Notes on Integrable Functions and the Riesz Representation Theorem Math 8445, Winter 06, Professor J. Segert. f(x) = f + (x) + f (x).

Similar documents
THEOREMS, ETC., FOR MATH 515

Finite-dimensional spaces. C n is the space of n-tuples x = (x 1,..., x n ) of complex numbers. It is a Hilbert space with the inner product

PROBLEMS. (b) (Polarization Identity) Show that in any inner product space

MATH 650. THE RADON-NIKODYM THEOREM

MATH MEASURE THEORY AND FOURIER ANALYSIS. Contents

CHAPTER II HILBERT SPACES

Real Analysis Notes. Thomas Goller

Notions such as convergent sequence and Cauchy sequence make sense for any metric space. Convergent Sequences are Cauchy

Reminder Notes for the Course on Measures on Topological Spaces

The Dirichlet s P rinciple. In this lecture we discuss an alternative formulation of the Dirichlet problem for the Laplace equation:

HILBERT SPACES AND THE RADON-NIKODYM THEOREM. where the bar in the first equation denotes complex conjugation. In either case, for any x V define

Recall that any inner product space V has an associated norm defined by

Math General Topology Fall 2012 Homework 11 Solutions

Math Tune-Up Louisiana State University August, Lectures on Partial Differential Equations and Hilbert Space

L p Spaces and Convexity

Overview of normed linear spaces

Chapter 3: Baire category and open mapping theorems

Linear Normed Spaces (cont.) Inner Product Spaces

08a. Operators on Hilbert spaces. 1. Boundedness, continuity, operator norms

MATHS 730 FC Lecture Notes March 5, Introduction

Problem Set 2: Solutions Math 201A: Fall 2016

An introduction to some aspects of functional analysis

Functional Analysis HW #5

Real Analysis: Part II. William G. Faris

Introduction to Empirical Processes and Semiparametric Inference Lecture 22: Preliminaries for Semiparametric Inference

l(y j ) = 0 for all y j (1)

Course 212: Academic Year Section 1: Metric Spaces

3 Orthogonality and Fourier series

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.

B. Appendix B. Topological vector spaces

1 Topology Definition of a topology Basis (Base) of a topology The subspace topology & the product topology on X Y 3

Recall that if X is a compact metric space, C(X), the space of continuous (real-valued) functions on X, is a Banach space with the norm

SPECTRAL PROPERTIES OF THE LAPLACIAN ON BOUNDED DOMAINS

Your first day at work MATH 806 (Fall 2015)

L p Functions. Given a measure space (X, µ) and a real number p [1, ), recall that the L p -norm of a measurable function f : X R is defined by

1 Functional Analysis

Math 209B Homework 2

Measure, Integration & Real Analysis

Tools from Lebesgue integration

i. v = 0 if and only if v 0. iii. v + w v + w. (This is the Triangle Inequality.)

Basic Properties of Metric and Normed Spaces

Convex Sets Strict Separation in Hilbert Spaces

Exercises to Applied Functional Analysis

ADJOINTS, ABSOLUTE VALUES AND POLAR DECOMPOSITIONS

Math 350 Fall 2011 Notes about inner product spaces. In this notes we state and prove some important properties of inner product spaces.

UNBOUNDED OPERATORS ON HILBERT SPACES. Let X and Y be normed linear spaces, and suppose A : X Y is a linear map.

Analysis Finite and Infinite Sets The Real Numbers The Cantor Set

1 The Projection Theorem

Introduction to Real Analysis Alternative Chapter 1

David Hilbert was old and partly deaf in the nineteen thirties. Yet being a diligent

Contents. 2 Sequences and Series Approximation by Rational Numbers Sequences Basics on Sequences...

Math 5210, Definitions and Theorems on Metric Spaces

Functional Analysis. Martin Brokate. 1 Normed Spaces 2. 2 Hilbert Spaces The Principle of Uniform Boundedness 32

Continuous Functions on Metric Spaces

Functional Analysis I

Chapter 1. Introduction

5 Compact linear operators

V. SUBSPACES AND ORTHOGONAL PROJECTION

Projection Theorem 1

Functional Analysis Exercise Class

Analysis Comprehensive Exam Questions Fall F(x) = 1 x. f(t)dt. t 1 2. tf 2 (t)dt. and g(t, x) = 2 t. 2 t

I teach myself... Hilbert spaces

Your first day at work MATH 806 (Fall 2015)

A VERY BRIEF REVIEW OF MEASURE THEORY

Measure Theory on Topological Spaces. Course: Prof. Tony Dorlas 2010 Typset: Cathal Ormond

Mathematics Department Stanford University Math 61CM/DM Inner products

Part III. 10 Topological Space Basics. Topological Spaces

Real Analysis, 2nd Edition, G.B.Folland Elements of Functional Analysis

4 Linear operators and linear functionals

4 Hilbert spaces. The proof of the Hilbert basis theorem is not mathematics, it is theology. Camille Jordan

CHAPTER I THE RIESZ REPRESENTATION THEOREM

Analysis Preliminary Exam Workshop: Hilbert Spaces

Another consequence of the Cauchy Schwarz inequality is the continuity of the inner product.

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

Vector Spaces. Vector space, ν, over the field of complex numbers, C, is a set of elements a, b,..., satisfying the following axioms.

Infinite-dimensional Vector Spaces and Sequences

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

96 CHAPTER 4. HILBERT SPACES. Spaces of square integrable functions. Take a Cauchy sequence f n in L 2 so that. f n f m 1 (b a) f n f m 2.

Functional Analysis. Franck Sueur Metric spaces Definitions Completeness Compactness Separability...

Problem Set 6: Solutions Math 201A: Fall a n x n,

McGill University Math 354: Honors Analysis 3

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?

Homework I, Solutions

FUNCTIONAL ANALYSIS LECTURE NOTES: COMPACT SETS AND FINITE-DIMENSIONAL SPACES. 1. Compact Sets

NAME: Mathematics 205A, Fall 2008, Final Examination. Answer Key

If Y and Y 0 satisfy (1-2), then Y = Y 0 a.s.

212a1214Daniell s integration theory.

Theorem (4.11). Let M be a closed subspace of a Hilbert space H. For any x, y E, the parallelogram law applied to x/2 and y/2 gives.

The weak topology of locally convex spaces and the weak-* topology of their duals

Week 5 Lectures 13-15

Hilbert Spaces. Hilbert space is a vector space with some extra structure. We start with formal (axiomatic) definition of a vector space.

Weak Convergence Methods for Energy Minimization

Best approximations in normed vector spaces

CONTENTS. 4 Hausdorff Measure Introduction The Cantor Set Rectifiable Curves Cantor Set-Like Objects...

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

1 Compact and Precompact Subsets of H

Math 240 (Driver) Qual Exam (5/22/2017)

SHARP BOUNDARY TRACE INEQUALITIES. 1. Introduction

MEASURE AND INTEGRATION. Dietmar A. Salamon ETH Zürich

1 Math 241A-B Homework Problem List for F2015 and W2016

Transcription:

References: Notes on Integrable Functions and the Riesz Representation Theorem Math 8445, Winter 06, Professor J. Segert Evans, Partial Differential Equations, Appendix 3 Reed and Simon, Functional Analysis, Chapter 1 Rudin Integrable and Square-Integrable Functions Let R be a measurable set. Let f : R be a measurable function. There exist unique nonnegative measurable functions f + : R and f : R such that f(x) = f + (x) f (x), f(x) = f + (x) + f (x). Definition: A measurable function f : R is integrable (or absolutely integrable, or summable) if both of the following integrals are finite (nonnegative real numbers): f + dx <, f dx <. The one-norm of an integrable function is the finite nonnegative real number f 1 = f dx = f + dx + f dx 0. The integral of an integrable function f is bounded above and below: f 1 f dx f 1. Definition: A measurable function f : R is square-integrable (or square-summable) if the following integral is finite (nonnegative real number): f 2 dx <. The two-norm of a square-integrable function is the nonnegative real number (here y 1 2 nonnegative square root) ( ) 1 ( ) 1 f 2 = f 2 2 dx = f 2 2 dx 0. = y 0 is the Exercise: Suppose f and g are square integrable. Prove the Cauchy inequality f g 1 f 2 g 2. The Cauchy inequality has a very important corollary: Corollary: The product f g of two square-integrable functions has finite integral: fg 1 f g dx fg 1. Exercise: Supposing that has finite Lebesgue measure, use the Cauchy inequality to show that every square-integrable function is integrable. Show that this is not true if has infinite Lebesgue measure. 1

We say that f : R is locally integrable (or locally summable) if the restriction to every subset Y of finite measure µ(y ) < is integrable. If has finite measure, then locally integrable is the same as integrable. But if has infinite measure, then a function may be locally integrable without being integrable. For example, consider the constant function f = 1 on = R. Vanishing Almost Everywhere Let f : R be a measurable function. We say that f vanishes everywhere if the preimage of {0} is the entire set, f 1 ({0}) =. Equivalently, f vanishes everywhere if the preimage of the complement R {0} is the empty set, f 1 (R {0}) =. We have often used the following fact and its generalization to higher dimensions, for example in the discussion of the Dirichlet principle: Lemma: Suppose = U is the closure of an open and bounded subset U R, and f : R is a continuous function. Then f vanishes everywhere if and only if f 1 = 0. The result remains true is the one-norm if replaced by the two-norm. We want a similar result that holds even for measurable functions which are not continuous. Note that the empty set has zero Lebesgue measure, µ( ) = 0, but there are also lots of non-emtpy sets with zero measure. In general, one says that a given property holds almost everywhere if it holds on some set of zero measure. In particular: Definition: A measurable function f : R vanishes almost everywhere if the preimage of R {0} has zero Lebesgue measure, µ(f 1 (R {0})) = 0. Exercise: Suppose f : R integrable. Show that f 1 = 0 if and only if f vanishes almost everywhere. Exercise: Suppose f : R is square-integrable. Show that f 2 = 0 if and only if f vanishes almost everywhere. MAIN RESULT: Return to the Riesz Representation Theorem We now consider an open bounded interval U = (a, b) R and its closure = U = [a, b]. Let V = C([a, b]) be the vector space of continuous functions f : R, with the inner product (f, g) = f g dx. We have already seen that this inner product space is not complete, so the strong form of the Riesz representation theorem does not apply. In particular, it is easy to construct a Cauchy sequence f n of continuous functions that does not converge to any continuous function. We now have the ingredients to fix this problem, although the fix introduces some complications. The following observation was already used implicitly in defining the inner product on V, let us state it explicitly: Lemma: C([a, b]) is a subspace of L 2 ([a, b]). Proof: We are claiming that a continuous function f : [a, b] R is square-integrable. Now [a, b] is a bounded and closed subset or R, so it is compact. A continuous function on a compact set attains a maximum and minimum, so f is bounded. Now [a, b] has finite measure µ([a, b]) = b a. A bounded measurable function on a set of finite measure is integrable, and square integrable. QED 2

We are now ready to state the main result. It says that the Riesz representation theorem works if we accept some modifications: Theorem: (Main Result) Let φ : C([a, b]) R be a bounded linear map. Then there exists a squareintegrable function q : [a, b] R such that φ(f) = (q, f) = [a,b] q f dx for every f C([a, b]). The function q is unique up to addition of a function that vanishes almost everywhere. In particular, we have to weaken continuity to square-integrability, and we have to accept a certain degree of non-uniqueness. Exercise: Let [a, b] = [0, 2]. Consider the bounded linear map φ : C([0, 2]) R defined by φ(f) = Find a square-integrable function q as described by the Theorem. Check that q cannot be continuous. [1,2] Continued... f dx. 3

Overview of the Technical Details This section outlines the ingredients used in the proof of the Main Result. This material is not strictly needed to understand and apply the main result. Definition: Suppose f and g are measurable functions. If the difference f g vanishes almost everywhere, we will say that f and g are equal almost everywhere, and will denote this by f g. Exercise: Show that the set of integrable functions is a vector space, and similarly for the set of squareintegrable functions. (Note that the set of all functions f : R is a vector space, we need to check that these subsets are vector subspaces.) The Lebesgue space L 1 Exercise: Show that is an equivalence relation on the set of integrable functions. Definition: The set of equivalence classes is called L 1 (). An element f L 1 is an integrable function defined up to addition of a function vanishing almost everywhere. It does not make sense to ask the value of f L 1 at a point x, since this can change if one adds a function vanishing almost everywhere (but not vanishing at the point x.) However, adding such a function does not affect the values of either of the two integrals f 1 = f dx, f dx. So both of these integrals are well-defined for an element f L 1, even though f(x) is not. The following Theorem and Remark are not needed for our applications, and may be skipped: Theorem: L 1 () with 1 is a complete normed space (real Banach space). Remark: f 1 does not satisfy the parallelogram identity, so it is not the norm associated to any inner product. The Lebesgue space L 2 Exercise: Show that is an equivalence relation on the set of square-integrable functions. Definition: The set of equivalence classes is called L 2 (). An element f L 2 is a square-integrable function defined up to addition of a function vanishing almost everywhere. It does not make sense to ask the value of f L 2 at a point x, since this can change if one adds a function vanishing almost everywhere (but not vanishing at the point x.) However, adding such a function does not affect the values of the integral ( f 2 ) 2 = f 2 dx = f 2 dx. So this integral is well-defined for an element f L 2, even though f(x) is not. In fact this norm is associated to an inner product. The integral (f, g) = f g dx. is well-defined for elements f, g L 2 (and is finite by the Cauchy inequality). Proposition: (f, g) is an inner product on the vector space L 2 (). The associated norm is of course (f, f) = f 2 dx = f 2. 4

The following two non-trivial theorems are essential: Theorem: (see Rudin Theorem 11.42) L 2 () with the inner product (f, g) is a complete inner-product space (real Hilbert space). This means that every Cauchy sequence f n L 2 () converges to some f L 2 () (and this f is unique). Now specialize to the closed interval = [a, b] R. Clearly this is a measurable set. Theorem: (Rudin Theorem 11.38) C([a, b]) is a dense subset of L 2 ([a, b]). All the technical details we need from Lebesgue theory are contained in the following statement, which combines the two previous theorems: Proposition: Every Cauchy sequence f n C([a, b]) has a unique limit f L 2 ([a, b]). Equivalently, L 2 ([a, b]) is a completion of C([a, b]) with respect to the inner product (f, g). Combining with the Riesz Representation Theorem, the Main Result follows. The End 5