Topics in Probability Theory and Stochastic Processes Steven R. Dunbar. The Weak Law of Large Numbers

Similar documents
In nite Sequences. Dr. Philippe B. Laval Kennesaw State University. October 9, 2008

A probabilistic proof of a binomial identity

Properties of MLE: consistency, asymptotic normality. Fisher information.

UC Berkeley Department of Electrical Engineering and Computer Science. EE 126: Probablity and Random Processes. Solutions 9 Spring 2006

A Recursive Formula for Moments of a Binomial Distribution

Sequences and Series

Overview of some probability distributions.

SECTION 1.5 : SUMMATION NOTATION + WORK WITH SEQUENCES

BINOMIAL EXPANSIONS In this section. Some Examples. Obtaining the Coefficients

Incremental calculation of weighted mean and variance

1 Computing the Standard Deviation of Sample Means

Department of Computer Science, University of Otago

Asymptotic Growth of Functions

Output Analysis (2, Chapters 10 &11 Law)

Lecture 4: Cauchy sequences, Bolzano-Weierstrass, and the Squeeze theorem

THE HEIGHT OF q-binary SEARCH TREES

Chapter 6: Variance, the law of large numbers and the Monte-Carlo method

Example 2 Find the square root of 0. The only square root of 0 is 0 (since 0 is not positive or negative, so those choices don t exist here).

CHAPTER 7: Central Limit Theorem: CLT for Averages (Means)

A RANDOM PERMUTATION MODEL ARISING IN CHEMISTRY

Chapter 7 - Sampling Distributions. 1 Introduction. What is statistics? It consist of three major areas:

Hypothesis testing. Null and alternative hypotheses

1. C. The formula for the confidence interval for a population mean is: x t, which was


Lecture 13. Lecturer: Jonathan Kelner Scribe: Jonathan Pines (2009)

University of California, Los Angeles Department of Statistics. Distributions related to the normal distribution

SAMPLE QUESTIONS FOR FINAL EXAM. (1) (2) (3) (4) Find the following using the definition of the Riemann integral: (2x + 1)dx

Annuities Under Random Rates of Interest II By Abraham Zaks. Technion I.I.T. Haifa ISRAEL and Haifa University Haifa ISRAEL.

An Efficient Polynomial Approximation of the Normal Distribution Function & Its Inverse Function

I. Chi-squared Distributions

Section 11.3: The Integral Test

Convexity, Inequalities, and Norms

Chapter 5: Inner Product Spaces

3 Basic Definitions of Probability Theory

The Stable Marriage Problem

THE ABRACADABRA PROBLEM

A Mathematical Perspective on Gambling

Chapter 7 Methods of Finding Estimators

MEI Structured Mathematics. Module Summary Sheets. Statistics 2 (Version B: reference to new book)

Our aim is to show that under reasonable assumptions a given 2π-periodic function f can be represented as convergent series

Central Limit Theorem and Its Applications to Baseball

Week 3 Conditional probabilities, Bayes formula, WEEK 3 page 1 Expected value of a random variable

Hypergeometric Distributions

.04. This means $1000 is multiplied by 1.02 five times, once for each of the remaining sixmonth

Factors of sums of powers of binomial coefficients

Lecture 5: Span, linear independence, bases, and dimension

Trigonometric Form of a Complex Number. The Complex Plane. axis. ( 2, 1) or 2 i FIGURE The absolute value of the complex number z a bi is

Non-life insurance mathematics. Nils F. Haavardsson, University of Oslo and DNB Skadeforsikring

Chapter 7: Confidence Interval and Sample Size

Acta Acad. Paed. Agriensis, Sectio Mathematicae 29 (2002) ALMOST SURE FUNCTIONAL LIMIT THEOREMS IN L p( ]0, 1[ ), WHERE 1 p <

Confidence Intervals for One Mean

Chapter 14 Nonparametric Statistics

MARTINGALES AND A BASIC APPLICATION

WHEN IS THE (CO)SINE OF A RATIONAL ANGLE EQUAL TO A RATIONAL NUMBER?

5: Introduction to Estimation

Taking DCOP to the Real World: Efficient Complete Solutions for Distributed Multi-Event Scheduling

Measures of Spread and Boxplots Discrete Math, Section 9.4

Math C067 Sampling Distributions

NOTES ON PROBABILITY Greg Lawler Last Updated: March 21, 2016

Definition. A variable X that takes on values X 1, X 2, X 3,...X k with respective frequencies f 1, f 2, f 3,...f k has mean

1 Review of Probability

On Formula to Compute Primes. and the n th Prime

Integer Factorization Algorithms

5.3. Generalized Permutations and Combinations

Doktori értekezés Katona Zsolt 2006

Unbiased Estimation. Topic Introduction

1. MATHEMATICAL INDUCTION

5 Boolean Decision Trees (February 11)

A note on the boundary behavior for a modified Green function in the upper-half space

Section 8.3 : De Moivre s Theorem and Applications

Case Study. Normal and t Distributions. Density Plot. Normal Distributions

Institute of Actuaries of India Subject CT1 Financial Mathematics

Discrete Mathematics and Probability Theory Spring 2014 Anant Sahai Note 13

Inverse Gaussian Distribution

Theorems About Power Series

Domain 1: Designing a SQL Server Instance and a Database Solution

Analyzing Longitudinal Data from Complex Surveys Using SUDAAN

Probabilistic Engineering Mechanics. Do Rosenblatt and Nataf isoprobabilistic transformations really differ?

Modified Line Search Method for Global Optimization

Class Meeting # 16: The Fourier Transform on R n

One-sample test of proportions


MINING RELAXED GRAPH PROPERTIES IN INTERNET

THIN SEQUENCES AND THE GRAM MATRIX PAMELA GORKIN, JOHN E. MCCARTHY, SANDRA POTT, AND BRETT D. WICK

How To Solve The Homewor Problem Beautifully

THE ARITHMETIC OF INTEGERS. - multiplication, exponentiation, division, addition, and subtraction

Elementary Theory of Russian Roulette

where: T = number of years of cash flow in investment's life n = the year in which the cash flow X n i = IRR = the internal rate of return

CHAPTER 3 DIGITAL CODING OF SIGNALS

Center, Spread, and Shape in Inference: Claims, Caveats, and Insights

Research Article Sign Data Derivative Recovery

A Combined Continuous/Binary Genetic Algorithm for Microstrip Antenna Design

A Faster Clause-Shortening Algorithm for SAT with No Restriction on Clause Length

AP Calculus AB 2006 Scoring Guidelines Form B

Present Values, Investment Returns and Discount Rates

Irreducible polynomials with consecutive zero coefficients

THE REGRESSION MODEL IN MATRIX FORM. For simple linear regression, meaning one predictor, the model is. for i = 1, 2, 3,, n

Maximum Likelihood Estimators.

Basic Elements of Arithmetic Sequences and Series

Tradigms of Astundithi and Toyota

Transcription:

Steve R. Dubar Departmet o Mathematics 203 Avery Hall Uiversity o Nebrasa-Licol Licol, NE 68588-0130 http://www.math.ul.edu Voice: 402-472-3731 Fax: 402-472-8466 Topics i Probability Theory ad Stochastic Processes Steve R. Dubar The Wea Law o Large Numbers Ratig Mathematicias Oly: prologed scees o itese rigor. 1

Questio o the Day Cosider a air (p = 1/2 = q) coi tossig game carried out or 1000 tosses. Explai i a setece what the law o averages says about the outcomes o this game. Be as precise as possible. Key Cocepts 1. Marov s Iequality: Let X be a radom variable taig oly oegative values. The or each a > 0 P [X > a] E [X] /a; 2. Chebyshev s Iequality: Let X be a radom variable. The or a > 0 P [ X E [X] a] Var [X] a 2 3. Wea Law o Large Numbers: For ɛ > 0 [ ] S P p > ɛ 0 as ad the covergece is uiorm i p. 4. Let be a real uctio that is deied ad cotiuous o the iterval [0, 1]. The ( ) ( ) sup (x) x (1 x) 0 as. =0 2

Vocabulary 1. The Wea Law o Large Numbers says that or ɛ > 0 [ ] S P p > ɛ 0 as ad the covergece is uiorm i p. 2. The polyomials B, (t) = are called the Berstei polyomials. ( ) x (1 x) Mathematical Ideas Proo o the Wea Law Usig Chebyshev s Iequality Propositio 1 (Marov s Iequality). Let X be a radom variable taig oly o-egative values. The or each a > 0 Proo. P [X a] E [X] /a. P [X a] = E [I X a ] = dp [] X a x dp [] a 1 a E [X] 3

Propositio 2 (Chebyshev s Iequality). Let X be a radom variable. The or a > 0 Var [X] P [ X E [X] a]. a 2 Proo. This immediately ollows rom Marov s iequality applied to the oegative radom variable (X E [X]) 2. Theorem 3 (Wea Law o Large Numbers). For ɛ > 0 [ ] S P p > ɛ 0 as ad the covergece is uiorm i p. Remar. The otatio P [] idicates that we are cosiderig a amily o probability measures o the sample space Ω. The Wea Law establishes the covergece o the sequece o measures i a particular way. Proo. The variace o the radom variable S is p(1 p). Rewrite the probability as the equivalet evet: [ ] S P p > ɛ = P [ S p > ɛ]. By Chebyshev s iequality P [ S p > ɛ] Var [S ] (ɛ) 2 = Sice p(1 p) 1/4, the proo is complete. Remar. This iequality demostrates that [ ] S P p > ɛ = O(1/) uiormly i p. p(1 p) ɛ 2 1. Remar. Jacob Beroulli origially proved the Wea Law o Large Numbers i 1713 or the special case whe the X i are biomial radom variables. Beroulli had to create a igeious proo to establish the result, sice Chebyshev s iequality was ot ow at the time. The theorem the 4

became ow as Beroulli s Theorem. Simeo Poisso proved a geeralizatio o Beroulli s biomial Wea Law ad irst called it the Law o Large Numbers. I 1929 the Russia mathematicia Alesadr Khichi proved the geeral orm o the Wea Law o Large Numbers preseted here. May other versios o the Wea Law are ow, with hypotheses that do ot require such striget requiremets as beig idetically distributed, ad havig iite variace. Remar. Aother proo o the Wea Law o Large Numbers usig momet geeratig uctios is i Mathematical Fiace/Cetral Limit Theorem Berstei s Proo o the Weierstrass Approximatio Theorem Theorem 4. Let be a real uctio deied ad cotiuous o the iterval [0, 1]. The ( ) ( ) sup (x) x (1 x) 0 as. =0 Proo. 1. Fix ɛ > 0. Sice cotiuous o the compact iterval [0, 1] it is uiormly cotiuous o [0, 1]. Thereore there is a η > 0 such that (x) (y) < ɛ i x y < η. 2. The expectatio E [(S /)] ca be expressed as a polyomial i p: [ E ( )] S = =0 ( ) P [S = ] = =0 ( ) ( ) p (1 p). 3. By the Wea Law o Large Numbers, or the give ɛ > 0, there is a 0 such that [ ] S P p > η < ɛ. 4. E [ ( ) S (p)] = =0 5 ( ( ) ) (p) P [S = ].

5. Apply the triagle iequality to the right had side ad express i terms o two summatios: ( ( ) ) (p) P [S = ] + p η ( ( ) ) + (p) P [S = ] p >η Note the secod applicatio o the triagle iequality o the secod summatio. 6. Now estimate the terms: ɛp [S = ] + p η p >η 2 sup (x) P [S = ] 7. Fially, do the additio over the idividual values o the probabilities over sigle values to re-write them as probabilities over evets: [ ] [ ] S = ɛp p η S + 2 sup (x) P p > η 8. Now apply the Wea Law to the secod term to see that: ( ) [ E S (p)] < ɛ + 2ɛ sup (x). This shows that E [ ( S ) (p) ] ca be made arbitrarily small, uiormly with respect to p, by picig suicietly large. Remar. The polyomials B, (t) = ( ) x (1 x) are called the Berstei polyomials. The Berstei polyomials have several useul properties: 6

1. B i, (t) = B i, (1 t) 2. B i, (t) 0 3. i=0 B i,(t) = 1 or 0 t 1. Corollary 1. A polyomial o degree uiormly approximatig the cotiuous uctio (x) o the iterval [a, b] is Sources =0 ( a + (b a) ) ( ) ( x a b a ) ( ) b x b a This sectio is adapted rom: Heads or Tails, by Emmauel Lesige, Studet Mathematical Library Volume 28, America Mathematical Society, Providece, 2005, Chapter 5, [3]. Problems to Wor or Uderstadig 1. Suppose X is a cotiuous radom variable with mea ad variace both equal to 20. What ca be said about P [0 X 40]? 2. Suppose X is a expoetially distributed radom variable with mea E [X] = 1. For x = 0.5, 1, ad 2, compare P [X x] with the Marov iequality boud. 3. Suppose X is a Beroulli radom variable with P [X = 1] = p ad P [X = 0] = 1 p = q. Compare P [X 1] with the Marov iequality boud. 4. Let X 1, X 2,..., X 10 be idepedet Poisso radom variables with mea 1. First use the Marov Iequality to get a boud o P [X 1 + + X 10 > 15]. Next id the exact probability that P [X 1 + + X 10 > 15] usig that the act that the sum o idepedet Poisso radom variables with parameters λ 1, λ 2 is agai Poisso with parameter λ 1 + λ 2. 7

5. Cosider a air (p = 1/2 = q) coi tossig game carried out or = 100 tosses. Calculate the exact value [ ] S P p > 1/10 ad compare it to the estimates i the proo o the Wea Law o Large Numbers. 6. Calculate the Berstei polyomial approximatio o si(πx) o degree 1, 2, ad 3 ad plot the graphs o si(πx) ad the approximatios. 7. Calculate the Berstei polyomial approximatio o cos(πx) o degree 1, 2, ad 3 ad plot the graphs o cos(πx) ad the approximatios. 8. Calculate the Berstei polyomial approximatio o exp(πx) o degree 1, 2, ad 3 ad plot the graphs o exp(πx) ad the approximatios. Readig Suggestio: Reereces [1] Leo Breima. Probability. SIAM, 1992. [2] William Feller. A Itroductio to Probability Theory ad Its Applicatios, Volume I, volume I. Joh Wiley ad Sos, third editio, 1973. QA 273 F3712. [3] Emmauel Lesige. Heads or Tails: A Itroductio to Limit Theorems i Probability, volume 28 o Studet Mathematical Library. America Mathematical Society, 2005. 8

Outside Readigs ad Lis: 1. Virtual Laboratories i Probability ad Statistics / Biomial 2. Weisstei, Eric W. Wea Law o Large Numbers. From MathWorld A Wolram Web Resource. Wea Law o Large Numbers. 3. Wiipedia, Wea Law o Large Numbers I chec all the iormatio o each page or correctess ad typographical errors. Nevertheless, some errors may occur ad I would be grateul i you would alert me to such errors. I mae every reasoable eort to preset curret ad accurate iormatio or public use, however I do ot guaratee the accuracy or timeliess o iormatio o this website. Your use o the iormatio rom this website is strictly volutary ad at your ris. I have checed the lis to exteral sites or useuless. Lis to exteral websites are provided as a coveiece. I do ot edorse, cotrol, moitor, or guaratee the iormatio cotaied i ay exteral website. I do t guaratee that the lis are active at all times. Use the lis here with the same cautio as you would all iormatio o the Iteret. This website relects the thoughts, iterests ad opiios o its author. They do ot explicitly represet oicial positios or policies o my employer. Iormatio o this website is subject to chage without otice. Steve Dubar s Home Page, http://www.math.ul.edu/~sdubar1 Email to Steve Dubar, sdubar1 at ul dot edu Last modiied: Processed rom L A TEX source o May 25, 2011 9