AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES

Size: px
Start display at page:

Download "AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES"

Transcription

1 Lithuanian Mathematical Journal, Vol. 4, No. 3, 00 AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES V. Bentkus Vilnius Institute of Mathematics and Informatics, Akademijos 4, LT-600 Vilnius, Lithuania ( Abstract. Let M n = X 1 + +X n be a martingale with bounded differences X m = M m M m 1 such that P{ X m σ m }=1 with some nonnegative σ m. Write σ = σ σ n. We prove the inequalities P{M n x} c(1 (x/σ, P{M n >x} 1 c(1 ( x/σ with a constant c 1/(1 ( 3 5. The result yields sharp inequalities in some models related to the measure concentration phenomena. Keywords: probabilities of large deviations, martingale, bounds for tail probabilities, inequalities, bounded differences and random variables, measure concentration phenomena, product spaces, Lipschitz functions, Hoeffding s inequalities, Azuma s inequality. 1. INTRODUCTION AND RESULTS Let M n be a martingale (we define M 0 = 0 with bounded differences X j = M j M j 1 of sizes σ j such that P{ X j σ j }=1forallj. Hence, M n = X 1 + +X n. Write σ = σ1 + +σ n. Let (x = x be the standard normal distribution function. Write ϕ(t dt, ϕ(t = 1 { exp t }, π D(x = min { exp { x /}, c(1 (x }, where c is an absolute constant such that c 1/(1 ( 3 5. Our result is the following upper and lower bounds for the tail probabilities of M n. THEOREM 1.1. For x R, we have P{M n x} D(x/σ and P{M n >x} 1 D( x/σ. (1.1 Research supported by Max Planck Institute for Mathematics, Bonn. Translated from Lietuvos Matematikos Rinkinys, Vol. 4, No. 3, pp , July September, 00. submitted April 0, 001. Original article /0/ $ Plenum Publishing Corporation 55

2 56 V. Bentkus The upper bound 5 for c in (1.1 is not optimal and can be improved. Extending the methods of the author [], one can show that sup sup P{M n x} (1 (x/σ, n M n where sup is taken over all Bernoulli type martingales M n such that σ 1 = =σ n and all n (we call a martingale of Bernoulli type if the differences conditionally have two point distributions. Hence, the constant c in (1.1 must satisfy c. In the case of i.i.d. Rademacher random variables X 1,...,X n of size 1/ n such that P{X 1 = 1/ n}= P{X 1 = 1/ n}=1/ andσ = 1, we obviously have sup P{M n x} 1 (x, n which, for larger x, differs by the factor c 5 compared to (1.1. Theorem 1.1 improves the inequality of Hoeffding [4]. In our setting, his inequality (Theorem of Hoeffding [4] reads as P{M n x} exp { x /(σ }, x 0 (1. For larger x c 1 σ (with some absolute constant c 1, the bound (1.1 is better, since exp { x /(σ } c(1 (x/σ as x. For independent X 1,...,X n, the inequality of Pinelis [9] improves (1. (see (1.3 below. The inequalities of type (1. (and, hence, of (1.1 as well have extensive applications in combinatorics, operational research, computer science, random graphs (see McDiarmid [6]. The result applies to some simple models in the measure concentration phenomena for Lipschitz functions on product spaces (see Section 3, as well as for some nonlinear statistics. In these models, the bounds implied are the sharpest among the known ones. The space for improvements is restricted (cf. the discussion above. For statistical applications, optimal bounds for finite (that is, fixed n are of interest. In this sense, our result is not optimal and can be improved extending the methods of the author []. Theorem 1.1 shows that the martingale-type dependence does not influence much the bounds for tail probabilities compared to the independent and i.i.d. cases. Theorem 1.1 is insufficiently general. Extensions to more general settings are desirable for applications in measure concentration phenomena (see Section 3. This is even more desirable for applications in statistics (see Bentkus and Van Zuijlen [3], where bounds must be explicit and must depend explicitly on parameters of interest. The proof of Theorem 1.1 is based on induction in n and multiply applications of the Chebyshev inequality. We understand Chebyshev s inequality as f g if f g, and we always apply it with a linear function g. Our methods are well designed for applications where we have martingale-type dependence. Seemingly, for the first time inductional proofs without Fourier transform in the context of probabilities of large deviations were used by the author [1]. We hope that the method is robust against generalizations. However, one has not to underestimate the complexity of technicalities that can occur by extensions of Theorem 1.1. Usually, bounds for tail probabilities are proved using exponential functions (which can be interpreted as a kind of the Fourier Laplace transform together with Chebyshev s inequality, which most probably leads to losses in the accuracy. We do not use exponential functions. For independent X 1,...,X n such that σ 1 = = σ n, a bound P{M n x} B n (x with some function B n (x essentially smaller than c(1 (x, was obtained earlier by the author []. One can show that the bound B n (x is sharp on martingales for integer x and all n. Heuristically, the basic ideas and methods are already contained in this paper. In the context of estimates of the convergence rate in the Central Limit Theorem in general spaces, inductional proofs is an extensively used tool (see, for example, a book of Paulauskas and Račkauskas [7].

3 An inequality for tail probabilities 57 The history of inequalities for tail probabilities is very rich (see, for example, books Petrov [8] and Shorack and Wellner [10]. The names of Chernoff, Bennett, Prokhorov, Hoeffding and other come to mind. It seems that the most precise inequalities for sums of bounded random variables were obtained by Hoeffding [4] almost 40 years ago. We are not aware of improvements of (1. for martingales. For independent X 1,...,X n, Pinelis [9] has proved that with the constant c 0 = e 3 / The bound (1.1 yields P{ M n x} c 0 (1 (x/σ (1.3 P{ M n x} c(1 (x/σ. (1.4 The constant in (1.3 is better than that in (1.4. However, the value of c in (1.1 and (1.4 can be improved. Our methods are completely different from those Hoeffding [4] and Pinelis [9].. THE PROOF We write I(x= ϕ(t dt, B(x = ci(x for x R. x By the definition of c, wehaveci( 3 = 1. Hence, B( 3 = 1andB(x > 1forx 3, and the function B is strictly decreasing. Proof of Theorem 1. The lower bound in (1.1 follows from the upper one, just replacing M n by M n and considering complementary events. Hence, it suffices to prove only the upper bound. We apply induction in n. Without loss of generality we assume that σ = 1. Throughout we write 1 = τ +, where τ = σ 1 and = σ + +σ n. Due to the assumption σ = 1, it suffices to prove that since, by the Hoeffding inequality, P{M n x} exp{ x /} for all x 0. P{M n x} B(x, (.1 The case n = 1. In this case M 1 = X 1, and we have to prove that the upper bound in (.1 holds for all possible martingales M 1,thatis,thatP{X 1 x} B(x for arbitrary random variables X 1 such that EX 1 = 0 and P{ X 1 1} =1. If x 1, then B(x > 1, and we trivially have that P{X 1 x} 1 <B(x. If x>1, then P{X 1 x} =0 <B(x,since X 1 1, which concludes the proof for n = 1. The case n>1. By the induction assumption we have P{Z n 1 x} B(x/β for all x R (. and for any martingale sequence Z 0 = 0,Z 1,...,Z n 1 such that P{ Z k Z k 1 β k }=1andβ = β1 + + βn 1. We can assume throughout that 0 <τ<1, 0 <<1, x> 3. Indeed, in the cases where one of τ or is zero, the martingale M n contains at most n 1 nonzero differences, and the result holds by the inductional assumption. If x 3, then B(x 1, and we trivially have P{M n x} 1 B(x. Using (. and conditioning on X 1,wehave X1 P{M n x} =EP{X + +X n x X 1 X 1 } EB, (.3

4 58 V. Bentkus since, for given X 1, the sequence Z 0 = 0,Z 1 = X,...,Z n 1 = X + +X n is a martingale sequence with differences such that P{ Z k Z k 1 σ k+1 }=1. To simplify the notation, write t ϱ(t = ci, EB X1 = Eϱ(X 1. Note that ϱ depends on n and x, which is not reflected in the notation. We have to prove that Eϱ(X 1 B(x. The function ϱ is convex in the interval (, 3], since we assume that x 3 and the function z I(z is convex for z 0. The random variable X 1 satisfies the inequality P{ X 1 τ} =1. The function ϱ is convex in the interval [ τ,τ], since τ 1andϱ is convex in (, 3]. Therefore, the linear function u(t def = τ t τ ϱ( τ+ t + τ τ ϱ(τ satisfies u( τ = ϱ( τ and u(τ = ϱ(τ and is an upper bound for ϱ(t, ϱ(t u(t for all t τ. (.4 Using (.4 and EX 1 = 0, we obtain Eϱ(X 1 Eu(X 1 = c + τ I + c τ I. (.5 Inequality (.5 reduces the proof of (.1 to checking that Let us prove (.6. We have w(τ def = I + τ + I τ + τ xτ + 1 τ w (τ = ϕ 3 ϕ I(x 0. (.6 xτ 1 3. (.7 A bit later we shall prove that w (τ 0for0 τ 1. Therefore, w(τ is a decreasing function of τ 0, and the proof of (.6 reduces to checking that w(0 0. But w(0 = 0, which concludes the proof of (.6. Let us prove that w (τ 0for0 τ 1. Using (.7 and the fact that ϕ(t = (π 1/ exp{ t /}, the inequality w (τ 0 is equivalent to Hence, we have to check (.8. We have v(τ def = xτ (xτ 1 exp {xa} 0, A = A(τ def = τ. (.8 v (τ = x + x exp {xa}+x(xτ 1A exp {xa}, v (0 = 0, and with ( v (τ = x exp {xa} xa + (xτ 1A + x(xτ 1A (.9 A = 1 + τ 4, A = (3τ + τ 3 6. (.10 Below we shall show that v (τ 0. This means that τ v(τ is a convex function such that v(0 = 0. Therefore, for τ 0, this function can assume only nonnegative values, which proves (.8.

5 An inequality for tail probabilities 59 It remains to show that v (τ 0. Due to (.9 and (.10, this is equivalent to the inequality x(1 + τ 4 + (xτ 1(3τ + τ x(xτ 1(1 + τ 8 0. (.11 Multiplying by 8,using = 1 τ, and collecting similar terms, we see that (.11 is equivalent to with Dividing by τ, inequality (.1 is equivalent to ax + bx + d 0 (.1 a = τ(1 + τ, b = 4τ 4, d = τ(3 τ τ 4. y(x def = (1 + τ x 4τ 3 x 3 + τ + τ 4 0. (.13 The function x y(x is a quadratic increasing function of x x τ,where It is easy to verify that x τ = τ 3 (1 + τ. x τ 1 for all 0 τ 1. Hence, the function x y(x from (.13 is an increasing function of x 1. Therefore, for x 3, inequality (.13 is implied by the stronger inequality y( 3 0, that is, by the inequality y( 3 = 8τ + 4τ 4 4 3τ 3 0 for 0 τ 1. (.14 The verification of (.14 is elementary. This proves (.11 and the theorem. 3. SOME APPLICATIONS TO THE MEASURE CONCENTRATION To simplify the notation, instead of a product of arbitrary metric (or measurable spaces, we consider only the case of the n-dimensional unit cube [ 1, 1] n. Let Z 1,...,Z n be independent random variables such that P{ Z j 1} =1. Let a function f :[ 1, 1] n R satisfy the separate Lipschitz condition in each of the variables, that is, f(x1,...,x j 1,x,x j+1,...,x n f(x 1,...,x j 1,y,x j+1,...,x n σj x y. (3.1 Recall that σ = σ 1 + +σ n. COROLLARY 3.1. Let f satisfy the Lipschitz condition (3.1. Then, for x 0, we have P{f Ef + x} D, P{f Ef x} D, (3. σ σ where f = f(z 1,...,Z n. If EZ j = 0 for all j, then the bound (3. improves to P{f Ef + x} D, P{f Ef x} D. (3.3 σ σ

6 60 V. Bentkus Inequalities (3.3 are rather precise already in the case of the linear function f(x= σ 1 x 1 + +σ n x n, (see Introduction. The Lipschitz condition is just a device to control the sizes of the differences X j = E j+1,...,n f(z 1,...,Z n E j,...,n f(z 1,...,Z n, where E j,...,n f(z 1,...,Z n = E ( f(z 1,...,Z n Z 1,...,Z j 1 (3.4 stands for the conditional expectation given Z 1,...,Z j 1. The bound (3.3 holds in the case of a product of arbitrary measurable spaces and arbitrary functions f such that the differences X j from (3.4 have the sizes σ j, that is, P{ X j σ j }=1. Proof of Corollary 3.1. Note that f Ef = X 1 + +X n is a martingale sequence with the differences X j. In order to apply Theorem 1.1, we have to estimate the sizes of X j. Let Y j be an independent copy of Z j, independent from other random variables as well. Let E be the conditional expectation given Z 1,...,Z j. Then, applying the Lipschitz condition (3.1, we have X j E f(...,zj 1,Z j,z j+1,... f(...,z j 1,Y j,z j+1,... σ j E Z j Y j. (3.5 To prove (3. we use the trivial estimate E Z j Y j. Then (3.5 yields X j σ j,andtheorem1.1 implies (3.. To prove (3.3 we note that E x Y 1 for all 1 x 1 (3.6 if a random variable Y satisfies EY = 0andP{ Y 1} =1. Using (3.6, estimate (3.5 yields X j σ j,and an application of Theorem 1.1 provides (3.3. Let us prove (3.6. It is clear that u(t def = x t v(t def 1 + x (1 t 1 x (1 + t = + (3.7 for x 1and t 1. Indeed, the function t u(t is a convex function of t. The function t v(t is a linear function of t such that u( 1 = v( 1 and u(1 = v(1. Hence, (3.7 holds. Replacing t by Y, taking the expectation in (3.7 and using EY = 0, we obtain E x Y 1 + x + 1 x 1 for x 1, which proves (3.6. The inequalities of types (3. and (3.3 have extensive applications in combinatorics and random graphs (see McDiarmid [6]. The known inequalities have the form P{f Ef + x} exp{ x /(σ }, which, for larger x, is worse than (3.3. The models we considered are rather simple ones in the measure concentration. The bounds (3. and (3.3 seem to be the sharpest among the known ones in these models. See Talagrand [11] and Ledoux [5] for a nice introduction into the topic.

7 An inequality for tail probabilities 61 REFERENCES 1. V. Bentkus, Large deviations in Banach spaces, Probab. Theory Appl., 31(4, ( V. Bentkus, An inequality for large deviation probabilities of sums of bounded i.i.d.r.v., Lith. Math. J., 41(, ( V. Bentkus and M. van Zuijlen, Upper confidence bounds for mean, Ann. Statist. (to appear. 4. W. Hoeffding, Probability inequalities for sums of bounded random variables, J. Amer. Statist. Assoc., 58, ( M. Ledoux, Concentration of measure and logarithmic Sobolev inequalities, in: Séminaire de Probabilités, XXXIII, Springer, Berlin (1999, pp C. McDiarmid, On the method of bounded differences, London Math. Soc. Lecture Note Ser., 141, ( V. Paulauskas and A. Račkauskas, Approximation Theory in the Central Limit Theorem. Exact Results in Banach Spaces, Kluwer, Dordrecht ( V. V. Petrov, Sums of Independent Random Variables, Springer, New York ( I. Pinelis, Extremal probabilistic problems and Hotelling s T test under a symmetry assumption, Ann. Statist., (4, ( G. R. Shorack and J. A. Wellner, Empirical Processes with Applications to Statistics, Wiley, New York ( M. Talagrand, Concentration of measure and isoperimetric inequalities in product spaces, Inst. Hautes Études Sci. Publ. Math., 81, (1995. (Translated by V. Bentkus

Entropy and Ergodic Theory Lecture 15: A first look at concentration

Entropy and Ergodic Theory Lecture 15: A first look at concentration Entropy and Ergodic Theory Lecture 15: A first look at concentration 1 Introduction to concentration Let X 1, X 2,... be i.i.d. R-valued RVs with common distribution µ, and suppose for simplicity that

More information

March 1, Florida State University. Concentration Inequalities: Martingale. Approach and Entropy Method. Lizhe Sun and Boning Yang.

March 1, Florida State University. Concentration Inequalities: Martingale. Approach and Entropy Method. Lizhe Sun and Boning Yang. Florida State University March 1, 2018 Framework 1. (Lizhe) Basic inequalities Chernoff bounding Review for STA 6448 2. (Lizhe) Discrete-time martingales inequalities via martingale approach 3. (Boning)

More information

Concentration inequalities and the entropy method

Concentration inequalities and the entropy method Concentration inequalities and the entropy method Gábor Lugosi ICREA and Pompeu Fabra University Barcelona what is concentration? We are interested in bounding random fluctuations of functions of many

More information

Lithuanian Mathematical Journal, 2006, No 1

Lithuanian Mathematical Journal, 2006, No 1 ON DOMINATION OF TAIL PROBABILITIES OF (SUPER)MARTINGALES: EXPLICIT BOUNDS V. Bentkus, 1,3 N. Kalosha, 2,3 M. van Zuijlen 2,3 Lithuanian Mathematical Journal, 2006, No 1 Abstract. Let X be a random variable

More information

Bounds for the stop loss premium for unbounded risks under the variance constraints

Bounds for the stop loss premium for unbounded risks under the variance constraints Bounds for the stop loss premium for unbounded risks under the variance constraints V. Bentkus 1,2 Abstract Let f be a function with convex derivative f. We consider the maximization problem max Ef(X)

More information

Constructive bounds for a Ramsey-type problem

Constructive bounds for a Ramsey-type problem Constructive bounds for a Ramsey-type problem Noga Alon Michael Krivelevich Abstract For every fixed integers r, s satisfying r < s there exists some ɛ = ɛ(r, s > 0 for which we construct explicitly an

More information

ON CONSERVATIVE CONFIDENCE INTERVALS

ON CONSERVATIVE CONFIDENCE INTERVALS Lithuanian Mathematical Journal, Vol. 43, No. 2, 2003 ON CONSERVATIVE CONFIDENCE INTERVALS V. Bentkus 1,2,3 Institute of Mathematics and Informatics, Akademijos 4, LT-2021 Vilnius, Lithuania (e-mail: bentkus@ktl.mii.lt)

More information

Spectral Gap and Concentration for Some Spherically Symmetric Probability Measures

Spectral Gap and Concentration for Some Spherically Symmetric Probability Measures Spectral Gap and Concentration for Some Spherically Symmetric Probability Measures S.G. Bobkov School of Mathematics, University of Minnesota, 127 Vincent Hall, 26 Church St. S.E., Minneapolis, MN 55455,

More information

Uniform concentration inequalities, martingales, Rademacher complexity and symmetrization

Uniform concentration inequalities, martingales, Rademacher complexity and symmetrization Uniform concentration inequalities, martingales, Rademacher complexity and symmetrization John Duchi Outline I Motivation 1 Uniform laws of large numbers 2 Loss minimization and data dependence II Uniform

More information

Outline. Martingales. Piotr Wojciechowski 1. 1 Lane Department of Computer Science and Electrical Engineering West Virginia University.

Outline. Martingales. Piotr Wojciechowski 1. 1 Lane Department of Computer Science and Electrical Engineering West Virginia University. Outline Piotr 1 1 Lane Department of Computer Science and Electrical Engineering West Virginia University 8 April, 01 Outline Outline 1 Tail Inequalities Outline Outline 1 Tail Inequalities General Outline

More information

Inequalities for Sums of Random Variables: a combinatorial perspective

Inequalities for Sums of Random Variables: a combinatorial perspective Inequalities for Sums of Random Variables: a combinatorial perspective Doctoral thesis Matas Šileikis Department of Mathematics and Computer Science, Adam Mickiewicz University, Poznań April 2012 Acknowledgements

More information

An estimate for the probability of dependent events

An estimate for the probability of dependent events Statistics and Probability Letters 78 (2008) 2839 2843 Contents lists available at ScienceDirect Statistics and Probability Letters journal homepage: www.elsevier.com/locate/stapro An estimate for the

More information

Lectures 6: Degree Distributions and Concentration Inequalities

Lectures 6: Degree Distributions and Concentration Inequalities University of Washington Lecturer: Abraham Flaxman/Vahab S Mirrokni CSE599m: Algorithms and Economics of Networks April 13 th, 007 Scribe: Ben Birnbaum Lectures 6: Degree Distributions and Concentration

More information

On probabilities of large and moderate deviations for L-statistics: a survey of some recent developments

On probabilities of large and moderate deviations for L-statistics: a survey of some recent developments UDC 519.2 On probabilities of large and moderate deviations for L-statistics: a survey of some recent developments N. V. Gribkova Department of Probability Theory and Mathematical Statistics, St.-Petersburg

More information

The symmetry in the martingale inequality

The symmetry in the martingale inequality Statistics & Probability Letters 56 2002 83 9 The symmetry in the martingale inequality Sungchul Lee a; ;, Zhonggen Su a;b;2 a Department of Mathematics, Yonsei University, Seoul 20-749, South Korea b

More information

On the Concentration of the Crest Factor for OFDM Signals

On the Concentration of the Crest Factor for OFDM Signals On the Concentration of the Crest Factor for OFDM Signals Igal Sason Department of Electrical Engineering Technion - Israel Institute of Technology, Haifa 3, Israel E-mail: sason@eetechnionacil Abstract

More information

OXPORD UNIVERSITY PRESS

OXPORD UNIVERSITY PRESS Concentration Inequalities A Nonasymptotic Theory of Independence STEPHANE BOUCHERON GABOR LUGOSI PASCAL MASS ART OXPORD UNIVERSITY PRESS CONTENTS 1 Introduction 1 1.1 Sums of Independent Random Variables

More information

A note on the convex infimum convolution inequality

A note on the convex infimum convolution inequality A note on the convex infimum convolution inequality Naomi Feldheim, Arnaud Marsiglietti, Piotr Nayar, Jing Wang Abstract We characterize the symmetric measures which satisfy the one dimensional convex

More information

TIGHT BERNOULLI TAIL PROBABILITY BOUNDS

TIGHT BERNOULLI TAIL PROBABILITY BOUNDS Vilnius University Dainius Dzindzalieta TIGHT BERNOULLI TAIL PROBABILITY BOUNDS Doctoral Dissertation Physical Sciences, Mathematics (01 P) Vilnius, 2014 Doctoral dissertation was prepared at the Institute

More information

Notes 6 : First and second moment methods

Notes 6 : First and second moment methods Notes 6 : First and second moment methods Math 733-734: Theory of Probability Lecturer: Sebastien Roch References: [Roc, Sections 2.1-2.3]. Recall: THM 6.1 (Markov s inequality) Let X be a non-negative

More information

Min-Wise independent linear permutations

Min-Wise independent linear permutations Min-Wise independent linear permutations Tom Bohman Department of Mathematical Sciences, Carnegie Mellon University, Pittsburgh PA523, USA E-mail:tbohman@andrewcmuedu Colin Cooper School of Mathematical

More information

On an uniqueness theorem for characteristic functions

On an uniqueness theorem for characteristic functions ISSN 392-53 Nonlinear Analysis: Modelling and Control, 207, Vol. 22, No. 3, 42 420 https://doi.org/0.5388/na.207.3.9 On an uniqueness theorem for characteristic functions Saulius Norvidas Institute of

More information

Hoeffding s inequality for supermartingales

Hoeffding s inequality for supermartingales Hoeffding s inequality for supermartingales Xiequan Fan, Ion Grama, Quansheng Liu To cite this version: Xiequan Fan, Ion Grama, Quansheng Liu. Hoeffding s inequality for supermartingales. Stochastic Processes

More information

STAT 200C: High-dimensional Statistics

STAT 200C: High-dimensional Statistics STAT 200C: High-dimensional Statistics Arash A. Amini May 30, 2018 1 / 59 Classical case: n d. Asymptotic assumption: d is fixed and n. Basic tools: LLN and CLT. High-dimensional setting: n d, e.g. n/d

More information

First Passage Percolation Has Sublinear Distance Variance

First Passage Percolation Has Sublinear Distance Variance arxiv:math/23262v5 [math.pr 25 May 27 First Passage Percolation Has Sublinear Distance Variance Itai Benjamini Gil Kalai Oded Schramm October 21, 22 Abstract Let < a < b

More information

Statistical Machine Learning

Statistical Machine Learning Statistical Machine Learning UoC Stats 37700, Winter quarter Lecture 2: Introduction to statistical learning theory. 1 / 22 Goals of statistical learning theory SLT aims at studying the performance of

More information

On the Bennett-Hoeffding inequality

On the Bennett-Hoeffding inequality On the Bennett-Hoeffding inequality of Iosif 1,2,3 1 Department of Mathematical Sciences Michigan Technological University 2 Supported by NSF grant DMS-0805946 3 Paper available at http://arxiv.org/abs/0902.4058

More information

Bennett-type Generalization Bounds: Large-deviation Case and Faster Rate of Convergence

Bennett-type Generalization Bounds: Large-deviation Case and Faster Rate of Convergence Bennett-type Generalization Bounds: Large-deviation Case and Faster Rate of Convergence Chao Zhang The Biodesign Institute Arizona State University Tempe, AZ 8587, USA Abstract In this paper, we present

More information

Lecture 4: Inequalities and Asymptotic Estimates

Lecture 4: Inequalities and Asymptotic Estimates CSE 713: Random Graphs and Applications SUNY at Buffalo, Fall 003 Lecturer: Hung Q. Ngo Scribe: Hung Q. Ngo Lecture 4: Inequalities and Asymptotic Estimates We draw materials from [, 5, 8 10, 17, 18].

More information

On the lines passing through two conjugates of a Salem number

On the lines passing through two conjugates of a Salem number Under consideration for publication in Math. Proc. Camb. Phil. Soc. 1 On the lines passing through two conjugates of a Salem number By ARTŪRAS DUBICKAS Department of Mathematics and Informatics, Vilnius

More information

ANALYSIS OF THE LAW OF THE ITERATED LOGARITHM FOR THE IDLE TIME OF A CUSTOMER IN MULTIPHASE QUEUES

ANALYSIS OF THE LAW OF THE ITERATED LOGARITHM FOR THE IDLE TIME OF A CUSTOMER IN MULTIPHASE QUEUES International Journal of Pure and Applied Mathematics Volume 66 No. 2 2011, 183-190 ANALYSIS OF THE LAW OF THE ITERATED LOGARITHM FOR THE IDLE TIME OF A CUSTOMER IN MULTIPHASE QUEUES Saulius Minkevičius

More information

Concentration Inequalities

Concentration Inequalities Chapter Concentration Inequalities I. Moment generating functions, the Chernoff method, and sub-gaussian and sub-exponential random variables a. Goal for this section: given a random variable X, how does

More information

A Criterion for the Compound Poisson Distribution to be Maximum Entropy

A Criterion for the Compound Poisson Distribution to be Maximum Entropy A Criterion for the Compound Poisson Distribution to be Maximum Entropy Oliver Johnson Department of Mathematics University of Bristol University Walk Bristol, BS8 1TW, UK. Email: O.Johnson@bristol.ac.uk

More information

Concentration function and other stuff

Concentration function and other stuff Concentration function and other stuff Sabrina Sixta Tuesday, June 16, 2014 Sabrina Sixta () Concentration function and other stuff Tuesday, June 16, 2014 1 / 13 Table of Contents Outline 1 Chernoff Bound

More information

Lecture 4: Applications: random trees, determinantal measures and sampling

Lecture 4: Applications: random trees, determinantal measures and sampling Lecture 4: Applications: random trees, determinantal measures and sampling Robin Pemantle University of Pennsylvania pemantle@math.upenn.edu Minerva Lectures at Columbia University 09 November, 2016 Sampling

More information

Inverse Brascamp-Lieb inequalities along the Heat equation

Inverse Brascamp-Lieb inequalities along the Heat equation Inverse Brascamp-Lieb inequalities along the Heat equation Franck Barthe and Dario Cordero-Erausquin October 8, 003 Abstract Adapting Borell s proof of Ehrhard s inequality for general sets, we provide

More information

MAJORIZING MEASURES WITHOUT MEASURES. By Michel Talagrand URA 754 AU CNRS

MAJORIZING MEASURES WITHOUT MEASURES. By Michel Talagrand URA 754 AU CNRS The Annals of Probability 2001, Vol. 29, No. 1, 411 417 MAJORIZING MEASURES WITHOUT MEASURES By Michel Talagrand URA 754 AU CNRS We give a reformulation of majorizing measures that does not involve measures,

More information

On the martingales obtained by an extension due to Saisho, Tanemura and Yor of Pitman s theorem

On the martingales obtained by an extension due to Saisho, Tanemura and Yor of Pitman s theorem On the martingales obtained by an extension due to Saisho, Tanemura and Yor of Pitman s theorem Koichiro TAKAOKA Dept of Applied Physics, Tokyo Institute of Technology Abstract M Yor constructed a family

More information

THE LINDEBERG-FELLER CENTRAL LIMIT THEOREM VIA ZERO BIAS TRANSFORMATION

THE LINDEBERG-FELLER CENTRAL LIMIT THEOREM VIA ZERO BIAS TRANSFORMATION THE LINDEBERG-FELLER CENTRAL LIMIT THEOREM VIA ZERO BIAS TRANSFORMATION JAINUL VAGHASIA Contents. Introduction. Notations 3. Background in Probability Theory 3.. Expectation and Variance 3.. Convergence

More information

FIXED POINT THEOREMS OF KRASNOSELSKII TYPE IN A SPACE OF CONTINUOUS FUNCTIONS

FIXED POINT THEOREMS OF KRASNOSELSKII TYPE IN A SPACE OF CONTINUOUS FUNCTIONS Fixed Point Theory, Volume 5, No. 2, 24, 181-195 http://www.math.ubbcluj.ro/ nodeacj/sfptcj.htm FIXED POINT THEOREMS OF KRASNOSELSKII TYPE IN A SPACE OF CONTINUOUS FUNCTIONS CEZAR AVRAMESCU 1 AND CRISTIAN

More information

Random convolution of O-exponential distributions

Random convolution of O-exponential distributions Nonlinear Analysis: Modelling and Control, Vol. 20, No. 3, 447 454 ISSN 1392-5113 http://dx.doi.org/10.15388/na.2015.3.9 Random convolution of O-exponential distributions Svetlana Danilenko a, Jonas Šiaulys

More information

Applied Analysis (APPM 5440): Final exam 1:30pm 4:00pm, Dec. 14, Closed books.

Applied Analysis (APPM 5440): Final exam 1:30pm 4:00pm, Dec. 14, Closed books. Applied Analysis APPM 44: Final exam 1:3pm 4:pm, Dec. 14, 29. Closed books. Problem 1: 2p Set I = [, 1]. Prove that there is a continuous function u on I such that 1 ux 1 x sin ut 2 dt = cosx, x I. Define

More information

ITERATIVE APPROXIMATION OF SOLUTIONS OF GENERALIZED EQUATIONS OF HAMMERSTEIN TYPE

ITERATIVE APPROXIMATION OF SOLUTIONS OF GENERALIZED EQUATIONS OF HAMMERSTEIN TYPE Fixed Point Theory, 15(014), No., 47-440 http://www.math.ubbcluj.ro/ nodeacj/sfptcj.html ITERATIVE APPROXIMATION OF SOLUTIONS OF GENERALIZED EQUATIONS OF HAMMERSTEIN TYPE C.E. CHIDUME AND Y. SHEHU Mathematics

More information

Expansion and Isoperimetric Constants for Product Graphs

Expansion and Isoperimetric Constants for Product Graphs Expansion and Isoperimetric Constants for Product Graphs C. Houdré and T. Stoyanov May 4, 2004 Abstract Vertex and edge isoperimetric constants of graphs are studied. Using a functional-analytic approach,

More information

arxiv: v2 [math.st] 20 Feb 2013

arxiv: v2 [math.st] 20 Feb 2013 Exact bounds on the closeness between the Student and standard normal distributions arxiv:1101.3328v2 [math.st] 20 Feb 2013 Contents Iosif Pinelis Department of Mathematical Sciences Michigan Technological

More information

arxiv: v1 [math.pr] 12 Jun 2012

arxiv: v1 [math.pr] 12 Jun 2012 arxiv: math.pr/ The missing factor in Bennett s inequality arxiv:6.59v [math.pr] Jun Xiequan Fan Ion Grama and Quansheng Liu Université de Bretagne-Sud, LMBA, UMR CNRS 65, Campus de Tohannic, 567 Vannes,

More information

THE SKOROKHOD OBLIQUE REFLECTION PROBLEM IN A CONVEX POLYHEDRON

THE SKOROKHOD OBLIQUE REFLECTION PROBLEM IN A CONVEX POLYHEDRON GEORGIAN MATHEMATICAL JOURNAL: Vol. 3, No. 2, 1996, 153-176 THE SKOROKHOD OBLIQUE REFLECTION PROBLEM IN A CONVEX POLYHEDRON M. SHASHIASHVILI Abstract. The Skorokhod oblique reflection problem is studied

More information

Viscosity Iterative Approximating the Common Fixed Points of Non-expansive Semigroups in Banach Spaces

Viscosity Iterative Approximating the Common Fixed Points of Non-expansive Semigroups in Banach Spaces Viscosity Iterative Approximating the Common Fixed Points of Non-expansive Semigroups in Banach Spaces YUAN-HENG WANG Zhejiang Normal University Department of Mathematics Yingbing Road 688, 321004 Jinhua

More information

Weak max-sum equivalence for dependent heavy-tailed random variables

Weak max-sum equivalence for dependent heavy-tailed random variables DOI 10.1007/s10986-016-9303-6 Lithuanian Mathematical Journal, Vol. 56, No. 1, January, 2016, pp. 49 59 Wea max-sum equivalence for dependent heavy-tailed random variables Lina Dindienė a and Remigijus

More information

Concentration of Measures by Bounded Size Bias Couplings

Concentration of Measures by Bounded Size Bias Couplings Concentration of Measures by Bounded Size Bias Couplings Subhankar Ghosh, Larry Goldstein University of Southern California [arxiv:0906.3886] January 10 th, 2013 Concentration of Measure Distributional

More information

Concentration, self-bounding functions

Concentration, self-bounding functions Concentration, self-bounding functions S. Boucheron 1 and G. Lugosi 2 and P. Massart 3 1 Laboratoire de Probabilités et Modèles Aléatoires Université Paris-Diderot 2 Economics University Pompeu Fabra 3

More information

Lecture 5: Probabilistic tools and Applications II

Lecture 5: Probabilistic tools and Applications II T-79.7003: Graphs and Networks Fall 2013 Lecture 5: Probabilistic tools and Applications II Lecturer: Charalampos E. Tsourakakis Oct. 11, 2013 5.1 Overview In the first part of today s lecture we will

More information

ON THE CONVEX INFIMUM CONVOLUTION INEQUALITY WITH OPTIMAL COST FUNCTION

ON THE CONVEX INFIMUM CONVOLUTION INEQUALITY WITH OPTIMAL COST FUNCTION ON THE CONVEX INFIMUM CONVOLUTION INEQUALITY WITH OPTIMAL COST FUNCTION MARTA STRZELECKA, MICHA L STRZELECKI, AND TOMASZ TKOCZ Abstract. We show that every symmetric random variable with log-concave tails

More information

Subelliptic mollifiers and a basic pointwise estimate of Poincaré type

Subelliptic mollifiers and a basic pointwise estimate of Poincaré type Math. Z. 226, 147 154 (1997) c Springer-Verlag 1997 Subelliptic mollifiers and a basic pointwise estimate of Poincaré type Luca Capogna, Donatella Danielli, Nicola Garofalo Department of Mathematics, Purdue

More information

Viscosity approximation method for m-accretive mapping and variational inequality in Banach space

Viscosity approximation method for m-accretive mapping and variational inequality in Banach space An. Şt. Univ. Ovidius Constanţa Vol. 17(1), 2009, 91 104 Viscosity approximation method for m-accretive mapping and variational inequality in Banach space Zhenhua He 1, Deifei Zhang 1, Feng Gu 2 Abstract

More information

APPROXIMATING SOLUTIONS FOR THE SYSTEM OF REFLEXIVE BANACH SPACE

APPROXIMATING SOLUTIONS FOR THE SYSTEM OF REFLEXIVE BANACH SPACE Bulletin of Mathematical Analysis and Applications ISSN: 1821-1291, URL: http://www.bmathaa.org Volume 2 Issue 3(2010), Pages 32-39. APPROXIMATING SOLUTIONS FOR THE SYSTEM OF φ-strongly ACCRETIVE OPERATOR

More information

CSE 525 Randomized Algorithms & Probabilistic Analysis Spring Lecture 3: April 9

CSE 525 Randomized Algorithms & Probabilistic Analysis Spring Lecture 3: April 9 CSE 55 Randomized Algorithms & Probabilistic Analysis Spring 01 Lecture : April 9 Lecturer: Anna Karlin Scribe: Tyler Rigsby & John MacKinnon.1 Kinds of randomization in algorithms So far in our discussion

More information

Lecture 4: September Reminder: convergence of sequences

Lecture 4: September Reminder: convergence of sequences 36-705: Intermediate Statistics Fall 2017 Lecturer: Siva Balakrishnan Lecture 4: September 6 In this lecture we discuss the convergence of random variables. At a high-level, our first few lectures focused

More information

A New Estimator for a Tail Index

A New Estimator for a Tail Index Acta Applicandae Mathematicae 00: 3, 2003. 2003 Kluwer Academic Publishers. Printed in the Netherlands. A New Estimator for a Tail Index V. PAULAUSKAS Department of Mathematics and Informatics, Vilnius

More information

Renormings of c 0 and the minimal displacement problem

Renormings of c 0 and the minimal displacement problem doi: 0.55/umcsmath-205-0008 ANNALES UNIVERSITATIS MARIAE CURIE-SKŁODOWSKA LUBLIN POLONIA VOL. LXVIII, NO. 2, 204 SECTIO A 85 9 ŁUKASZ PIASECKI Renormings of c 0 and the minimal displacement problem Abstract.

More information

FKN Theorem on the biased cube

FKN Theorem on the biased cube FKN Theorem on the biased cube Piotr Nayar Abstract In this note we consider Boolean functions defined on the discrete cube { γ, γ 1 } n equipped with a product probability measure µ n, where µ = βδ γ

More information

2 Statement of the problem and assumptions

2 Statement of the problem and assumptions Mathematical Notes, 25, vol. 78, no. 4, pp. 466 48. Existence Theorem for Optimal Control Problems on an Infinite Time Interval A.V. Dmitruk and N.V. Kuz kina We consider an optimal control problem on

More information

Statistics 300B Winter 2018 Final Exam Due 24 Hours after receiving it

Statistics 300B Winter 2018 Final Exam Due 24 Hours after receiving it Statistics 300B Winter 08 Final Exam Due 4 Hours after receiving it Directions: This test is open book and open internet, but must be done without consulting other students. Any consultation of other students

More information

Notes 1 : Measure-theoretic foundations I

Notes 1 : Measure-theoretic foundations I Notes 1 : Measure-theoretic foundations I Math 733-734: Theory of Probability Lecturer: Sebastien Roch References: [Wil91, Section 1.0-1.8, 2.1-2.3, 3.1-3.11], [Fel68, Sections 7.2, 8.1, 9.6], [Dur10,

More information

Bounds for pairs in partitions of graphs

Bounds for pairs in partitions of graphs Bounds for pairs in partitions of graphs Jie Ma Xingxing Yu School of Mathematics Georgia Institute of Technology Atlanta, GA 30332-0160, USA Abstract In this paper we study the following problem of Bollobás

More information

(2m)-TH MEAN BEHAVIOR OF SOLUTIONS OF STOCHASTIC DIFFERENTIAL EQUATIONS UNDER PARAMETRIC PERTURBATIONS

(2m)-TH MEAN BEHAVIOR OF SOLUTIONS OF STOCHASTIC DIFFERENTIAL EQUATIONS UNDER PARAMETRIC PERTURBATIONS (2m)-TH MEAN BEHAVIOR OF SOLUTIONS OF STOCHASTIC DIFFERENTIAL EQUATIONS UNDER PARAMETRIC PERTURBATIONS Svetlana Janković and Miljana Jovanović Faculty of Science, Department of Mathematics, University

More information

Random Bernstein-Markov factors

Random Bernstein-Markov factors Random Bernstein-Markov factors Igor Pritsker and Koushik Ramachandran October 20, 208 Abstract For a polynomial P n of degree n, Bernstein s inequality states that P n n P n for all L p norms on the unit

More information

Large deviations of empirical processes

Large deviations of empirical processes Large deviations of empirical processes Miguel A. Arcones Abstract. We give necessary and sufficient conditions for the large deviations of empirical processes and of Banach space valued random vectors.

More information

Zeros of lacunary random polynomials

Zeros of lacunary random polynomials Zeros of lacunary random polynomials Igor E. Pritsker Dedicated to Norm Levenberg on his 60th birthday Abstract We study the asymptotic distribution of zeros for the lacunary random polynomials. It is

More information

Cramér large deviation expansions for martingales under Bernstein s condition

Cramér large deviation expansions for martingales under Bernstein s condition Cramér large deviation expansions for martingales under Bernstein s condition Xiequan Fan, Ion Grama, Quansheng Liu To cite this version: Xiequan Fan, Ion Grama, Quansheng Liu. Cramér large deviation expansions

More information

Chapter 2 Vector-matrix Differential Equation and Numerical Inversion of Laplace Transform

Chapter 2 Vector-matrix Differential Equation and Numerical Inversion of Laplace Transform Chapter 2 Vector-matrix Differential Equation and Numerical Inversion of Laplace Transform 2.1 Vector-matrix Differential Equation A differential equation and a set of differential (simultaneous linear

More information

arxiv: v1 [math.pr] 29 Dec 2015

arxiv: v1 [math.pr] 29 Dec 2015 Royen s proof of the Gaussian correlation inequality arxiv:5.8776v [math.pr] 9 Dec 5 Rafa l Lata la and Dariusz Matlak Abstract We present in detail Thomas Royen s proof of the Gaussian correlation inequality

More information

Concentration of Measures by Bounded Couplings

Concentration of Measures by Bounded Couplings Concentration of Measures by Bounded Couplings Subhankar Ghosh, Larry Goldstein and Ümit Işlak University of Southern California [arxiv:0906.3886] [arxiv:1304.5001] May 2013 Concentration of Measure Distributional

More information

Stein s Method: Distributional Approximation and Concentration of Measure

Stein s Method: Distributional Approximation and Concentration of Measure Stein s Method: Distributional Approximation and Concentration of Measure Larry Goldstein University of Southern California 36 th Midwest Probability Colloquium, 2014 Concentration of Measure Distributional

More information

Random fields and central limit theorem in some generalized Hölder spaces

Random fields and central limit theorem in some generalized Hölder spaces 1 Central limit theorems in Hölder spaces Random fields and central limit theorem in some generalized Hölder spaces A. RAČKAUSKAS, CH. SUQUET Department of Mathematics, Vilnius University, Naugarduko 24,

More information

ON THE LIMIT POINTS OF THE FRACTIONAL PARTS OF POWERS OF PISOT NUMBERS

ON THE LIMIT POINTS OF THE FRACTIONAL PARTS OF POWERS OF PISOT NUMBERS ARCHIVUM MATHEMATICUM (BRNO) Tomus 42 (2006), 151 158 ON THE LIMIT POINTS OF THE FRACTIONAL PARTS OF POWERS OF PISOT NUMBERS ARTŪRAS DUBICKAS Abstract. We consider the sequence of fractional parts {ξα

More information

Phenomena in high dimensions in geometric analysis, random matrices, and computational geometry Roscoff, France, June 25-29, 2012

Phenomena in high dimensions in geometric analysis, random matrices, and computational geometry Roscoff, France, June 25-29, 2012 Phenomena in high dimensions in geometric analysis, random matrices, and computational geometry Roscoff, France, June 25-29, 202 BOUNDS AND ASYMPTOTICS FOR FISHER INFORMATION IN THE CENTRAL LIMIT THEOREM

More information

Burkholder s inequality for multiindex martingales

Burkholder s inequality for multiindex martingales Annales Mathematicae et Informaticae 32 (25) pp. 45 51. Burkholder s inequality for multiindex martingales István Fazekas Faculty of Informatics, University of Debrecen e-mail: fazekasi@inf.unideb.hu Dedicated

More information

LAWS OF LARGE NUMBERS AND TAIL INEQUALITIES FOR RANDOM TRIES AND PATRICIA TREES

LAWS OF LARGE NUMBERS AND TAIL INEQUALITIES FOR RANDOM TRIES AND PATRICIA TREES LAWS OF LARGE NUMBERS AND TAIL INEQUALITIES FOR RANDOM TRIES AND PATRICIA TREES Luc Devroye School of Computer Science McGill University Montreal, Canada H3A 2K6 luc@csmcgillca June 25, 2001 Abstract We

More information

The Moment Method; Convex Duality; and Large/Medium/Small Deviations

The Moment Method; Convex Duality; and Large/Medium/Small Deviations Stat 928: Statistical Learning Theory Lecture: 5 The Moment Method; Convex Duality; and Large/Medium/Small Deviations Instructor: Sham Kakade The Exponential Inequality and Convex Duality The exponential

More information

Wald for non-stopping times: The rewards of impatient prophets

Wald for non-stopping times: The rewards of impatient prophets Electron. Commun. Probab. 19 (2014), no. 78, 1 9. DOI: 10.1214/ECP.v19-3609 ISSN: 1083-589X ELECTRONIC COMMUNICATIONS in PROBABILITY Wald for non-stopping times: The rewards of impatient prophets Alexander

More information

Concentration inequalities for non-lipschitz functions

Concentration inequalities for non-lipschitz functions Concentration inequalities for non-lipschitz functions University of Warsaw Berkeley, October 1, 2013 joint work with Radosław Adamczak (University of Warsaw) Gaussian concentration (Sudakov-Tsirelson,

More information

STAT 992 Paper Review: Sure Independence Screening in Generalized Linear Models with NP-Dimensionality J.Fan and R.Song

STAT 992 Paper Review: Sure Independence Screening in Generalized Linear Models with NP-Dimensionality J.Fan and R.Song STAT 992 Paper Review: Sure Independence Screening in Generalized Linear Models with NP-Dimensionality J.Fan and R.Song Presenter: Jiwei Zhao Department of Statistics University of Wisconsin Madison April

More information

On Z p -norms of random vectors

On Z p -norms of random vectors On Z -norms of random vectors Rafa l Lata la Abstract To any n-dimensional random vector X we may associate its L -centroid body Z X and the corresonding norm. We formulate a conjecture concerning the

More information

Hamming Cube and Other Stuff

Hamming Cube and Other Stuff Hamming Cube and Other Stuff Sabrina Sixta Tuesday, May 27, 2014 Sabrina Sixta () Hamming Cube and Other Stuff Tuesday, May 27, 2014 1 / 12 Table of Contents Outline 1 Definition of a Hamming cube 2 Properties

More information

Formulas for probability theory and linear models SF2941

Formulas for probability theory and linear models SF2941 Formulas for probability theory and linear models SF2941 These pages + Appendix 2 of Gut) are permitted as assistance at the exam. 11 maj 2008 Selected formulae of probability Bivariate probability Transforms

More information

EXPLICIT DISTRIBUTIONAL RESULTS IN PATTERN FORMATION. By V. T. Stefanov 1 and A. G. Pakes University of Western Australia

EXPLICIT DISTRIBUTIONAL RESULTS IN PATTERN FORMATION. By V. T. Stefanov 1 and A. G. Pakes University of Western Australia The Annals of Applied Probability 1997, Vol. 7, No. 3, 666 678 EXPLICIT DISTRIBUTIONAL RESULTS IN PATTERN FORMATION By V. T. Stefanov 1 and A. G. Pakes University of Western Australia A new and unified

More information

Weak and strong moments of l r -norms of log-concave vectors

Weak and strong moments of l r -norms of log-concave vectors Weak and strong moments of l r -norms of log-concave vectors Rafał Latała based on the joint work with Marta Strzelecka) University of Warsaw Minneapolis, April 14 2015 Log-concave measures/vectors A measure

More information

Entropy and Ergodic Theory Lecture 17: Transportation and concentration

Entropy and Ergodic Theory Lecture 17: Transportation and concentration Entropy and Ergodic Theory Lecture 17: Transportation and concentration 1 Concentration in terms of metric spaces In this course, a metric probability or m.p. space is a triple px, d, µq in which px, dq

More information

ON AN INEQUALITY OF V. CSISZÁR AND T.F. MÓRI FOR CONCAVE FUNCTIONS OF TWO VARIABLES

ON AN INEQUALITY OF V. CSISZÁR AND T.F. MÓRI FOR CONCAVE FUNCTIONS OF TWO VARIABLES ON AN INEQUALITY OF V. CSISZÁR AND T.F. MÓRI FOR CONCAVE FUNCTIONS OF TWO VARIABLES BOŽIDAR IVANKOVIĆ Faculty of Transport and Traffic Engineering University of Zagreb, Vukelićeva 4 0000 Zagreb, Croatia

More information

a factors The exponential 0 is a special case. If b is any nonzero real number, then

a factors The exponential 0 is a special case. If b is any nonzero real number, then 0.1 Exponents The expression x a is an exponential expression with base x and exponent a. If the exponent a is a positive integer, then the expression is simply notation that counts how many times the

More information

MAT 570 REAL ANALYSIS LECTURE NOTES. Contents. 1. Sets Functions Countability Axiom of choice Equivalence relations 9

MAT 570 REAL ANALYSIS LECTURE NOTES. Contents. 1. Sets Functions Countability Axiom of choice Equivalence relations 9 MAT 570 REAL ANALYSIS LECTURE NOTES PROFESSOR: JOHN QUIGG SEMESTER: FALL 204 Contents. Sets 2 2. Functions 5 3. Countability 7 4. Axiom of choice 8 5. Equivalence relations 9 6. Real numbers 9 7. Extended

More information

Lecture 2 One too many inequalities

Lecture 2 One too many inequalities University of Illinois Department of Economics Spring 2017 Econ 574 Roger Koenker Lecture 2 One too many inequalities In lecture 1 we introduced some of the basic conceptual building materials of the course.

More information

On the minimum of certain functional related to the Schrödinger equation

On the minimum of certain functional related to the Schrödinger equation Electronic Journal of Qualitative Theory of Differential Equations 2013, No. 8, 1-21; http://www.math.u-szeged.hu/ejqtde/ On the minimum of certain functional related to the Schrödinger equation Artūras

More information

Markov s Inequality for Polynomials on Normed Linear Spaces Lawrence A. Harris

Markov s Inequality for Polynomials on Normed Linear Spaces Lawrence A. Harris New Series Vol. 16, 2002, Fasc. 1-4 Markov s Inequality for Polynomials on Normed Linear Spaces Lawrence A. Harris This article is dedicated to the 70th anniversary of Acad. Bl. Sendov It is a longstanding

More information

Extended dynamic programming: technical details

Extended dynamic programming: technical details A Extended dynamic programming: technical details The extended dynamic programming algorithm is given by Algorithm 2. Algorithm 2 Extended dynamic programming for finding an optimistic policy transition

More information

Selected Exercises on Expectations and Some Probability Inequalities

Selected Exercises on Expectations and Some Probability Inequalities Selected Exercises on Expectations and Some Probability Inequalities # If E(X 2 ) = and E X a > 0, then P( X λa) ( λ) 2 a 2 for 0 < λ

More information

ROBUSTNESS OF THE GAUSSIAN CONCENTRATION INEQUALITY AND THE BRUNN-MINKOWSKI INEQUALITY

ROBUSTNESS OF THE GAUSSIAN CONCENTRATION INEQUALITY AND THE BRUNN-MINKOWSKI INEQUALITY ROBUSTNESS OF THE GAUSSIAN CONCENTRATION INEQUALITY AND THE BRUNN-MINKOWSKI INEQUALITY M. BARCHIESI AND V. JULIN Abstract. We provide a sharp quantitative version of the Gaussian concentration inequality:

More information

Lesson 9 Exploring Graphs of Quadratic Functions

Lesson 9 Exploring Graphs of Quadratic Functions Exploring Graphs of Quadratic Functions Graph the following system of linear inequalities: { y > 1 2 x 5 3x + 2y 14 a What are three points that are solutions to the system of inequalities? b Is the point

More information

A Note on Jackknife Based Estimates of Sampling Distributions. Abstract

A Note on Jackknife Based Estimates of Sampling Distributions. Abstract A Note on Jackknife Based Estimates of Sampling Distributions C. Houdré Abstract Tail estimates of statistics are given; they depend on jackknife estimates of variance of the statistics of interest. 1

More information