

Machine Learning
Precise asymptotics of ruin probabilities for a class of multivariate heavy-tailed distributions

Miriam Hägele

DOI: https://doi.org/10.1016/j.spl.2020.108871

2020-11-01

Abstract:<p>This article studies asymptotic approximations of ruin probabilities of multivariate random walks with heavy-tailed increments. Under our assumptions, the distributions of the increments are closely connected to multivariate subexponentiality and admit dependence between components.</p>

statistics & probability
Hanson-Wright inequality in Banach spaces

Radosław Adamczak,Rafał Latała,Rafał Meller

DOI: https://doi.org/10.1214/19-AIHP1041

2020-02-20

Abstract:We discuss two-sided bounds for moments and tails of quadratic forms in Gaussian random variables with values in Banach spaces. We state a natural conjecture and show that it holds up to additional logarithmic factors. Moreover in a certain class of Banach spaces (including $L_r$-spaces) these logarithmic factors may be eliminated. As a corollary we derive upper bounds for tails and moments of quadratic forms in subgaussian random variables, which extend the Hanson-Wright inequality.

Functional Analysis,Statistics Theory
Quantitative bounds for large deviations of heavy tailed random variables

Quirin Vogel

DOI: https://doi.org/10.30757/ALEA.v20-61

2023-08-07

Abstract:The probability that the sum of independent, centered, identically distributed, heavy-tailed random variables achieves a very large value is asymptotically equal to the probability that there exists a single summand equalling that value. We quantify the error in this approximation. We furthermore characterise of the law of the individual summands, conditioned on the sum being large.

Probability
Background risk model in presence of heavy tails under dependence

Dimitrios G. Konstantinides,Charalampos D. Passalidis

2024-05-06

Abstract:We study the background risk model under a various forms of dependence and some distribution classes of heavy tails. First, we study the asymptotic behavior of tail expectation of a portfolios with unequal heavytailedness, of non-random sums and randomly stopped sums, under some dependence structure which contains the independence as a special case. Further we investigate the asymptotic behavior of a pair of randomly weighted sums, generalizing the dependence structure among random vector components, our results contains the finite ruin probability in bi-dimensional discrete time risk model with unequal heavytailedness. Next, we carry out asymptotic analysis of the tail distortion risk measures in background risk model under various forms of dependence, with multivariate regularly varying risk distributions in each portfolio.

Probability

Some notes on moment inequalities for heavy-tailed distributions

On Log-Concave-Tailed Chaos and the Restricted Isometry Property

Concentration and moment inequalities for heavy-tailed random matrices

Concentration Inequalities for General Functions of Heavy-Tailed Random Variables

The generalized lognormal distribution and the Stieltjes moment problem

Self-normalized partial sums of heavy-tailed time series

Probability and moment inequalities for additive functionals of geometrically ergodic Markov chains

On Log-Concave-Tailed Chaoses and the Restricted Isometry Property

A new approach in two-dimensional heavy-tailed distributions

From the Central Limit Theorem to Heavy-Tailed Distributions

Some informational inequalities involving generalized trigonometric functions and a new class of generalized moments

Probability Inequalities for Weighted Distributions

Hybrid Statistics of a Random Model of Zeta over Intervals of Varying Length

Heavy-Tailed Distributions and Their Applications

On asymptotic properties of high moments of compound Poisson distribution

A note on generalization bounds for losses with finite moments

Precise asymptotics of ruin probabilities for a class of multivariate heavy-tailed distributions

Hanson-Wright inequality in Banach spaces

Quantitative bounds for large deviations of heavy tailed random variables

Background risk model in presence of heavy tails under dependence