Introductory Ergodic Theory and the Birkhoff Ergodic Theorem

Similar documents
A Proof of Birkhoff s Ergodic Theorem

FUNDAMENTALS OF REAL ANALYSIS by. V.1. Product measures

Product measures, Tonelli s and Fubini s theorems For use in MAT3400/4400, autumn 2014 Nadia S. Larsen. Version of 13 October 2014.

BIRKHOFF ERGODIC THEOREM

The Pointwise Ergodic Theorem and its Applications

Measure and Measurable Functions

Introduction to Probability. Ariel Yadin. Lecture 7

Advanced Stochastic Processes.

Integrable Functions. { f n } is called a determining sequence for f. If f is integrable with respect to, then f d does exist as a finite real number

B Supplemental Notes 2 Hypergeometric, Binomial, Poisson and Multinomial Random Variables and Borel Sets

It is always the case that unions, intersections, complements, and set differences are preserved by the inverse image of a function.

Introduction to Probability. Ariel Yadin. Lecture 2

HOMEWORK #4 - MA 504

Chapter 0. Review of set theory. 0.1 Sets

s = and t = with C ij = A i B j F. (i) Note that cs = M and so ca i µ(a i ) I E (cs) = = c a i µ(a i ) = ci E (s). (ii) Note that s + t = M and so

If a subset E of R contains no open interval, is it of zero measure? For instance, is the set of irrationals in [0, 1] is of measure zero?

Lecture 3 The Lebesgue Integral

Axioms of Measure Theory

Introduction to Ergodic Theory and its Applications to Number Theory. Karma Dajani

f n (x) f m (x) < ɛ/3 for all x A. By continuity of f n and f m we can find δ > 0 such that d(x, x 0 ) < δ implies that

FUNDAMENTALS OF REAL ANALYSIS by

Infinite Sequences and Series

1 Convergence in Probability and the Weak Law of Large Numbers

University of Colorado Denver Dept. Math. & Stat. Sciences Applied Analysis Preliminary Exam 13 January 2012, 10:00 am 2:00 pm. Good luck!

lim za n n = z lim a n n.

Singular Continuous Measures by Michael Pejic 5/14/10

Archimedes - numbers for counting, otherwise lengths, areas, etc. Kepler - geometry for planetary motion

MATH301 Real Analysis (2008 Fall) Tutorial Note #7. k=1 f k (x) converges pointwise to S(x) on E if and

5 Birkhoff s Ergodic Theorem

MATH 413 FINAL EXAM. f(x) f(y) M x y. x + 1 n

1 Lecture 2: Sequence, Series and power series (8/14/2012)

Advanced Real Analysis

Sequences and Series of Functions

Topics. Homework Problems. MATH 301 Introduction to Analysis Chapter Four Sequences. 1. Definition of convergence of sequences.

The Borel hierarchy classifies subsets of the reals by their topological complexity. Another approach is to classify them by size.

Lecture 3 : Random variables and their distributions

Sequences. Notation. Convergence of a Sequence

Chapter 3. Strong convergence. 3.1 Definition of almost sure convergence

Real Numbers R ) - LUB(B) may or may not belong to B. (Ex; B= { y: y = 1 x, - Note that A B LUB( A) LUB( B)

Lecture Notes for Analysis Class

Probability and Random Processes

The Boolean Ring of Intervals

MA131 - Analysis 1. Workbook 2 Sequences I

Distribution of Random Samples & Limit theorems

Real and Complex Analysis, 3rd Edition, W.Rudin

Relations Among Algebras

2.1. The Algebraic and Order Properties of R Definition. A binary operation on a set F is a function B : F F! F.

Integration Theory: Lecture notes 2013

Lecture 6: Integration and the Mean Value Theorem. slope =

Math 341 Lecture #31 6.5: Power Series

62. Power series Definition 16. (Power series) Given a sequence {c n }, the series. c n x n = c 0 + c 1 x + c 2 x 2 + c 3 x 3 +

Definition 4.2. (a) A sequence {x n } in a Banach space X is a basis for X if. unique scalars a n (x) such that x = n. a n (x) x n. (4.

Recitation 4: Lagrange Multipliers and Integration

5 Many points of continuity

2 Banach spaces and Hilbert spaces

MEASURE-PRESERVING DYNAMICAL SYSTEMS AND APPROXIMATION TECHNIQUES

Math 525: Lecture 5. January 18, 2018

Review Problems 1. ICME and MS&E Refresher Course September 19, 2011 B = C = AB = A = A 2 = A 3... C 2 = C 3 = =

SOME TRIBONACCI IDENTITIES

6.3 Testing Series With Positive Terms

Notes 27 : Brownian motion: path properties

Exercises 1 Sets and functions

Solution. 1 Solutions of Homework 1. Sangchul Lee. October 27, Problem 1.1

Convergence of random variables. (telegram style notes) P.J.C. Spreij

Metric Space Properties

INFINITE SEQUENCES AND SERIES

A) is empty. B) is a finite set. C) can be a countably infinite set. D) can be an uncountable set.

Chapter 6 Infinite Series

Lecture Chapter 6: Convergence of Random Sequences

Sequences I. Chapter Introduction

Sequence A sequence is a function whose domain of definition is the set of natural numbers.

Mathematical Description of Discrete-Time Signals. 9/10/16 M. J. Roberts - All Rights Reserved 1

18.440, March 9, Stirling s formula

Statistics 310B, Lecture Notes, Wednesday, February 25, 2015

Let us give one more example of MLE. Example 3. The uniform distribution U[0, θ] on the interval [0, θ] has p.d.f.

Solutions to HW Assignment 1

DANIELL AND RIEMANN INTEGRABILITY

Dupuy Complex Analysis Spring 2016 Homework 02

Riesz-Fischer Sequences and Lower Frame Bounds

PAPER : IIT-JAM 2010

6 Infinite random sequences

7 Sequences of real numbers

Lecture 6: Integration and the Mean Value Theorem

MAT1026 Calculus II Basic Convergence Tests for Series

Lecture 3: Convergence of Fourier Series

INTRODUCTION TO SPECTRAL THEORY

Introduction to Functional Analysis

Fall 2013 MTH431/531 Real analysis Section Notes

Chapter 10: Power Series

Chapter IV Integration Theory

(A sequence also can be thought of as the list of function values attained for a function f :ℵ X, where f (n) = x n for n 1.) x 1 x N +k x N +4 x 3

Mathematical Methods for Physics and Engineering

Math Solutions to homework 6

4 The Sperner property.

Math 2784 (or 2794W) University of Connecticut

Here are some examples of algebras: F α = A(G). Also, if A, B A(G) then A, B F α. F α = A(G). In other words, A(G)

THE STRONG LAW OF LARGE NUMBERS FOR STATIONARY SEQUENCES

MA131 - Analysis 1. Workbook 3 Sequences II

Lecture 12: November 13, 2018

y X F n (y), To see this, let y Y and apply property (ii) to find a sequence {y n } X such that y n y and lim sup F n (y n ) F (y).

Transcription:

Itroductory Ergodic Theory ad the Birkhoff Ergodic Theorem James Pikerto Jauary 14, 2014 I this expositio we ll cover a itroductio to ergodic theory. Specifically, the Birkhoff Mea Theorem. Ergodic theory is geerally described as the study of dyamical systems that have a ivariat measure. This reaches has relatios to flows (such as the Poicaré s Recurrece Theorem, etc.). Let s begi by discussig what it meas for a dyamical system to be ergodic. We ll start by defiig a probability space. Defiitio 0.1. A probability space is a triple (X, F, µ), where: X is a set of possible outcomes F is a set of evets, where each evet is a set cotaiig outcomes µ is a assigmet of probabilities to each evet i F. Defiitio 0.2. Ergodicity Let T : X X be a measure-preservig trasformatio. T is ergodic with respect to µ if F F s.t. T 1 (F ) = F either µ(f ) = 0 or µ(f ) = 1. Here are some examples: Example 0.3. If X is fiite ad has uiform measure, T : X X is ergodic IFF it s a cycle. Example 0.4. If T is ergodic, the so is T 1. T is ot ecessarily ergodic for all, as we ca see from the previous example (this property total ergodicity). 1

Ituitively, we ca thik of T as a trasformatio w.r.t. time. The, a ergodic fuctio describes a radom process for which the time average of oe sequece equals the total average. Now there is a atural aalogy to measurable flows; I discussed flows i my presetatio earlier i the course. Let T t be a measurable flow o a probability space (X, F, µ). A evet F F is ivariat mod 0 uder T t if t R, X(T t (F ) F ) = 0 ( meas symmetric differece). Now that we uderstad the defiitio of a probability space ad of Ergodicity, we ca begi provig the Birkhoff Theorem. We will first prove a lemma. The best summary of the probability of a evet is the frequecy of its occurrece over some large spa of time. This is captured i ergodic theory by the measure preservig map, T, from the probability space to itself. T is the chage from oe outcome of a radom series of evets to the ext. So suppose we have some measurable fuctio f. The we defie the average of measures over time, the Cesáro average, which is defied as follows: Defiitio 0.5. Cesáro Average A (f, x) = 1 1 f(t i (x)). Defiitio 0.6. σ-fiite measure space A measure µ defied o a σ-algebra Σ of subsets of a set X is called fiite if µ(x) R. It is σ-fiite if X ca be writte as the coutable uio of measurable sets with fiite measure. A set i a measure space is of σ- fiite measure if it ca be writte as the coutable uio of sets with fiite measure. Lemma 0.7. Maximal Ergodic Theorem Let T be a measure-preservig trasformatio o the σ-fiite measure space (X, F, µ). Suppose f L 1 (µ). Set E = {x A j (f, x) > 0 for some j }. The E fdµ 0. 2

Proof. The trick here is to fid a o-decreasig series of fuctios ad the redefie our itegral i terms of differeces i the series (so that we are itegratig over a etirely positive space). j 1 Let F (x) = max(0, f(t i (x)) : j ). This will be our o-decreasig series. This gives us F +1 = max(0, f + F T ). But ote that the secod term is always at least 0 over E +1, meaig we have F +1 = f + F T = f = F +1 F T over E +1. Now take the itegral: fdµ = (F +1 F T )dµ. E +1 E +1 But this is coveiet sice F +1 = 0 everywhere except E +1 ad F T 0 everywhere, givig F +1 F T 0 everywhere but E +1. So we ca chage the boudaries o the itegral to prove our theorem! fdµ = (F +1 F T )dµ (F +1 F T )dµ E +1 E +1 X = (F +1 F )dµ 0. X So we ve just show that if the Cesáro average of a L 1 fuctio if positive for a small eough time frame over a subset, the so is its itegral. Corollary 0.8. Set E = =1 E ad E fdµ 0. Theorem 0.9. Birkhoff Ergodic Theorem Let G be a measure-preservig trasformatio o a σ-fiite measure space (X, F, µ), ad f L 1 (µ). There exists a f with A (f, x) f(x) 3

almost everywhere. Proof. This is the deep theorem which we ve bee aimig for. First let s cosider some ratioal u, v. Ad let s set E u,v = {x lima (f, x) > v > u > lima (f, x) > 0}. Now we ca say: if x E u,v for ay u, v the lim A (f, x) exists. So assumig v > 0, otherwise replace f by f ad u > 0. Assume µ(e u,v ) > 0. From its defiitio, T (x) E u,v, the x is also. I other words, T 1 (E u,v ) E u,v, so we may, w.l.o.g. assume E u,v is the etire measure space. So x X = E u,v, s.t.. 1 1 (f(t i (x)) v) > 0 So, by the maximal ergodic lemma(!), we have (f vχ a )dµ 0. Meaig, X X fdµ vµ(a) because X = E from the first Corollary. Now there are several iterestig corollaries that result from Birkhoff s Theorem. Corollary 0.10. Defiig the map L(f) = f,for f L 1 (µ), L(f) 1 f 1 ad so L(f) is a cotiuous projectio from L 1 (µ) oto the subspace of T -ivariat L 1 -fuctios. Proof. As A (g) 1 = g 1 for g 0, ad sice A (f) coverges poitwise to f, L(f) dµ L( f )dµ lim A ( f )dµ = f dµ. ad L(F ) 1 f 1. 4

So we have: L(f)(T (x) = lim A (f, T (x)) = lim 1 f(t i (x)) i=1 = lim( + 1 A +1(f, x) f(x) ) = L(f)(x). So L(f) is T -ivariat. As L(f g) 1 = L(f) L(g) 1 f g 1, L is a cotiuous projectio oto the T -ivariat L 1 fuctios. Corollary 0.11. If (X, F, µ) has o T -ivariat subsets of fiite measure, the L(f) 0, f L 1 (µ). Proof. If X has o T -ivariat sets of fiite measure, the oly T -ivariat L 1 fuctio is idetically equal to 0. Corollary 0.12. If µ(x) <, the A (f) L(f) 1 0. Proof. Defie A = {f L 1 (µ) : A (f) L(f) 1 0}. As the operator L is a cotractio i L 1, A is L 1 -closed (if f i is Cauchy so is L(f i )). f < = all A (f) have the same boud. By the domiated covergece theorem, A (f) L(f) i L 1. L 1 (µ) is the oly closed subspace of L 1 (µ) that cotais all bouded fuctios. Last, we should uderstad why the Birkhoff Theorem is importat i Ergodic Theory. This lies i the relatioship to ergodic maps. I fact we ca use the theorem to directly characterize ergodic maps! Corollary 0.13. Applicatio of Birkhoff s Theorem If {A i } is a coutable collectio of sets L 1 dese i the collectio of all sets ad 1 1 χ Aj (T i (x)) µ(a j ) i=1 for all i ad almost every x the T is ergodic. 5

Proof. Pick A measure-ivariat ad A j arbitrarily from our collectio. µ(a)µ(a c j) = µ(a c i)dµ ( 1 = lim 1 = lim 1 A A χ A c j (T i (x))dµ ) χ T i(a A c j ) (x)dµ = µ(a A c j). Selectig A j so µ(a A j ) 0, µ(a)µ(a c ) = 0. So µ(a) {0, 1}. Refereces: A Itroductio to Ergodic Theory by Walters Fudametals of Measurable Dyamics by Rudolph 6