Introductory Analysis 2 Spring 2010 Exam 1 February 11, 2015

Similar documents
1/12/05: sec 3.1 and my article: How good is the Lebesgue measure?, Math. Intelligencer 11(2) (1989),

d(x n, x) d(x n, x nk ) + d(x nk, x) where we chose any fixed k > N

Selected solutions for Homework 9

MATH 131A: REAL ANALYSIS (BIG IDEAS)

In N we can do addition, but in order to do subtraction we need to extend N to the integers

MATH 202B - Problem Set 5

1.4 Outer measures 10 CHAPTER 1. MEASURE

g 2 (x) (1/3)M 1 = (1/3)(2/3)M.

Solutions Final Exam May. 14, 2014

7. Let X be a (general, abstract) metric space which is sequentially compact. Prove X must be complete.

µ (X) := inf l(i k ) where X k=1 I k, I k an open interval Notice that is a map from subsets of R to non-negative number together with infinity

Continuity. Chapter 4

Math212a1411 Lebesgue measure.

Analysis Qualifying Exam

3 (Due ). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?

MATH41011/MATH61011: FOURIER SERIES AND LEBESGUE INTEGRATION. Extra Reading Material for Level 4 and Level 6

Midterm 1. Every element of the set of functions is continuous

Introductory Analysis I Fall 2014 Homework #9 Due: Wednesday, November 19

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

Economics 204 Fall 2011 Problem Set 2 Suggested Solutions

Continuity. Chapter 4

Solutions to Problem Set 5 for , Fall 2007

Problem set 1, Real Analysis I, Spring, 2015.

Metric Spaces and Topology

In N we can do addition, but in order to do subtraction we need to extend N to the integers

Exercises from other sources REAL NUMBERS 2,...,

Math 4317 : Real Analysis I Mid-Term Exam 1 25 September 2012

Quick Tour of the Topology of R. Steven Hurder, Dave Marker, & John Wood 1

h(x) lim H(x) = lim Since h is nondecreasing then h(x) 0 for all x, and if h is discontinuous at a point x then H(x) > 0. Denote

Walker Ray Econ 204 Problem Set 3 Suggested Solutions August 6, 2015

Real Analysis - Notes and After Notes Fall 2008

Economics 204 Summer/Fall 2011 Lecture 5 Friday July 29, 2011

Solutions Manual for: Understanding Analysis, Second Edition. Stephen Abbott Middlebury College

JORDAN CONTENT. J(P, A) = {m(i k ); I k an interval of P contained in int(a)} J(P, A) = {m(i k ); I k an interval of P intersecting cl(a)}.

Some Function Problems SOLUTIONS Isabel Vogt Last Edited: May 24, 2013

1 The topology of metric spaces

We are going to discuss what it means for a sequence to converge in three stages: First, we define what it means for a sequence to converge to zero

Analysis Finite and Infinite Sets The Real Numbers The Cantor Set

General Notation. Exercises and Problems

ANALYSIS QUALIFYING EXAM FALL 2017: SOLUTIONS. 1 cos(nx) lim. n 2 x 2. g n (x) = 1 cos(nx) n 2 x 2. x 2.

First In-Class Exam Solutions Math 410, Professor David Levermore Monday, 1 October 2018

Solutions Final Exam May. 14, 2014

Math 140A - Fall Final Exam

Real Analysis Comprehensive Exam Fall A(k, ε) is of Lebesgue measure zero.

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi

MA5206 Homework 4. Group 4. April 26, ϕ 1 = 1, ϕ n (x) = 1 n 2 ϕ 1(n 2 x). = 1 and h n C 0. For any ξ ( 1 n, 2 n 2 ), n 3, h n (t) ξ t dt

MATH31011/MATH41011/MATH61011: FOURIER ANALYSIS AND LEBESGUE INTEGRATION. Chapter 2: Countability and Cantor Sets

Defining the Integral

2 (Bonus). Let A X consist of points (x, y) such that either x or y is a rational number. Is A measurable? What is its Lebesgue measure?

Undergraduate Notes in Mathematics. Arkansas Tech University Department of Mathematics

We are now going to go back to the concept of sequences, and look at some properties of sequences in R

Spring 2014 Advanced Probability Overview. Lecture Notes Set 1: Course Overview, σ-fields, and Measures

One-to-one functions and onto functions

Three hours THE UNIVERSITY OF MANCHESTER. 24th January

A LITTLE REAL ANALYSIS AND TOPOLOGY

MATH 51H Section 4. October 16, Recall what it means for a function between metric spaces to be continuous:

Math 118B Solutions. Charles Martin. March 6, d i (x i, y i ) + d i (y i, z i ) = d(x, y) + d(y, z). i=1

Math212a1413 The Lebesgue integral.

REAL ANALYSIS I Spring 2016 Product Measures

DRAFT MAA6616 COURSE NOTES FALL 2015

FUNDAMENTALS OF REAL ANALYSIS by. II.1. Prelude. Recall that the Riemann integral of a real-valued function f on an interval [a, b] is defined as

Part V. 17 Introduction: What are measures and why measurable sets. Lebesgue Integration Theory

Chapter 3 Continuous Functions

MATH 23b, SPRING 2005 THEORETICAL LINEAR ALGEBRA AND MULTIVARIABLE CALCULUS Midterm (part 1) Solutions March 21, 2005

converges as well if x < 1. 1 x n x n 1 1 = 2 a nx n

Mid Term-1 : Practice problems

n [ F (b j ) F (a j ) ], n j=1(a j, b j ] E (4.1)

Proof. We indicate by α, β (finite or not) the end-points of I and call

Introductory Analysis 1 Fall 2009 Homework 4 Solutions to Exercises 1 3

Measures and Measure Spaces

MATH NEW HOMEWORK AND SOLUTIONS TO PREVIOUS HOMEWORKS AND EXAMS

Measures. Chapter Some prerequisites. 1.2 Introduction

U e = E (U\E) e E e + U\E e. (1.6)

Problem List MATH 5143 Fall, 2013

Real Analysis Problems

Metric spaces and metrizability

Summary of Real Analysis by Royden

Iowa State University. Instructor: Alex Roitershtein Summer Homework #5. Solutions

1 Measurable Functions

Measures. 1 Introduction. These preliminary lecture notes are partly based on textbooks by Athreya and Lahiri, Capinski and Kopp, and Folland.

MATH MEASURE THEORY AND FOURIER ANALYSIS. Contents

From now on, we will represent a metric space with (X, d). Here are some examples: i=1 (x i y i ) p ) 1 p, p 1.

Introduction to Proofs in Analysis. updated December 5, By Edoh Y. Amiran Following the outline of notes by Donald Chalice INTRODUCTION

DIFFERENTIAL GEOMETRY Multivariable Calculus, a refresher

HW 4 SOLUTIONS. , x + x x 1 ) 2

MT804 Analysis Homework II

F (x) = P [X x[. DF1 F is nondecreasing. DF2 F is right-continuous

Set, functions and Euclidean space. Seungjin Han

Math 320-2: Final Exam Practice Solutions Northwestern University, Winter 2015

Supplementary Notes for W. Rudin: Principles of Mathematical Analysis

REAL VARIABLES: PROBLEM SET 1. = x limsup E k

Principle of Mathematical Induction

The Lebesgue Integral

MATH5011 Real Analysis I. Exercise 1 Suggested Solution

Partial Solutions to Folland s Real Analysis: Part I

Bonus Homework. Math 766 Spring ) For E 1,E 2 R n, define E 1 + E 2 = {x + y : x E 1,y E 2 }.

A List of Problems in Real Analysis

2.1 Convergence of Sequences

Analysis III. Exam 1

Lecture Notes in Advanced Calculus 1 (80315) Raz Kupferman Institute of Mathematics The Hebrew University

Transcription:

Introductory Analysis 2 Spring 21 Exam 1 February 11, 215 Instructions: You may use any result from Chapter 2 of Royden s textbook, or from the first four chapters of Pugh s textbook, or anything seen in class. Anything more advanced needs to be justified (and probably should not be used). You may also use results from previous homework. 1. Let g : [, 1] R be continuous. Show there exists a unique continuous function f : [, 1] R such that f(x) = g(x) + e (x t)2 f(t) dt. Hint: All you need to do here are some simple computations, and quoting past results correctly. Make sure you quote them correctly. Something that may or may not be useful (depending on how you decide to solve the problem) is: If φ : [a, b] R is continuous, φ(x) for all x [a, b] and φ is NOT identically zero, then b φ(t) dt >. a Solution. The idea in solving this exercise is essentially the same as in Picard s Theorem. The space M = C([, 1]) = {f : [, 1] R} is (as seen in part 1) a complete metric space with the metric defined by d(f, g) = f g, where Given g M, Define L : M M by L(f)(x) = g(x) + f = max x1 f(x). e (x t)2 f(t) dt. It is clear that Lf : [, 1] [, 1] is continuous; that is Lf M if f M so that we do indeed have L mapping from M into M. If f, h M, then for x [, 1], L(f)(x) L(h)(x) = e (x t)2 f(t) dt e (x t)2 h(t) dt e (x t)2 f(t) h(t) dt ( ) ( ) e (x t)2 dt = e t2 dt f g ( 1 ) e t2 dt f g. This being true for all x [, 1], we proved L(f) L(h) α f h ; i.e., d(l(f), L(h)) αd(f, h) for all f, h M, where α = 1 dt. We notice that < α < 1. In fact, e t2 e t2 < 1 for all t (, 1] so that! e t2 > for all t (, 1], hence 1 α = 1 (1 e t2 ) dt >. It follows that L is a contraction from M to M. since (as mentioned) M is complete, L has a fixed point; there is f M such that f = Lf.

2. If I = (a, b) is an open, bounded, interval in R, define λ(i) = e b e a. If A R define µ (A) = inf{ λ(i n ) : {I n } n N ranges over all families of bounded open intervals s.t. A I n }. Prove that µ is an outer measure in R; that is, prove: (a) µ ( ) =. (b) A B implies µ (A) µ (B). (c) µ ( A n) µ (A n ) for all countable families A 1, A 2,... of subsets of R. Solution. The proof of all these properties is identical to the proof that the outer measure m defined in the textbook is indeed an outer measure. I allowed the covering intervals to be empty, which simplifies the proof. However, that is not necessary. Proof that µ ( ) =. Since x e x is continuous, given ϵ >, n N we can find a n, b n R, a n < b n such that e bn e an < ϵ/2 n+1. Setting I n = (a n, b n ) we have λ(i n ) < ϵ/2 n+1. Now I n, hence µ ϵ ( ) λ(i n ) < = ϵ. 2n+1 Since ϵ > was arbitrary, we are done. Proof that µ is monotone. Assume A B R. If {I n } is any family of open intervals covering B, it also covers A hence µ (A) λ(i n ). This implies that µ (A) is all numbers in the set of which µ (B) is the greatest lower bound. Thus µ (A) µ (B). Proof of σ-subadditivity. Assume A, A 1, A 2,... are subsets of R and A A n. Let ϵ > be given. For each n N there is a family {I nk } k N of open and bounded intervals such that A n I nk and λ(i nk) < µ (A n ) + ϵ 2. Then {I n+1 nk } n,k N is a covering of A by a countable family of open intervals,hence µ (A) k,n N λ(i nk ) = λ(i nk ) < Since ϵ > is arbitrary, we are done. ( µ (A n ) + ϵ 2 n+1 ) = Bonus: Is µ (I) = λ (I) if I is a bounded open interval? µ (A n )+ϵ. The answer is yes, the proof is similar (slightly easier because you have to deal only with open, bounded intervals) than the proof that the outer measure of an interval is its length. The fact that µ (I) λ(i) is trivial by the definition. For the converse inequality, assume I = (a, b) I n, where each I n is an open bounded interval. One needs to prove that λ(i ) λ(i n). One proceeds as for Lebesgue outer measure but 2

with a bit of care. The idea is identical, but shrinking an interval by ϵ > does not decrease its λ value by ϵ. We need an ϵ and a δ. We can argue as follows. Because I is open we can assume I n I for all n. In fact, otherwise replace I n by I n I; I n I is againa an open interval and because x e x is strictly increasing, λ(i n I) λ(i n ), so if we prove λ(i ) λ(i n I), then λ(i ) λ(i n) follows. It may be convenient to define f : R R by f(x) = e x. Because it is continuous, it is uniformly continuous on bounded sets; given ϵ > there is δ > such that x y δ; x, y I implies f(x) f(y) < ϵ/2. Let J = [a+δ, b δ] so that J is a closed and bounded interval, J I. By compactness, there is N N such that I 1,..., I N cover J. We relabel these intervals as follows as in Royden. There is one of these intervals containing a + δ. Relabel it as I 1 = (a 1, b 1 ). If b 1 < b δ we are done. If not, there is another one of this finite set of intervals containing b 2. Relabel it as I 2 = (a 2, b 2 ). And so forth. The process has to end, eventually one of the intervals contains b δ, since we only have a finite number of intervals. So we have for some m N a 1 < a + δ < a 2 < b 1 < b 2 < a 3 < < a m < b δ < b m. Now, as in Royden (p. 32), but adding the function f. Since f is increasing, it preserves order relations. Notice that (f(a m ) f(b m 1 )) + (f(a m 1 ) f(b m 2 )) + (f(a 2 ) f(b 1 )) < because b k (a k+1, b k+1 ); subtracting this expressions increases quantities. So, as in Royden, but with an f around points, f (b δ) f (a + δ) < f(b m ) f(a m ) < f(b m ) (f(a m ) f(b m 1 )) (f(a 2 ) f(b 1 )) f(a 1 ) = (f(b m ) f(a m )) + (f(b m 1 ) f(a m 1 ) + + (f(b 1 ) f(a 1 )) m = λ(i k ) This proves that f(b δ) f(a + δ) < λ(i n ), where now the I n s are again the original I n s. By choice of δ, f(b δ) > f(b) ϵ/2, f(a + δ) < f(a) + ϵ/2 so that λ(i) = f(b) f(a) < f(b δ) f(a + δ) + ϵ < m λ(i k ) + ϵ. Since ϵ > is arbitrary, this proves λ(i) m λ(i k); since {I n } was an arbitrary covering of I by open, bounded intervals, this proves λ(i) µ (I). 3. Let E be a measurable subset of R. Prove: m(e) = sup{m(c) : C is a compact subset of E}. Solution. Maybe I should have made this easier by adding the assumption that E is bounded. Then it is an immediate consequence of Theorem 11 in Royden, since closed bounded sets are compact. For an unbounded E there are several ways to proceed. one way is to prove it first for a 3

bounded set. So let s do this, assume first E is bounded. Let ϵ >. By Theorem 11 (p.4) in Royden, there is a closed sibset F of E such that m(e\f ) < ϵ. Since bounded sets have finite measure, we see that m(e) m(f ) < ϵ, hence m(e) < m(f ) + ϵ sup{m(c) : C closed,c E} + ϵ. Since ϵ > was arbitrary we see that m(e) is the sup of the measure of all included closed sets; since the opposite inequality is obvious we have that m(e) = sup{m(c) : C is a closed subset of E} = sup{m(c) : C is a compact subset of E} since compact and closed are the same thing for bounded subsets of R. For the general case, let E n = E [ n, n]. There is then a compact set C n E n such that m(e n ) < m(c n )+1/n. Here is the place where we have to be a little bit careful. So far, to say something like m(a) < m(b) + ϵ for measurable sets A, B, B A, as long as A has finite measure, is equivalent to m(a\b) < ϵ. But if the measures get infinite, things change. Back to E n, C n. We have E 1 E 2 E 3 and E = E n, thus m(e) = lim n m(e n ). We now have unfortunately to consider two cases. Case 1: m(e) <. Given ϵ >, there is then n such that m(e n ) + ϵ/2 > m(e) and such that 1/n < ϵ/2. Then m(e) < m(e n ) + ϵ 2 < m(c n) + 1 n + ϵ 2 < m(c n) + ϵ. We proved that for every ϵ >, E contains a compact subset with measure > m(e) ϵ; the conclusion follows. Case 2. m(e) =. In this case lim n m(e n ) = thus also lim n m(c n ) =. Thus for every R > there is n such that m(c n ) > R; in other words, for every R >, E contains a compact subset of measure > R. It follows that sup{m(c) : C is a compact subset of E} = = m(e). 4. Prove or disprove: There exists a closed subset of the interval [, 1], consisting exclusively of irrational numbers, having positive measure. That is, can there exist F closed, F [, 1]\Q, m(f ) >? Solution. Since the set of irrational numbers is a measurable set of infinite measure, by the previous exercise it contains compact, hence closed, subsets of arbitrarily large measure. So there certainly exists such a set F. Many such sets, even compact such sets. 5. Let E R, assume m (E) <. (a) Show there exists an F σ set A and a G δ set U such that A E U and m(a) = m (E) = m(u). Solution. This is exercise 18 (p. 43) in Royden. Half of it is false. But first we first show the existence of the G δ set. One doesn t really need m (E) < for this part; if m (E) = we can take U = R. But the proof for the case m (E) < is a bit more involved. So assume m (E) <. Let ϵ > be given. There exists then a family {I n } of bounded open intervals such that E I n and 4

m (E) l(i n) < m (E) + ϵ. Setting U ϵ = I n, U ϵ is open, E U ϵ and The set U = m (E) m(u ϵ ) m(i n ) = l(i n ) < m (E) + ϵ. U 1/k is a G δ set, E U and since m(u) m(u k ) < m (E) + 1 k for all k, we see that m(u) m (E); since E U we also have m (E) m (U) = m(u). This takes care of the existence of the G δ set. But the F σ set might not exist. I will refer to my answers to Homework 3, Exercise 1. The same argument I use there proves that if E is a measurable set of positive finite measure, if A = C E (defined as in Royden), then m (A) > (has to be since A is not measurable, m (A) < (because E A and m(a) < ), and every measurable subset (in particular every F σ subset) of A has measure. In other words, there is NO F σ subset of A with m(f σ ) = m (A). 6. Show there is a continuous, strictly increasing, function on the interval [, 1] that maps a set of positive measure onto a set of measure. Hint: What goes up must come down. problem seems hard, invert it. Or as Abel used to say, if a Solution. If the problem had stated: Show there is a continuous, strictly increasing, function on the interval [, 2] (instead of [, 1]) that maps a set of positive measure onto a set of measure, it would have been a trivial consequence of Proposition 21 (p.52) of Royden. In fact, by Proposition 21, the function ψ : [, 1] [, 2] defined there is onto, strictly increasing, continuous and maps C onto a set of positive measure. Since ψ is strictly increasing and onto [, 2], it is invertible and ψ 1 : [, 2] [, 1] is strictly increasing and onto. Moreover, ψ 1 is continuous, since ψ is continuous and [, 1] is compact. Almost finally, if we let E = ψ(c), then by Proposition 11, E is a measurable set of positive measure (in fact, of measure 1). Thus ψ 1 maps a measurable set of positive measure, namely E, onto a set of measure, namely C. The transition to the interval [, 1] is quite easy. Let E be as above. The easiest way to proceed is to say that either E 1 = E [, 1] or E 2 = E [, 2] has positive measure. If m(e 1 ) >, then let f = ψ 1 [,1] ; f is clearly strictly increasing, continuous, and maps the non-null measurable set E 1 onto a subset of the Cantor set C, thus onto a null set. If m(e 1 ) =, so m(e 2 ) >, define g : [, 1] [, 1] by g(x) = ψ 1 (x + 1). Once again it is immediate that g is strictly increasing, continuous. Letting E 3 = E 2 1 (E 2 translate by one to the left), we see that E 3 is a measurable subset f [, 1] of positive measure and g(e 3 ) = ψ 1 (E 2 ) C is a null set. Done. There are other ways one can apply Royden s Proposition. For example one can define the sought for function by x ψ 1 (2x); but then one needs to prove that if E is a subset of [, 2] of positive measure, then {x [, 1] : 2x E} also has positive measure. This is fairly easy, but a bit more time consuming than what I did. 5