Chapter 3 Inner Product Spaces. Hilbert Spaces

Similar documents
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)

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.

Introduction to Optimization Techniques

Homework 2. Show that if h is a bounded sesquilinear form on the Hilbert spaces X and Y, then h has the representation

Abstract Vector Spaces. Abstract Vector Spaces

Math Solutions to homework 6

HILBERT SPACE GEOMETRY

Solutions to home assignments (sketches)

LECTURE 8: ORTHOGONALITY (CHAPTER 5 IN THE BOOK)

Lecture Notes for Analysis Class

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

Brief Review of Functions of Several Variables

Riesz-Fischer Sequences and Lower Frame Bounds

5. Matrix exponentials and Von Neumann s theorem The matrix exponential. For an n n matrix X we define

TENSOR PRODUCTS AND PARTIAL TRACES

Singular Continuous Measures by Michael Pejic 5/14/10

2 Banach spaces and Hilbert spaces

HOMEWORK #4 - MA 504

Lecture 10: Bounded Linear Operators and Orthogonality in Hilbert Spaces

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

Sequences and Series of Functions

Introduction to Optimization Techniques. How to Solve Equations

M A T H F A L L CORRECTION. Algebra I 1 4 / 1 0 / U N I V E R S I T Y O F T O R O N T O

Introduction to Functional Analysis

McGill University Math 354: Honors Analysis 3 Fall 2012 Solutions to selected problems

5 Birkhoff s Ergodic Theorem

Math 61CM - Solutions to homework 3

PRELIM PROBLEM SOLUTIONS

MATH 205 HOMEWORK #2 OFFICIAL SOLUTION. (f + g)(x) = f(x) + g(x) = f( x) g( x) = (f + g)( x)

PAPER : IIT-JAM 2010

Measure and Measurable Functions

HILBERT-SCHMIDT AND TRACE CLASS OPERATORS. 1. Introduction

Axioms of Measure Theory

Ma 4121: Introduction to Lebesgue Integration Solutions to Homework Assignment 5

Homework 1 Solutions. The exercises are from Foundations of Mathematical Analysis by Richard Johnsonbaugh and W.E. Pfaffenberger.

A Characterization of Compact Operators by Orthogonality

Apply change-of-basis formula to rewrite x as a linear combination of eigenvectors v j.

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?

REAL ANALYSIS II: PROBLEM SET 1 - SOLUTIONS

Almost Surjective Epsilon-Isometry in The Reflexive Banach Spaces

n=1 a n is the sequence (s n ) n 1 n=1 a n converges to s. We write a n = s, n=1 n=1 a n

We are mainly going to be concerned with power series in x, such as. (x)} converges - that is, lims N n

ANSWERS TO MIDTERM EXAM # 2

Topologie. Musterlösungen

Theorem 3. A subset S of a topological space X is compact if and only if every open cover of S by open sets in X has a finite subcover.

Final Solutions. 1. (25pts) Define the following terms. Be as precise as you can.

(VII.A) Review of Orthogonality

Advanced Real Analysis

Properties of Fuzzy Length on Fuzzy Set

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

CHAPTER 5. Theory and Solution Using Matrix Techniques

On n-collinear elements and Riesz theorem

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

, then cv V. Differential Equations Elements of Lineaer Algebra Name: Consider the differential equation. and y2 cos( kx)

lim za n n = z lim a n n.

b i u x i U a i j u x i u x j

Lecture 3 : Random variables and their distributions

A REMARK ON A PROBLEM OF KLEE

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

McGill University Math 354: Honors Analysis 3 Fall 2012 Solutions to selected problems. x y

A General Iterative Scheme for Variational Inequality Problems and Fixed Point Problems

Linear Elliptic PDE s Elliptic partial differential equations frequently arise out of conservation statements of the form

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 21 11/27/2013

Algebra of Least Squares

Homework 4. x n x X = f(x n x) +

MAS111 Convergence and Continuity

1 lim. f(x) sin(nx)dx = 0. n sin(nx)dx

A Proof of Birkhoff s Ergodic Theorem

Geometry of LS. LECTURE 3 GEOMETRY OF LS, PROPERTIES OF σ 2, PARTITIONED REGRESSION, GOODNESS OF FIT

SOLVED EXAMPLES

Math 220B Final Exam Solutions March 18, 2002

Lecture 4: Grassmannians, Finite and Affine Morphisms

Math 210A Homework 1

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

Differentiable Convex Functions

Chapter 2. Periodic points of toral. automorphisms. 2.1 General introduction

Matrix Algebra from a Statistician s Perspective BIOS 524/ Scalar multiple: ka

INTRODUCTION TO SPECTRAL THEORY

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

Introduction to Probability. Ariel Yadin. Lecture 7

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 +

Machine Learning Theory Tübingen University, WS 2016/2017 Lecture 11

FUNDAMENTALS OF REAL ANALYSIS by

5.1. The Rayleigh s quotient. Definition 49. Let A = A be a self-adjoint matrix. quotient is the function. R(x) = x,ax, for x = 0.

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

NBHM QUESTION 2007 Section 1 : Algebra Q1. Let G be a group of order n. Which of the following conditions imply that G is abelian?

Chapter 0. Review of set theory. 0.1 Sets

A TYPE OF PRIMITIVE ALGEBRA*

It is often useful to approximate complicated functions using simpler ones. We consider the task of approximating a function by a polynomial.

(bilinearity), a(u, v) M u V v V (continuity), a(v, v) m v 2 V (coercivity).

8. Applications To Linear Differential Equations

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

1+x 1 + α+x. x = 2(α x2 ) 1+x

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

Notes #3 Sequences Limit Theorems Monotone and Subsequences Bolzano-WeierstraßTheorem Limsup & Liminf of Sequences Cauchy Sequences and Completeness

Several variables and partial derivatives

Functional Analysis I

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

The Boolean Ring of Intervals

Linearly Independent Sets, Bases. Review. Remarks. A set of vectors,,, in a vector space is said to be linearly independent if the vector equation

Transcription:

Chapter 3 Ier Product Spaces. Hilbert Spaces 3. Ier Product Spaces. Hilbert Spaces 3.- Defiitio. A ier product space is a vector space X with a ier product defied o X. A Hilbert space is a complete ier product space. A ier product o X is a mappig of X X ito the scalar field of X such that for all x y ad z i X ad ay scalar we have (Ip) < x + y z > = < x z > + < y z >. (Ip) < x y > = < x y >. (Ip 3) < x y > = y x the complex cojugate of < y x >. (Ip4) < x x > 0 ad < x x > = 0 if ad oly if x = 0. otes. () A ier product o X defies a orm o X give by x = x x > ad a metric o X give by d( x y ) = x y = x y x y >. () The ier product spaces are ormed spaces Hilbert spaces are Baach spaces. (3) For all x y ad z i a ier product space ad ay scalar a) < x + y z > = < x z > + < y z >. b) < x y > = < x y >. c) < x y + z > = < x y > + < x z >. The the ier product is sesquiliear ( liear i the first factor ad cojugate liear i the secod ). 4) Parallelogram equality. If X is a ier product space the for all x yx x + y + x y = ( x + y ). ( How?) ( From 4) we have: if parallelogram equality is ot satisfied i a ormed space X the X is ot a ier product space.) 3.-3 Defiitio. A elemet x i a ier product space X is said to be orthogoal to yx if < x y > = 0 a we write x y. If A B are subsets of X the x A if x a for all aa ad A B if a b for all aa ad all bb. Examples. 3.-4 Euclidea space R is a Hilbert spaces with ier product defied by < x y > = i i. This ier product iduces the orm x = ( i ) i ad the metric d( x y ) = ad y = (... ). i i i where x = i ( ( ) ) (... )

3.-5 Uitary space C is a Hilbert spaces with ier product defied by < x y > = i. This ier product iduces the orm x = ( ) i i i ad the metric d( x y ) = ad y = (... ). i i i where x = i ( ) (... ) 3.-6 Space L [a b]. The completio of the space of all cotiuous real-valued fuctios o [a b] with orm defied by x = b ier product < x y > = x ( t ) y ( t ) dt. a b x t dt ad the ( ( ) ) ote. The fuctio x(t) ca be exteded to be complex valued o [a b] ad the correspodig ier product is < x y > = x = b b a x ( t ) y ( t ) dt ad the orm is x t dt. The L [a b] becomes as the completio of the ( ( ) ) a space of all all cotiuous complex-valued fuctios o [a b] correspodig to this orm. I each case real or complex L [a b] is a Hilbert space. 3.-7 Hilbert sequece space product defied by < x y > = a. This space is a Hilbert spaces with ier i i. This ier product iduces the orm x = ( i ) where x = ( i ) ad y = ( i ). i p 3.-8 Space. This space with p is ot a ier product space hece is ot a Hilbert spaces. Proof. Cosider x = ( 0 0 0 ) ad y = ( - 0 0 0 ). The x y p ad x + y + x y = ( 0 0 0 0 ) + (0 0 0 0 ) = 8. But ( x + y ) = ( ( 0 0 0 ) + ( - 0 0 0 ) p p ) = [ ( ) + ( ) ] = 4( p ) 8 whe p. Therefore x + y + x y ( x + y ). p Hece by ote 4) above the space with p is ot a ier product space hece is ot a Hilbert spaces i

3.-9 Space C[a b]. This space is ot a ier product space hece is ot a Hilbert spaces. t a Proof. Cosider x(t) = ad y(t) = o the closed iterval J b a ta = [a b]. The x y C[a b] ad x = y = max = ta ta bt x + y = max = ad x y = max = max =. tj ba Hece ( x + y ) = 4 but x + y + x y = 5. The x + y + x y ( x + y ). Hece by ote 4) above the space C[a b] is ot a ier product space hece is ot a Hilbert spaces 3.-0 Remars. () For ay x y i a real ier product space tj < x y > = ( x + y 4 - x y ). () For ay x y i a complex ier product space a) Re < x y > = ( x + y 4 - x y ). a) Im < x y > = ( x + iy 4 - x iy ). ba tj tj ba ba H. W. -9. H.W.* 5 6 8. 3

3. Further Properties of Ier Product Spaces. 3.- Lemma ( Schwarz iequality triagle iequality). If x ad y are elemets i a ier product space the a) < x y > x y ( Schwarz iequality) where the equality sig holds if ad oly if { x y } is a liearly depedet set. b) x + y x + y ( Triagle iequality) where the equality sig holds if ad oly if y =0 or x = cy for some c 0. Proof. a) If y = 0 the < x y > = 0 ad the result is trivial. Let y 0 ad be a scalar we have 0 x - y = < x - y x - y > = < x x > - < x y > - [ < y x > - < y y > ]...() Put yx = yy we have 0 < x x > - yx yy < x y > = x - xy. Hece < x y > x y ad the the result follows. y If { x y } is liearly depedet the there is a scalar such that x = y ad < x y > = < y y > = y = y y = x y. Coversly if equality holds the by () either y = 0 or x - y = 0. Hece { x y } is liearly depedet. b) By usig Schwarz iequality we have x + y = < x + y x + y > = x + < x y > + < y x > + y x + < x y > + < y x > + y x + x y + y = ( x + y ). Hece x + y x + y. If equality holds the < x y > + < y x > = x y. However < x y > + < y x > = Re< x y >. The Re< x y > = x y < x y >. Sice the real part ca t exceed the modulas the Re< x y > = x y = < x y > 0 the by a) { x y } is liearly depedet ad so y = 0 or x = cy. Sice 0 < x y > = < cy y > = c y the c 0. Coversly if y = 0 or x = cy the it is a easy calculatio for gettig x + y = x + y 3.- Lemma ( Cotiuity of ier product). If i a ier product space x x ad y y the < x y > < x y >. H. W. 4-8. H.W.* 8. 4

3.3 Orthogoal Complemets ad Direct Sums Remar. A subset M of a vector space X is said to be covex if for all x ym ad all [0 ] x + (-)ym. Hece every subspace of X is covex ad the itersectio of covex sets is covex. 3.3- Theorem. Let X be a ier product space ad M a o empty covex subset which is complete. The for every xx there exists a uique ym such that = if x y = x y. ym Proof. a) Existece. Sice = if x y the there is a sequece (y ) ym i M such that where = x y. We show that (y ) is Cauchy. Let y x = v. The = v ad v + v m = y + y m x = (y + y m ) x where (y + y m ) M because M is covex. Furthermore y y m = v v m. By usig parallelogram equality we have y y m = v v m = v + v m + ( v + v m ) () + ( + m ). As m y y m 0. Hece for every > 0 there is sufficietly large such that y y m < for all m >. Hece (y ) is a Cauchy sequece i M. But M is complete the (y ) coverges say y ym. The x y. However x y x y + y y = + y y. The x y. Therefore x y =. b) Uiqueess. Suppose that there are y y 0 M with x y = ad x y 0 =. By parallelogram equality y y 0 = (y x) (y 0 x) = y x + y 0 x (y x) + (y 0 x) = + 4 (y + y 0 ) x...() Sice M is covex the (y + y 0 )M which implies that (y + y 0 ) x. By this ad () y y 0 + 4 = 0. Hece y y 0 = 0 that is y = y 0. This proves the uiqueess 3.3- Lemma. I Theorem 3.3- let M be a complete subspace Y ad xx fixed. The z = x y is orthogoal to Y. Proof. Suppose that z = x y is ot orthogoal to Y. The there is y Y such that < z y > = 0. Clearly y 0 otherwise < z y > = 0. Furthermore for ay scalar z y = < z z > < z y > [ < y z > < y y > ] = < z z > [ < y y > ] Choose = y y to get z y = z -. However y z = x y = ad 0 the z y < z =..() Sice Y is a subspace the ( y y ) Y ad so z y = x ( y y ). This is a cotradictio with (). Therefore z = x y is orthogoal to Y 5

3.3-3 Defiitio. A vector space X is said to be the direct sum of two subspaces Y ad Z of X writte X = Y Z if each xx has a uique represetatio x = y + z yy ad zz. The Z is called the algebraic complemet of Y i X. 3.3-4 Theorem ( Projectio Theorem ). Let Y be ay closed subspace of a Hilbert space H. The H = Y Z where Z = Y = { zh : z Y}. Proof. Sice Y is a closed subspace of a Hilbert space H the Y is complete. Sice Y is covex the Theorem 3.3- ad lemma 3.3- imply that for ay xx there is yy such that z = x yy. The x = y + z yy ad zy = Z..() Suppose that there are y y Y ad z z Z with x = y + z = y + z. The y y = z z. However y y Y ad z z Z. The y y YY = {0}. Hece y = y ad z = z. This proves the uiqueess Remar. From () a bove we ca defie a mappig ( called the orthogoal projectio of H oto Y ) P: H Y defied by Px = y where x = y + z yy ad zz. This mappig has the followig properties: () P is liear ad bouded. () P maps H oto Y. (3) P(Y) = Y. (4) P(Y ) = {0}. (5) P = P ad the restrictio of P o Y is the idetity operator o Y. 3.3-5 Lemma. The orthogoal complemet Y of a closed subspace Y of a Hilbert space H is the ull space (P) of the orthogoal projectio P of H oto Y. Remar 3. If M is a oempty subset of a ier product space X the M is a closed vector subspace of X ad M is a subset of M. 3.3-6 Lemma. If Y is a closed subspace of a Hilbert space H the Y = Y. Proof. By Remar 3 Y Y. Coversely let xy. By Theorem 3.3-4 there is a uique yy such that x = y + z. However Y Y ad Y is a vector space the z = x y Y that is z Y. Sice Y is a closed subspace of a Hilbert space the it is complete ad so by Lemma 3.3- z Y. Hece z z ad so z = 0. So that x = y ad xy. Therefore Y Y. Hece Y = Y Remar 4. If Y is a closed subspace of a Hilbert space H ad Z =Y the Z =Y = Y H = Z + Z ad P z x = z defies a projectio P z :HZ. 6

3.3-7 Lemma. For ay o empty subset M of a Hilbert space H the spa of M is dese if ad oly if M = {0}. Proof. Let xm ad assume that V = spa M to be dese i H. The x V = H the there is a sequece (x ) of elemets i V such that x x. Sice xm the for all mm < x m > = 0 which implies that < x v > = 0 for all vv i particular < x x > = 0 for all. By the cotiuity of the ier product < x x >< x x >. Hece < x x > = 0 ad so x =0. Therefore M = {0}. Coversely suppose that M = {0}. Sice M V the V M = {0}. Hece V = {0}. By Theorem 3.3-4 H = V V = V {0} = V. Hece V is dese i H H. W. 6-0. H.W.* 8 Remars 3 ad Lemma 3.3-5. 7

3.4 Orthoormal Sets ad Sequeces 3.4- Defiitio. A orthogoal set M i a ier product space X is a subset M of X whose elemets are pairwise orthogoal. A orthoormal subset M of X is a orthogoal set i X whose elemets have orm. That is 0 if x y for all x ym < x y > =. if x = y otes. a) If a orthogoal or a orthoormal set M is coutable we ca arrage it i a sequece (x ) ad call it a orthogoal or a orthoormal sequece respectively. b) A family (x ) I is called orthogoal if x x for all I ad. The family is orthoormal if it is orthogoal ad all x have orm 0 if so that for all I < x x > =. if = Remar. For a orthogoal elemets x ad y we have x + y = x + y ( Pythagorea Relatio ). I geeral if {x x. x } is a orthogoal set the x + x +. + x = x + x +. + x. 3.4- Lemma. A orthoormal set is liearly idepedet. Examples: 3.4-3 Euclidea space R. The stadard basis of R forms a orthoormal set. ( How?) 3.4-4 Hilbert sequece space orthoormal sequece i. The Schauder basis (e ) of. ( How?) forms a 3.4-5 Cotiuous fuctios. Let X be the ier product space of all cotiuous real-valued fuctios o [0 ] with < x y > = x ( t ) y ( t ) dt. The (cost ) = 0.. ad ( sit ) are orthogoal sequeces. Moreover ( cost cos t si t. ) ad ( ) are orthoormal sequeces. 0 8

3.4-6 Theorem ( Bessel's Iequality). Let (e ) be a orthoormal sequece i a ier product space X. The for every xx xe x. Proof. Cosider the fiite subset { e e. e } from ( e ). The 0 x x ei ei = < x i x < x x ei ei x i x e e > = x e e > < x ei ei x > + < i x e e > = x x e x e i i x ei ei i x ei ei x + x ei x e ei e = x x e x e + x e x e = x By lettig we get xe. Hece xe x xe x. otes. a) The ier product < x e > above is called the Fourier coefficiets of x with respect to the orthoormal sequece (e ). b) If dim(x) is fiite the ay orthoormal set i X must be fiite because it is liearly idepedet. Gram-Schmidit process for orthoormalizig a liearly idepedet sequece (xj) i a ier product space X. x x st step. The first elemet of (e ) is e = d step. Let v = x < x e > e. Sice (x j ) is liearly idepedet the v 0. Also v e where < v e > = < x e > < x e >< e e > = 0. So we ca tae e = v. v 3 rd step. Let v 3 = x 3 < x 3 e > e < x 3 e > e. As above v 3 0 v 3 e ad v 3 e (how?). So we ca tae e 3 = v 3. th step. Let v = x x e e 3. v. As above v 0 ad v e i for all i = -. So we ca tae e =. Therefore we have (e ) as a v orthoormal sequece. v H. W. 3 5 7-0. H.W.* 8 0. 9

3.5 Series Related to Orthoormal Sets ad Sequeces 3.5- Theorem. Let (e ) be a orthoormal sequece i a Hilbert space H. The for ay scalars 3. a) b) If e coverges ( i the orm of H ) if ad oly if coverges. e coverges the 's are the Fourier coefficiets < x e > where x deotes the sum x = c) For ay xh e ad so x e e coverges. e = x = x e e. Proof. a) Let S = e + e +.+ e ad σ = + +.+. Sice (e ) is orthoormal the for > m S S m = m+ e m+ + m+ e m+ +.+ e = < m+ e m+ + m+ e m+ +.+ e m+ e m+ + m+ e m+ +.+ e > = m+ e m+ + m+ e m+ +.+ e = m+ + m+ +.+ = σ σ m. Therefore (S ) is Cauchy i H if ad oly if (σ ) is Cauchy i R. However both H ad R are complete the (S ) coverges if ad oly if (σ ) coverges. Hece e coverges (i the orm of H) if ad oly if b) Let S x; that is x = e coverges i R.. Let be fixed the for a fixed j = ad we have < S e j > = < e + e +.+ e e j > = < j e j e j > = j. By the cotiuity of the ier product we have j = <S e j ><x e j > as. The we ca tae ( ) as large as we please. Hece j = lim = < x e j > for all j = 3. Therefore x = b) Let x be ay elemet i H. By Bessel s iequality Hece xe coverges ad so by a) e = x e e. xe x. x e e coverges 3.5- Lemma. Ay x i a ier product space X ca have at most accoutably may o zero Fourier coefficiets < x e > with respect to a orthoormal family (e ) I i X. Remar. a) The Bessel iequality holds i the case of (e ) I as a orthoormal family that it is the parseval relatio. H.W.* 4. xe x. If the equality holds we say j 0

3.6 Total Orthoormal Sets ad Sequeces 3.6- Defiitio. A total set (fudametal set) i a ormed space X is a subset M of X whose spa is dese i X. A total orthoormal set i X is a orthoormal set which is total. Remar. a) I every Hilbert space H {0} there exists a total orthoormal set. b) All total orthoormal sets i a give Hilbert space H {0} have the same cardiality where the cardiality of the total orthoormal set i a Hilbert space H {0} is called the Hilbert dimesio or orthogoal dimesio of H. If H = {0} the Hilbert dimesio is defied to be zero. 3.6- Theorem. Let M be a subset of a ier product space X. The a) If M is total i X the xx ad x M implies x = 0. b) If X is complete ad if the coditio ( xx ad x M implies x = 0 ) is satisfied the M is total i X. Proof. a) Let H be the completio of X. The X ca be regarded as a subspace of H which is dese i H. However M is total i X the spam is dese i X the it is dese i H. Hece by Lemma3.3-7 M = {0}. Therefore xx ad x M implies x = 0. b) Use Lemma3.3-7 ad the defiitio of total set to get the result 3.6-3 Theorem. A orthoormal set M i a Hilbert space H is total i H if ad oly if xe = x for all xh. Proof. If M is ot total the by Theorem3.6- b) there is xh x 0 ad x M. The < x e > = 0 for all e M. Hece xe = 0 x. Therefore if xe = x for all xh the M is total i H. Coversely suppose that M is total i H. Let x be ay elemet i H. ad arrage all its ozero Fourier coefficiets i a sequece < x e > < x e >. or writte i some defiite order if there are oly fiitely may of them. Defie y by y = x e e () Sice M is orthoormal the for every e j occurig i () we have < x y e j > = < x e j > < y e j > = < x e j > x e e ej = < x e j > < x e j > = 0. But for all vm ot cotaied i () we have < x v > = 0. So that < x y v > = < x v > x e e v = 0. Hece x y M. However M is total i H the by Lemma3.3-7 M = {0} ad so x y = 0 that is x = y. Therefore x = < y y > = < x e e = x e x em e em = m xe x em em > m

3.6-4 Theorem. Let H be a Hilbert space. The a) If H is separable the every orthoormal set i H is coutable. b) If H cotais a orthoormal sequece which is total i H the H is separable. Proof. a) Let H be separable B ay dese set i H ad M ay orthoormal set. Sice M is orthoormal the for ay x ym with x y we have x y = < x y x y > = < x x > + < y y > =. Hece spherical eighborhoods x of x ad y of y of radius 3 are disjoi ( why?). Sice B is dese i H the for ay xm x B. Hece there is a x B ad b y B. Therefore a b. If M is ucoutable the we have uaccoutably may pair wise disjoit spherical eiborhoods so that B would be ucoutable. Sice B was ay dese set this maes that H would ot cotai a coutable dese set which cotradicts the separability of H. Therefore M must be coutable. a) Let (e m ) be a total orthoormal sequece i H ad A = { ( ) e : = ( ) ( ) a + i ( ) b ( ) a ( ) b Q }. ( b = 0 whe H is real ). A is coutable (how?). We show that A is dese i H. Let x be ay fixed elemet i H. Sice (e m ) is total i H the spa( e m) = H. The for every > 0 there is wspa(e m ) such that x w <. Hece wy = spa{e e e } for some. By Lemma3.3- there is yy such that x y Y ad x y x w <. By (8a&b) i 3.4[ see the text boo] y ca be writte as y = x e e.the x dese i R the for ay < x e > there is such that ( ) [ x e ] e ( ) ( ) x e e <. Sice Q is = ( ) a + ib ( ) <. Hece there is v = that satisfies x v = x ( ) x e e e < dese i H + ( ) e x a ( ) b ( ) e ( ) Q A x e e + =. Hece v B(x; ) A. Thus A is

3.6-5 Theorem. Two Hilbert spaces H ad H both real or both complex are isomorphic if ad oly if they have the same Hilbert dimesio. Proof. Suppose that H is isomorphic with H the there is a bijective liear mappig T :H H that satisfies < Tx Ty > = < x y > for all x yh. Hece orthoormal elemets i H have orthoormal images uder T. However T is bijective the T maps every total orthoormal set i H oto a total orthoormal set i Hilbert dimesio. Coversly suppose that H ad H ( how?) Therefore H ad H have the same H have the same Hilbert dimesio. The case that H = {0} ad H = {0} is trivial. Let H {0} the H {0} ad ay total; orthoormal sets M i H ad M i H have the same cardiality. So we ca idex them by the same idex set {} ad write M = (e ) ad M = ( e ). ow defie T:H H by Tx= e x e. This is well defied because for all xh we have x = x e e ad by Bessel s iequality xe coverges. The by Theorem 3.5- e x e coverges so that Tx H. Let x = x e e ad y = y e e be ay elemets i H ad ay scalar T(x+ y) = x y e e = x e e + y e e = Tx +Ty. Hece T is liear. Sice ( e ) is orthoormal the < Tx Tx > = < x e e x e e > = m xe = x. For ay x y H (if H is real) < Tx Ty > = 4 x e x em e em = ( Tx + Ty - Tx Ty ) = ( T(x + y) 4 - T(x y) ) = ( x + y 4 - x y ) = < x y >. Similarly for the complex case. Hece T preserves the ier product which implies that T is -. Give ay x = By Bessel's iequality e H. coverges (how?) ad so e is a fiite sum or a series which coverges to xh by Theorem3.5- ad = <x e > by the same theorem. Hece x = e x e = Tx. Thus T is oto. Therefore T is a isomorphism so H ad H are isomorphic H. W. 6 7 9 0. H.W.* 6 0. 3

3.8 Represetatio of Fuctioals o Hilbert Spaces 3.8- Riesz's Theorem. Every bouded liear fuctioal f o a Hilbert space H ca be represeted i terms of ier product amely f(x) = < x z > where z depeds o f is uiquely determied by f with orm z = f. Proof. The case f = 0 is a trivial case (why?). Let f 0. The (f) H. However (f) is a closed subspace of H the by Theorem 3.3-4 H = (f) (f). Hece (f) {0}. Let z 0 0 ad z 0 (f). Set v = z 0 f(x) xf(z 0 ) where x is arbitrary elemet i H. Sice f is liear the f(v) = f(z 0 )f(x) f(x)f(z 0 ) = 0 ad so v(f). Hece z 0 v ad 0 = < v z 0 > = < z 0 f(x) xf(z 0 ) z 0 > = f(x)< z 0 z 0 > f(z 0 ) < x z 0 >. The f(x) = f ( z0 ) z z < x z 0 > = < x 0 0 the by taig z = f ( z0 ) z z 0 0 f ( z0 ) z z 0 0 z 0 we get f(x) = < x z >. z 0 >. Sice x was arbitrary To prove the uiqueess of z. Suppose that there are z ad z such that f(x) = < x z > = < x z > for all xh. The < x z - z > = 0 for all xh ad for x = z - z we have < z - z z - z > = 0. Hece z = z ad the uiqueess is proved 3.8- Lemma. If < v w > = < v w > for all w i a ier product space X the v = v. I particular < v w > = 0 for all wx implies that v = 0. 3.8-3 Defiitio. Let X ad Y be vector spaces over a field. A sesquiliear form (sesquiliear fuctioal) h o X Y is a mappig h : X Y such that for all x x x X y y y Y ad all scalars it satisfies a) h(x + x y) = h(x y) + h(x y) b) h(x y + y ) = h(x y ) + h(x y ) c) h(x y) = h(x y) d) h(x y) = h(x y). otes. ) If = R the h above is biliear. ) If X ad Y are ormed spaces ad there is c > 0 such that for all xx ad all yy h(x y) c x y the h is bouded ad the umber h = sup{ h ( x y ) x y : xx\{0} & yy\{0} } = sup{h(x y): x = y = } is called the orm of h the h(x y) h x y for all x y. 3) The ier product is sesquiliear ad bouded (how?) 4

3.8-4 Theorem (Riesz's Represetatio). Let H ad H be Hilbert spaces over a field ad h : H H a bouded sesquiliear form. The h has a represetatio h(x y) = < Sx y > where S : H H is a bouded liear operator. Moreover S is uiquely determied by h ad has the orm S = h. Proof. Cosider h ( x y ) which is liear i y. ow for ay fixed xh we apply Theorem 3.8- to get a uique zh that depeds o xh such that h( x y ) = < y z > for all yh ad so h(x y) = < z y >. Hece S : H H give by Sx = z is a operator ad h(x y) = < Sx y >. For all x x H y H ad all scalars < S(x + x ) y > = h(x + x y) = h(x y) + h(x y) = <Sx y > +< Sx y > = <Sx + Sx y >. The by Lemma3.8- S(x + x ) = Sx +Sx so S is liear. If S=0 the it is bouded. If S 0 the Sx y Sx Sx h = sup{ : xh \{0} & yh \{0}} sup{ : xh \{0}} = sup{ Sx x y x Sx x : xh \{0}} = S. Hece S is bouded ad h S. Sx y Sx y But h = sup{ x y : xh \{0} & yh \{0}} sup{ x y : xh \{0} & yh \{0}} = S. Therefore h = S. To prove the uiqueess Suppose that there is a liear operator T : H H such that h(x y) = <Sx y > = <Tx y > for all xh ad yh. The by Lemma3.8- Sx = Tx for all xh. This proves the uiqueess of S H. W.-35 6 0 3. H.W.* 0. 5

3.9 Hilbert adjoit operator 3.9- Defiitio. Let H ad H be two Hilbert spaces ad T:H H a bouded liear operator. The Hilbert adjoit operator T* of T is the operator T*:H H such that for all xh ad all yh < Tx y > = < x T*y >. 3.9- Theorem. The Hilbert adjoit operator T* of T i Defiitio 3.9- exists is uique is liear ad is bouded with orm T* = T. Proof. Cosider h(y x) = < y Tx > which defies a sesquiliear form o H H ( The details is left to the reader ). By Schwarz iequality h(y x) = < y Tx > y Tx y x T this implies that h is bouded ad h T. But h = sup{ h ( y x ) y x : yh \{0} & xh \{0} } = sup{ y Tx : yh \{0} & xh \{0}} sup{ Tx Tx y x Tx x : xh \{0}} = T. Therefore h = T. Sice h is a bouded sesquiliear form the by Theorem3.8-4 there exists a bouded liear operator call it T* that is uiquely determied by h ad h(y x) = < T*y x > with T* = h. Hece T* = T. However h(y x) = <y Tx> the < y Tx > = < T*y x > ad so < Tx y > = < x T*y > 3.9-3 Lemma. Let X ad Y be ier product spaces ad Q:XY a bouded liear operator. The a) Q = 0 if ad oly if < Qx y > = 0 for all x X ad all yy. b) If Q:XX X is complex ad < Qx x > = 0 for all xx the Q =0. Proof. a) Left to the reader. b)from assumptio < Qv v > = 0 for all v = x + yx; that is 0 = < Q(x + y) x + y > = < Qx x > + < Qy y > + < Qx y > + < Qy x > = < Qx y > + < Qy x >. Put = ad the = i to get < Qx y > + < Qy x > = 0 ad < Qx y > - < Qy x > = 0 respectively. By additio < Qx y > = 0 ad so Q = 0 follows from a) Remar. If X is real the b) above eed ot holds. To see this cosider the mappig Q : R R give by Q( ) = ( -). It is clear that for ay x = ( )R < Qx x > = ( -).( ) = 0 but Q 0 3.9-4 Theorem. Let H ad H be Hilbert spaces S T:H H bouded liear operators ad ay scalar. The we have a) < T*y x > = < y Tx > for all xh ad yh. b) (S +T)* = S* + T* ad (T )* = T*. c) ( T* )* = T ad i the case H = H (ST)* = T*S*. d) T*T = TT* = T. e) T*T = 0 if ad oly if T = 0. Proof. We prove d) ad left the proof of the other parts to the reader. First ote that T*T:H H but TT*:H H. The Tx = <Tx Tx> = < T*Tx x > T*Tx x T*T x. The T = sup{ Tx : x = xh } T*T T* T = T T = T. Hece T*T = T. Replacig T by T* to get T**T* = T*. However T**T* = TT* ad T = T* Therefore T*T = TT* = T H. W.-8. H.W.* 4 6. 6

3.0 Self- adjoit Uitary ad ormal Ooperators 3.0- Defiitio. A bouded liear operator T : H H o a Hilbert space H is said to be ) self-adjoit or Hermitia if T* = T. ) Uitary if T is bijective ad T* = T -. 3) ormal if TT* = T*T. 3.0- Remar. a) If T is self-adjoit the < Tx y > = < x Ty > for all x ad y. b) If T is self-adjoit or uitary the T is ormal. c) ormal operators eed ot be self-adjoi or uitary. To see this cosider the idetity operator I : H H o a Hilbert space H. It is easy to see that T = ii is ormal but it is either self-adjoit or uitary. 3.0-3 Theorem. Let T : H H be a bouded liear operator o a Hilbert space H. The a) If T is self-adjoit the < Tx x > is real for all xh. b) If H is complex ad <Tx x > is real for all xh the T is self-adjoit. Proof. a) Suppose that T is self-adjoit. The for all xh Tx x = < x Tx > = <Tx x >. Hece <Tx x > is real. b) Suppose that H is complex ad <Tx x > is real for all xh. The < Tx x > = Tx x = x T * x = < T*x x > ad 0 = < Tx x > - < T*x x > = < (T-T*)x x >. The by Lemma3.9-3(b) T-T* = 0. Hece T is self-adjoit 3.0-4 Theorem. The product of two bouded self-adjoit liear operators S ad T o a Hilbert space H is self adjoit if ad oly if ST = TS. 3.0-5 Theorem. Let ( T ) be a sequece of bouded self-adjoit liear operators T : H H o a Hilbert space H. Suppose that T T that is T - T 0 where this orm is the orm o the space B(H H). The T is a bouded self-adjoit liear operator o H. 3.0-6 Theorem. Let the operators U V : H H be uitary o a Hilbert space H. The : a) U is isometric thus Ux = x for all xh. b) U = provided that H {0}. c) U - ad UV are uitary. d) U is ormal. Furthermore c) A bouded liear operator T o a complex Hilbert space H is uitary if ad oly if T is isometric ad oto. 3.0-7 Remar. Isometric operators eed ot be uitary. Proof. Cosider the bouded liear operator T: give by T(.) = (0.). It is easy to see that T is a isometric but T is ot oto where there is (.) ad 0 but there is o x with Tx = (.). Hece we have a isometric which is ot uitary H. W.-4 8-5. H.W.* 4 4. 7