The extreme points of symmetric norms on R^2

Size: px
Start display at page:

Download "The extreme points of symmetric norms on R^2"

Transcription

1 Graduate Theses and Dissertations Iowa State University Capstones, Theses and Dissertations 2008 The extreme points of symmetric norms on R^2 Anchalee Khemphet Iowa State University Follow this and additional works at: Part of the Mathematics Commons Recommended Citation Khemphet, Anchalee, "The extreme points of symmetric norms on R^2" (2008). Graduate Theses and Dissertations This Thesis is brought to you for free and open access by the Iowa State University Capstones, Theses and Dissertations at Iowa State University Digital Repository. It has been accepted for inclusion in Graduate Theses and Dissertations by an authorized administrator of Iowa State University Digital Repository. For more information, please contact

2 The extreme points of symmetric norms on R 2 by Anchalee Khemphet A thesis submitted to the graduate faculty in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Major: Mathematics Program of Study Committee: Justin Peters, Major Professor Krishna Athreya Paul Sacks Iowa State University Ames, Iowa 2008 Copyright c Anchalee Khemphet, All rights reserved.

3 ii DEDICATION I would like to dedicate this thesis to my father Pleng and to my mother Jumrus without whose support I would not have been able to complete this work. I would also like to thank my friends Chayanon Devin Jyy-I and Natsima for their loving guidance and caring assistance during the writing of this work.

4 iii TABLE OF CONTENTS LIST OF FIGURES v ACKNOWLEDGEMENTS ABSTRACT vi vii CHAPTER 1. OVERVIEW Introduction Definitions and examples Definition of a norm Example of norms Definition of a convex set Remark Definition of equivalence of norms Equivalence of norms on R n Norms and unit balls on R n Theorem Theorem CHAPTER 2. NORMS ON R Norm space of norms on R Extreme points of norms on R Definition of an extreme point Example of an extreme point of C Example of a non-extreme point of C

5 iv 2.3 Compactness on C Definition of uniform equivalence on a set of norms Uniform equivalence on C Definition of equicontinuity Theorem Equicontinuity on C Arzela-Ascoli Theorem CHAPTER 3. SYMMETRIC NORMS ON R Properties of symmetric norms on R Definition of invariance Invariance under the dihedral group Extreme points of C G Compactness of C G Extreme points of a compact convex set Krein-Milman Theorem Theorem CHAPTER 4. EXTREME POINTS OF SYMMETRIC NORMS ON R Symmetric norms with piecewise linear boundaries Lemma Theorem Theorem CHAPTER 5. SUMMARY AND DISCUSSION APPENDIX. KNOWN THEOREMS BIBLIOGRAPHY

6 v LIST OF FIGURES Figure 1.1 The unit balls of l 1, l 2 and l on R Figure 2.1 The unit balls of p 1, p 2 and l Figure 2.2 The unit balls of l 1, l ɛ l1 and p Figure 2.3 The unit ball of p n Figure 3.1 The unit ball of p t Figure 4.1 The unit ball of a norm with piecewise linear boundary Figure 4.2 The unit ball of a norm with piecewise linear boundary

7 vi ACKNOWLEDGEMENTS I would like to take this opportunity to express my thanks to those who helped me with various aspects of conducting research and the writing of this thesis. First and foremost, Dr. Justin Peters for his guidance, patience and support throughout this research and the writing of this thesis. His insights and words of encouragement have often inspired me and renewed my hopes for completing my graduate education. I would also like to thank my committee members for their efforts and contributions to this work: Dr. Krishna Athreya and Dr. Paul Sacks. I would additionally like to thank Mrs. Erickson for her guidance throughout the initial stages of my graduate career.

8 vii ABSTRACT Let C G be a set of symmetric norms on R 2 such that p(0, 1) = p(1, 0) = 1 for any norm p in C G. Our questions are: Do the extreme points of C G exist? What are all the extreme points of C G? To answer these questions, compactness plays an important role in completing the sufficient conditions of known theorems. Results show that C G has extreme points and all its extreme points are found.

9 1 CHAPTER 1. OVERVIEW Let C G be a set of norms on R 2 such that for any norm p in C G, p(1, 0) = p(0, 1) = 1 and p is G-invariant where G is the dihedral group generated by a counterclockwise rotation α by an angle of 90 and a reflection β in the y-axis. By the Arzela-Ascoli Theorem, C G is a nonempty compact convex set. Consequently, the Krien-Milman Theorem guarantees that this set has an extreme point and the closed convex hull of its extreme points is itself. For example, the minimum norm l 1 (x, y) = x + y and the maximum norm l (x, y) = max { x, y } are both extreme points of C G. Let y x, if x p t (x, y) = t ; x+y 1+t, if t < y x 1, where 0 t 1. We have that p t is a norm in C G. Notice that p 0 = l 1 and p 1 = l. Consider that {p t : 0 t 1} is a closed set. Moreover, the closed convex hull of {p t : 0 t 1} is a subset of C G since C G is a closed convex set. Conversely, if C G is a subset of the closed convex hull of {p t : 0 t 1}, then the following known theorem provides that {p t : 0 t 1} is the only extreme points of C G. Theorem Let X be a norm space and K a compact convex subset of X. If F is a closed subset of extreme points of K such that co(f ) = K, then F = ext(k). To show that C G is a subset of the closed convex hull of {p t : 0 t 1}, we have the next theorems as our main results.

10 2 Theorem The convex hull of {p t : 0 t 1} is the set of all norms in C G with piecewise linear boundaries. Theorem Any p C G can be approximated by norms in C G with piecewise linear boundaries. From these theorems, if p C G, then there is a sequence p n in C G such that p n converges to p where each p n is a convex combination of {p t : 0 t 1}. Thus, C G is contained in the closed convex hull of {p t : 0 t 1}. Hence, we complete our expectation. To emphasize, the set {p t : 0 t 1} is the set of all extreme points of C G. 1.1 Introduction In this paper, we begin by recalling the definition of a norm, together with some examples and basic properties. If p and q are norms, then r = α p + β q is also a norm for α, β > 0. We consider a convex combination of norms, a relationship between the unit balls and also a convex combination of the unit balls. Next, we restrict our attention to the set of norms on R 2 with p(1, 0) = p(0, 1) = 1. To find its extreme points, the behavior of this set is studied. Moreover, we determine whether some specific norms are extreme points of the set. Furthermore, the invariance under the dihedral group is added to the set of such norms, namely, symmetric norms. With this new set, some extreme points are found. We claim that these extreme points are, in fact, the only extreme points of symmetric norms. Some known theorems are raised in order to support the proof of the claim. Finally, we construct our main theorems to finish the claim. One of the interesting points of the new theorems is that whatever norm we pick from this set, we can approximate this norm by convex combinations of these extreme points.

11 3 1.2 Definitions and examples Definition of a norm Let V be a vector space over R. A function p : V R, is called a norm if for v, w V i) p(v) 0 and p(v) = 0 if and only if v = 0 ; ii) p(α v) = α p(v) for α R ; iii) p(v + w) p(v) + p(w). The set {v V : p(v) < 1}, is called the unit ball B p Example of norms Figure 1.1 The unit balls of l 1, l 2 and l on R 2 Let u = (u 1, u 2,..., u n ) R n, define l r (u) = ( u 1 r + u 2 r u n r ) 1 r for r 1. Note that l r satisfies all conditions above, so it is a norm. More specifically, l 1 (u) = u 1 + u u n is called the minimum norm, l 2 (u) = ( u u u n 2) 1 2 is called the Euclidean norm, and l (u) = max { u 1, u 2,..., u n } is called the maximum norm.

12 Definition of a convex set Let V be a vector space over R. A subset U of V is called a convex set if for any v, w U, λ v + (1 λ) w U, 0 < λ < 1. Let W be any subset of V. For w 1, w 2,..., w n W, a convex combination of these elements is a linear combination of the form λ i w i for λ i > 0 and λ i = 1. Then the convex hull of W is the set of all convex combinations from W, denoted by co(w ). In fact, this is the smallest convex set containing W. Similarly, the closed convex hull of W is the smallest closed convex set containing W, denoted by co(w ). Note that the set of norms on a vector space over R is convex Remark Let p, q be norms on a vector space over R. i) p q if and only if B q B p. ii) B α p = 1 α B p for α > 0. iii) If either p(v) q(v) for all v B p or p(v) q(v) for all v B q, then p q Definition of equivalence of norms Let V be a vector space over R and let p, q be norms on V. Then p, q is said to be equivalent if there exist m, M > 0 such that m p q M p. The equivalence of norms is an equivalence relation. i) Reflexive: It is easy to see that p p. ii) Symmetric: If p q, then m p q M p for some m, M > 0. Thus, 1 M q p 1 m q. That is, q p. iii) Transitive: Assume that p q and q r. Then, m 1 p q M 1 p and m 2 q r M 2 q for some m 1, m 2, M 1, M 2 > 0. Therefore, m 2 m 1 p m 2 q r M 2 q M 2 M 1 p. That is, p r.

13 Equivalence of norms on R n Theorem Any two norms on R n are equivalent. That is, let p, q be norms on R n. Then, there exist m, M > 0 such that m p(v) q(v) M p(v) for v R n. This result is a well-known theorem in functional analysis. The proof is provided in the Appendix. Consequently, we have that the sets of open sets with respect to norms on R n are all the same. Notice that for any norm p on R n, we can induce the metric d by d(u, v) = p(u v) for any u, v R n. Let p, q be norms on R n. Let A be an open set with respect to the norm p. Let v A. Then, there is a δ > 0, B p (v, δ) = {u R n : p(u v) < δ} A. Choose γ = m δ > 0. Then, B q (v, γ) = {u R n : q(u v) < γ}. Note that p(u v) 1 m q(u v). Thus, v B q (v, γ) B p (v, δ) A. Therefore, A is also an open set with respect to q. Conversely, if B is an open set with respect to the norm q, with the equivalent relation similar to above, then B is an open set with respect to p as well. 1.3 Norms and unit balls on R n For any subset B of R n, one could ask whether there exists a norm p on R n such that B p = B. Consider any triangle T as a subset of R 2. Let l be one of the sides of T. If p is a norm with a unit ball T, then p = 1 on l and so p = 1 on l. Thus, l is also a side of T, which is impossible. The following theorem gives sufficient conditions to be a unit ball Theorem Let p be a norm on R n, B p = {u R n : p(u) < 1}. With respect to the Euclidean norm l n, we have that i) B p is convex ; ii) B p is open ; iii) B p is bounded ;

14 6 iv) B p = B p. Conversely, let B R n. If B is nonempty and has properties i)-iv), then B is the unit ball of some norm p. Proof. For any norm p on R n, we will show that the unit ball B p has properties i)-iv). i) Let u, v B p and 0 < λ < 1. Then p(λ u + (1 λ) v) λ p(u) + (1 λ) p(v) < 1. Thus, λ u + (1 λ) v B p. ii) We will show that B p is open with respect to the norm p. Let v B p. Then, p(v) < 1. Choose δ = 1 p(v) 2 > 0. Claim that B p (v, δ) B p. Let w B p (v, δ). Then p(v w) < δ. Therefore, p(w) p(w v) + p(v) < δ + p(v) = p(v) < 1. Thus, we have the claim. By 1.2.6, the equivalence of norms on R n, B p is also open with respect to the Euclidean norm l n. iii) Since any two norms on R n are equivalent from 1.2.6, there exist m, M > 0 such that m l n p M l n. Then, l n (u) 1 m p(u) < 1 m for any u B p. iv) u B p p(u) < 1 p( u) < 1 u B p u B p. Conversely, let B be a nonempty subset of R n satisfying properties i)- vi). Define p : R n R + {0} by p(u) = inf{α R + {0} : u α B} for u R n. Then, p is well-defined since B is bounded. We will show that p is a norm. Clearly, p(u) 0 for all u R n. If u B, then, by vi), u B and so, by i), 0 = 1 2 u ( u) B. Note that p(u) = 0 if and only if u = 0. If u = 0, then u 0 B and so p(u) = 0. If p(u) = 0, then p(u) ɛ for any ɛ > 0, i.e. u ɛ B and so u = ( ) 0. Let ɛ > 0. For β R, consider that u (p(u) + ɛ) B. Then u 1 β p(β u) + ɛ B. ( ) Thus, p(u) 1 β p(β u) + ɛ. Therefore, p(u) 1 β p(β u), i.e. β p(u) p(β u). Similarly, β u (p(β u) + ɛ) B implies β u ( β p(u) + ɛ) B. Therefore, p(β u) ( β p(u) + ɛ). Then, p(β u) β p(u). Hence, p(βu) = β p(u) for any β R. Note that (a + b) B a B + b B for ( ) a, b > 0. For x, y B, a x + b y = (a + b) a a+b x + b a+b y (a + b) B since B is convex. Thus, (a + b) B = a B + b B for a, b > 0. By definition of p, u ( p(u) + ɛ 2) B and v ( p(v) + ɛ 2) B.

15 7 Then, u + v ( p(u) + ɛ 2) B + ( p(v) + ɛ 2) B, that is, u + v (p(u) + p(v) + ɛ) B. Therefore, p(u + v) p(u) + p(v) + ɛ, i.e. p(u + v) p(u) + p(v). Hence, p is a norm. Next, we will show that B p = B. If u B p, then p(u) < 1 and so u B. Conversely, let u B. Then p(u) 1. Suppose p(u) = 1. Since B is open, there exists δ > 0 such that B p (u, δ) B. Notice that p ( u ( 1 + 2) δ ) ( u = p δ 2 u) = δ 2 < δ. Thus, ( 1 + 2) δ u Bp (u, δ) and so ( 1 + 2) δ (( ) ) u B. This is a contradiction since p 1 + δ 2 u = 1 + δ 2 > 1. Therefore, p(u) < 1, that is, u B p. Hence, B p = B. Notice that a convex combination of any two norms is also a norm. Specifically, if p and q are norms on R n, 0 < λ < 1, then so is r = λ p + (1 λ) q. How does its unit ball compare with those of p and q? The answer is shown in the next theorem Theorem If p, q are distinct norms on R n, 0 < λ < 1, let r = λ p+(1 λ) q, then B r λ B p +(1 λ) B q. Proof. Let v B r, i.e., r(v) 1 and assume that v 0. Then p(α v) = 1 and q(β v) = 1 for some α, β > 0. Thus, p(v) = 1 α and q(v) = 1 β. For 0 < λ < 1, let w = λ (α v) + (1 λ) (β v) λ B p + (1 λ) B q and γ = λ α + (1 λ) β > 0. Then w = γ v. By the inequality of arithmetic and geometric mean, λ a + (1 λ) b a λ b 1 λ for a, b > 0. Then, r(w) = λ p(w) + (1 λ) q(w) = λ (γ p(v)) + (1 λ) (γ q(v)) = λ γ α + (1 λ) γ β = = ( γ ) ( ) λ γ 1 λ α β γ α λ β 1 λ λ α + (1 λ) β α λ β 1 λ αλ β 1 λ α λ β 1 λ = 1.

16 8 Since r(v) 1 and r(γ v) = r(w) 1, γ 1 and so 1 γ 1. Therefore, v = 1 γ w λ B p + (1 λ) B q. Moreover, observe that if α β, then r(w) > 1. Hence, B r λ B p + (1 λ) B q.

17 9 CHAPTER 2. NORMS ON R 2 In this chapter, we will consider a norm space of norms on R 2 and then restrict our attention to the set of norms on R 2 where (0, 1), (1, 0) are on the boundary of its unit ball. 2.1 Norm space of norms on R 2 We will show that any norm on R 2 is a continuous function. Let p be any norm on R 2. Notice that p(α (1, 0)) = 1 and p(β (0, 1)) = 1 for some α, β > 0. Let ɛ > 0 be given. Choose δ = αβ α+β ɛ > 0. If l 2((x, y) (a, b)) < δ, then p(x, y) p(a, b) p(x a, y b) p(x a, 0) + p(0, y b) = x a p(1, 0) + y b p(0, 1) p(α (1, 0)) < δ α = α + β αβ δ = ɛ. + δ β(0, 1) β Thus, p is continuous. Then, p C(R 2, R). Let H = {f C(R 2, R) : f(γ v) = γ f(v) for v R 2, γ R} be the family of homogeneous continuous functions. Define a norm on H by f = sup { f(v) : v R 2, l 2 (v) 1}. Note that {v R 2 : l 2 (v) 1} is closed and bounded with respect to the Euclidean norm l 2. That is, this set is compact. Since any f H is continuous, f <. Notice that every norm on R 2 is in H. Then, we will consider the set of norms on R 2 as a subspace of the norm space H.

18 Extreme points of norms on R 2 Let C be a set of norms on R 2 such that for any p C, p(1, 0) = p(0, 1) = 1. For any p 1, p 2 C, let p = λ p 1 + (1 λ) p 2 for 0 < λ < 1. Then p is a norm in R 2. Note that p(1, 0) = λ p 1 (1, 0) + (1 λ) p 2 (1, 0) = 1. Similarly, p(0, 1) = 1. Therefore, p C. Hence, C is convex Definition of an extreme point Let K be a convex set. A norm p is called an extreme point of K if no such distinct p 1, p 2 K exist such that p = λ p 1 + (1 λ) p 2 for 0 < λ < 1. The set of all extreme points of K is denoted by ext(k) Example of an extreme point of C Recall the norm l r as defined in Example We have that l r belongs to C for each r 1. Note that p(x, y) p(x, 0) + p(0, y) = x + y = l 1 (x, y) for any p C. That is, p l 1 for all p C. We will show that the minimum norm l 1 is an extreme point of C. Suppose that there exist p 1, p 2 C such that l 1 = λ p 1 + (1 λ) p 2 where 0 < λ < 1. For any (x, y) on the boundary of the unit ball of l 1, 1 = l 1 (x, y) = λ p 1 (x, y) + (1 λ) p 2 (x, y). Since p 1, p 2 l 1, if either p 1 (x, y) or p 2 (x, y) is less than 1, then l 1 (x, y) < 1, a contradiction. Therefore, p 1 (x, y) = p 2 (x, y) = 1 for all (x, y) on the boundary of the unit ball of l 1. Hence, l 1 = p 1 = p 2 and so l 1 is an extreme point of C Example of a non-extreme point of C Recall the maximum norm l from Example 1.2.2, we will show that l is not an extreme point of C.

19 11 Figure 2.1 The unit balls of p 1, p 2 and l It suffices to consider only on the set {(x, y) R 2 : x, y 0}. Let p 1, p 2 be norms on R 2 defined by and p 1 (x, y) = p 2 (x, y) = x 1 2 y if y x ; 1 2 x + y if y > x x y if y x ; 1 2 x + y if y > x. for (x, y) R 2. Thus, Hence, l = 1 2 p p (p 1 + p 2 ) = x if y x ; y if y > x.

20 Compactness on C Definition of uniform equivalence on a set of norms Let V be a vector space over R. A set C of norms on V is said to be uniformly equivalent if there exist m, M > 0 such that m p q M p for all p, q C Uniform equivalence on C From 1.2.6, we have that any two norms in C are equivalent. However, we will show that C is not uniformly equivalent. Figure 2.2 The unit balls of l 1, l ɛ l1 and p Since l 1 C, it is enough to show that for any ɛ > 0 we can always find p C such that p ɛ l 1, that is, B p B ɛ l1 = 1 ɛ B l 1. Let ɛ > 0 be given. If ɛ > 1, then q l 1 < ɛ l 1 for all q C. Suppose that 0 < ɛ < 1. Without loss of generality, consider (x, y) R 2 when x, y 0.

21 13 Let p(x, y) = x (1 ɛ) y if y x ; (1 ɛ) x + y if y > x. Then p( 1 ɛ, 1 ɛ ) = 1. Thus, ( 1 ɛ, 1 ɛ ) B p. However, ( 1 ɛ, 1 ɛ ) / B ɛ l 1 since ɛ l 1 ( 1 ɛ, 1 ɛ ) = l 1(1, 1) > Definition of equicontinuity Let F be a collection of functions from R 2 to R. F is said to be equicontinuous on R 2 if for each u R 2, and ɛ > 0, there is a δ > 0 such that if l 2 (u, v) < δ, then f(u) f(v) < ɛ for all f F. Next, we will consider a relationship between uniform equivalence and equicontinuity Theorem Let F be a set of norms on R 2. If F is uniformly equivalent, then it is equicontinuous. Proof. Assume F is uniformly equivalent. Let q F and α = max {q(1, 0), q(0, 1)}. Then there is an M > 0 such that p M q for all p F. Let ɛ > 0 and choose δ = l 2 ((x, y) (a, b)) < δ, then ɛ 2 M α > 0. If p(x, y) p(a, b) p(x a, y b) p(x a, 0) + p(0, y b) = x a p(1, 0) + y b p(0, 1) < δ M q(1, 0) + δ M q(0, 1) 2 δ M α = ɛ for all p F.

22 Equicontinuity on C then Now consider the set C. Let ɛ > 0 be given. Choose δ = ɛ 2 > 0. If l 2((x, y) (a, b)) < δ, p(x, y) p(a, b) p(x a, y b) p(x a, 0) + p(0, y b) = x a p(1, 0) + y b p(0, 1) < 2 δ = ɛ for all p C. Thus, C is equicontinuous. However, we have shown that C is not uniformly equivalent. Hence, the converse of the previous theorem is not necessarily true Arzela-Ascoli Theorem Let (X, d) be a compact metric space. Let B(X, R) be the set of all bounded continuous functions from X to R. For any subset C of B(X, R), C is compact if and only if C is closed, bounded, and equicontinuous. The proof of the Arzela-Ascoli Theorem is provided in the sense of a metric space in the Appendix. Note that the set C is not compact since it is not closed. Let B n be a subset of R 2 with piecewise linear boundary as in the figure 2.3. Then, by Theorem 1.3.1, B n is the unit ball of a norm p n for some p n C. For each n N, p n ( n 2, n 2 ) = 1 and then p n( 1 2, 1 2 ) = 1 n. Thus, p n ( 1 2, 1 2 ) converges to 0 which cannot be a value of any norm on R2 at the point ( 1 2, 1 2 ). Hence, C is not closed. In other words, one can say that the unit ball B n converges to a subset of R 2 which is not a unit ball.

23 15 Figure 2.3 The unit ball of p n

24 16 CHAPTER 3. SYMMETRIC NORMS ON R 2 Next, we will consider the set of norms in C with the invariance under the dihedral group. 3.1 Properties of symmetric norms on R Definition of invariance Let V be a vector space over R and p a norm on V. If α is an automorphism of V, then p is said to be α-invariant if p = p α. If G is a group of automorphisms of V, p is said to be G-invariant if p = p α for every α G. We will show that p α is also a norm. If p α(v) = 0, then α(v) = 0 and so v = 0. Thus, (p α)(v) = 0 if and only if v = 0. Clearly, (p α)(av) = a(p α)(v) for a > 0. Consider (p α)(v + w) = p(α(v) + α(w)) p(α(v)) + p(α(w)) = (p α)(v) + (p α)(w) Invariance under the dihedral group Let α be a counterclockwise rotation by an angle of 90 and β a reflection in the y-axis. The group G generated by α, β is called the dihedral group. Let C G be a set of norms on R 2 such that p(1, 0) = p(0, 1) = 1 and p is invariant under the dihedral group for any p C G. Note that C G is a convex subset of C. Consider a unit ball of any norm in C G. It is symmetric with respect to the x-axis, y-axis, and the lines y = x, and y = x. Without loss of generality, from this point, we will consider the norms in C G restricted to the set A = {(x, y) R 2 : 0 y x}.

25 Extreme points of C G Note that l 1 also belongs to C G. Similar to what we have shown in Example 2.2.2, l 1 is an extreme point of C G. We will show that l is an extreme point of C G. First, suppose that there exists a norm p C G such that l (x, y) = x > 1 = p(x, y) for some (x, y) A. Since p α 2 β = p, p( x, y) = p(x, y) = 1. Then 1 2 (x, y) (x, y) B p since B p is convex. Thus, (x, 0) B p. However, p(x, 0) = x p(1, 0) = x > 1, a contradiction. Hence, l p for all p C G. Assume that l = λ p + (1 λ) q for some p, q C G, 0 < λ < 1. Let (x, y) be any point on the boundary of B l. If either p(x, y) or q(x, y) is greater than 1, then l (x, y) > 1, a contradiction. Therefore, p(x, y) = q(x, y) = 1 for all (x, y) B l. Hence, p = q = l and so l is an extreme point of C G. Figure 3.1 The unit ball of p t For 0 < t < 1, define p t (x, y) = x if 0 y x t ; (3.1) x+y 1+t if t < y x 1. By Theorem 1.3.1, p t is a norm in C G by considering a unit ball of p t. We will show that p t is an extreme point. Assume that p t = λ p + (1 λ) q for some p, q C G.

26 18 Case i) : 0 y x t Consider p t (1, y) = λ p(1, y)+(1 λ) q(1, y) for 0 y t 1. Then p(1, y) l (1, y) 1 and q(1, y) l (1, y) 1. Thus p(1, y) = q(1, y) = 1 since p t (1, y) = 1. That is, p t (1, y) = p(1, y) = q(1, y) for 0 y t 1. Case ii) : t < y x 1 Note that p(1, t) = 1, q(1, t) = 1, and p( 1+t 2, 1+t 2 ) = 1, q( 1+t 2, 1+t 2 ) = 1. Let l be a line segment {λ ( 1+t 2, 1+t 2 ) + (1 λ) (1, t) : 0 λ 1}. Then p t = 1 on l and, by convexity, l lies both B p and B q. Thus, p 1 and q 1 on l. Hence, p = q = 1 on l. From 2.1, we define H = {f C(R 2, R) : f(γ v) = γ f(v) for v R 2, γ R} as the norm space of norms on R 2 where the norm on H defined by f = sup { f(v) : v R 2, l 2 (v) 1}. Let H G be the restriction of H to the set where the function f H is G-invariant. Then, H G is also a norm space fo norms with the invariance under the dihedral group. Next, we will show that {p t : 0 t 1} is a closed set with respect to the norm space H G. Let f : [0, 1] C G be a function defined by f(t) = p t for t [0, 1]. Claim that f is continuous. Let ɛ > 0 be given. Choose δ = ɛ 2 > 0. Let t 1, t 2 [0, 1]. Without loss of generality, assume that 0 t 1 < t 2 1 and 0 < t 1 t 2 < δ. Then Case i) : 0 y x t 1 p t1 (x, y) p t2 (x, y) = x x = 0 < ɛ. Case ii) : t 1 < y x t 2 p t1 (x, y) p t2 (x, y) = x + y 1 + t 1 x y = t 1x 1 + t t 1 xt 2 xt t 1 t 2 t 1 < δ < ɛ.

27 19 Case iii: : t 2 < y x 1 p t1 (x, y) p t2 (x, y) = x + y 1 + t 1 x + y 1 + t 2 = (x + y)(t 2 t 1 ) (1 + t 1 )(1 + t 2 ) 2(t 2 t 1 ) < ɛ. Therefore, p t1 p t2 = sup { p t1 (x, y) p t2 (x, y) : l 2 (x, y) 1} < ɛ. Thus, we have the claim. Then, the image of f is compact. That is, {p t : 0 t 1} is compact. Hence, {p t : 0 t 1} is closed Compactness of C G Recall that l p l 1 for all norms p C G. Let p, q C G and (x, y) R 2. Then max { x, y } p(x, y) x + y and Thus, max { x, y } q(x, y) x + y. x + y max { x, y } + max { x, y } = 2 max { x, y }, i.e. 1 2 ( x + y ) max{ x, y }. Therefore, 1 2 p(x, y) 1 ( x + y ) q(x, y) 2 max { x, y } 2 p(x, y). 2 That is, 1 2 p q 2 p. Therefore, C G is uniformly equivalent. By Theorem 2.1.7, it is equicontinuous. From the Arzela-Ascoli Theorem, we will show that C G is compact. Since we have that C G is equicontinuous, it is enough to show that C G is closed and bounded. Let {p n } be a sequence in C G which converges to p. Clearly, p(1, 0) = p(0, 1) = 1, p α = p, and p β = p. We will show that p is a norm on R 2. Since p n 0 for all n, p 0. For any (x, y) R 2

28 20 such that p(x, y) = 0, if (x, y) 0, then 0 p n (x, y) l (x, y) > 0 for all n. Therefore, p(x, y) l (x, y) > 0, a contradiction. Thus, (x, y) = 0 and, clearly, p(0) = 0. Note that p n (α (x, y)) converges to p(α (x, y)) and α p n (x, y) converges α p(x, y). Since p n (α (x, y)) = α p n (x, y), p(α (x, y)) = α p(x, y). Since 0 p n ((x, y) + (a, b)) p n (x, y) + p n (a, b), 0 p((x, y) + (a, b)) p(x, y) + p(a, b). Thus, p C G. Hence, C G is closed. To show C G is bounded, we need to show that {p(x) : p C G and x B} for some ball B. Let q C G and B = B(0; 1) = B l2. Then sup v B q(v) sup l 1 (v) v B sup l 1 (v) = 2. Hence, C G is compact. v 2B l1 3.2 Extreme points of a compact convex set With compactness of C G, we have the known theorem, the Krien-Milman Theorem, which provides the essential facts Krein-Milman Theorem Let (X, d) be a norm space and let K be any nonempty compact convex subset of X. Then ext(k) and K is the closed convex hull of its extreme points. This theorem implies that the set of the extreme points of C G is nonempty. The examples are shown in the previous sections. Additionally, we know that the set C G is the closed convex hull of the set of the extreme points of C G. However, if we have the set of extreme points such that the closed convex hull of this set is equal to C G, the Krien-Milman Theorem does not guarantee that this set needs to be the only extreme points of C G. To complete this part, we raise another known result Theorem Let X be a norm space and K a compact convex subset of X. If F is a closed subset of extreme points of K such that co(f ) = K, then F = ext(k). The guideline from this theorem is that finding a closed set of extreme points of C G which the closed convex hull of this set is equal to C G.

29 21 CHAPTER 4. EXTREME POINTS OF SYMMETRIC NORMS ON R 2 In this chapter, we will show our main theorems. The statements of these theorems complete the sufficient conditions of being the only extreme points of C G. From 3.1.3, we define the norm space H G = {f C(R 2, R) : f(γ v) = γ f(v) for v R 2, γ R and f is G-invariant} with the norm on H G defined by f = sup { f(v) : v R 2, l 2 (v) 1}. Therefore, we consider the set C G as a subspace of H G. First, we will pay our attention to the norms in C G with piecewise linear boundaries. 4.1 Symmetric norms with piecewise linear boundaries Lemma Figure 4.1 The unit ball of a norm with piecewise linear boundary Let G = {(x, y) R 2 : 0 y x 1}. For (x 1, y 1 ), (x 2, y 2 ), (x 3, y 3 ) G such that

30 22 0 y 1 < y 2 < y 3 x 3 < x 2 < x 1 1, let If < m 1 < m 2 1, then m 1 = y 2 y 1 x 2 x 1 and m 2 = y 3 y 2 x 3 x 2. m 1 1 m 2 1 > 0. (4.1) m 1 x 1 + y 1 m 2 x 2 + y 2 Proof. Note that m 1 x 1 + y 1 = m 1 x 2 + y 2 and that m 2 x 2 + y 2 = m 2 x 3 + y 3. For t R, define a function f : R R by f(t) = t 1 tx 2 + y 2. (4.2) Then f (t) = x 2+y 2 (tx 2 +y 2 ) 2 > 0. Since m 1 > m 2, f( m 1 ) > f( m 2 ). Thus, by (4.2), the inequality (4.1) holds Theorem For 0 < t < 1, let p t be a norm defined in (3.1). Then, co{p t : 0 t 1} is the set of all norms in C G with piecewise linear boundaries. Proof. Without loss of generality, we will consider only (x, y) R 2 such that 0 y x 1. Let p co{p t : 0 t 1}. Then p = λ i p ti and λ i = 1 where 0 λ i 1, i = 1, 2,..., n. Clearly, p C G. Recall that for any 0 < t < 1, y x, if x p t (x, y) = t ; x+y 1+t, if t < y x 1. Assume that 0 = t 1 t 2... t n = 1. For t k < y x t k+1, p(x, y) = k λ i x + y + i=k+1 λ i x, where k = 1,..., n. Claim: k α i = 1 k 1 + t for some 0 < t < 1 and α i = 1.

31 23 We will use induction on k 1. Base: k = 2 Thus, let t = (1 α) t 2 1+α t 2 < 1. α (1 α) + = α + 1 α 1 + t t t 2 = α + α t 2 + (1 α) 1 + t 2 = 1 + α t t 2 1 = = 1+t 2 1+α t 2 Induction step: Assume the equation holds for k 2. k+1 Let α i = 1. Then k+1 for some 0 < t, t < 1. α i = k α i + α k t k+1 = (α 1 + α α k ) = (α 1 + α α k ) = t For t k < y x t k+1, k = 0, 1,..., n, p(x, y) = k λ i x + y + i=k+1 = (λ 0 + λ λ k ) = ( k ) λ i λ i x k x + y 1 + t + i=k (1 α) t. 2 1+α t 2 k α i α 1 +α α k + α k t k t + α k t k+1 λ i λ 0 +λ λ k (x + y) + λ i x i=k+1 λ i x for some 0 < t < 1. Now consider (x, y) R 2 on the boundary of the unit ball of p, i.e,

32 24 p(x, y) = 1. Then ( k λ i 1 + t + i=k+1 λ i ) x + ( k ) λ i 1 + t y = 1 is a linear equation for each t k < y x t k+1, k =, 1,..., n 1. Hence, p is a norm with a piecewise linear boundary. Conversely, let p C G be a norm with piecewise linear boundary. Assume that p l 1, l. Let (x 1, y 1 ),..., (x n, y n ) R 2 be the extreme points on the boundary of the unit ball of p such that 0 < y i x i < 1, i = 1, 2,..., n. Note that the slopes of linear equations on the boundary are less than or equal to -1 and strictly decreasing since the unit ball is convex. Assume that 0 < y 1 < y 2 <... < y n < x n < x n 1 <... < x 1 < 1. Let y 0 = max {y R : p(1, y) = 1} and x > 0 such that p(x, x ) = 1. Thus, 0 y 0 < y i < x < x i for all i = 1, 2,..., n. Let t 0 = y 0, t i = y i x i, i = 1, 2,..., n. Figure 4.2 The unit ball of a norm with piecewise linear boundary We will prove, by induction on n, that we can write p as a convex combination of {p ti, p 1 C G : i = 0, 1,..., n}. Base: n = 1. Let L 0, L 1, L 2 be the linear equations for the boundary of the unit ball of p on y x t 0, t 0 < y x t 1, and t 1 < y x 1, respectively. Note that if t 0 = 0, we can consider L 0 as a point

33 25 on the x-axis. Then m 1 = y 1 y 0 x 1 1 and m 2 = x y 1 x x 1 are slopes of L 1 and L 2, respectively. Note that m 1 < m 2 1. Then L 0 : x = 1, if 0 y x t 0 ; (4.3) L 1 : m 1 x + y = m 1 + y 0, if t 0 < y x t 1 ; (4.4) L 2 : m 2 x + y = m 2 x 1 + y 1, if t 1 < y x 1. (4.5) Assume that λ 0 p t0 (x, y) + λ 1 p t1 (x, y) + λ p 1 (x, y) = 1 where λ 0 + λ 1 + λ = 1. That is, x = 1, if 0 y x t 0 ; (4.6) ( ) λ0 + λ 1 + λ x + λ 0 y = 1, if t 0 < y 1 + t t 0 x t 1 ; (4.7) ( λ0 + λ ) ( 1 λ0 + λ x + + λ ) 1 y = 1, if t 1 < y 1. (4.8) 1 + t t t t 1 x Assume that the linear equations (4.3) to (4.5) for the boundary of p are equal to the linear equations (4.6) to (4.8), respectively. Then it is enough to show that 0 λ 0, λ 1, λ 1. For t 0 < y x t 1, multiply both sides of (4.7) by m 1 + y 0, then ( m 1 + y 0 ) ( ) λ0 λ 0 + λ 1 + λ x + ( m 1 + y 0 ) y = m 1 + y t t 0 By comparing coeffients with (4.4), we have that ( m 1 + y 0 ) ( ) λ0 + λ 1 + λ 1 + t 0 = m 1 (4.9) and ( m 1 + y 0 ) λ t 0 = 1. (4.10) Then λ 0 = 1 + t 0 m 1 + y 0 and so 0 < λ 0 < 1. By subtracting (4.10) from (4.9), λ 1 + λ = m 1 1 m 1 + y 0. (4.11)

34 26 For t 1 < y x 1, multiply both sides of (4.8) by m 2x 1 + y 1, then ( m 2 x 1 +y 1 ) ( λ0 + λ ) ( 1 λ0 + λ x+( m 2 x 1 +y 1 ) + λ ) 1 y = m 2 x 1 +y t t t t 1 Similarly, by comparing coeffients with (4.5), we have that ( m 2 x 1 + y 1 ) ( λ0 + λ ) 1 + λ 1 + t t 1 = m 2 (4.12) and ( m 2 x 1 + y 1 ) ( λ0 + λ ) t t 1 = 1. (4.13) By subtracting (4.13) to (4.12), λ = m 2 1 m 2 x 1 + y 1. Thus, λ 0, and by (4.11), λ 1 = m 1 1 m 1 + y 0 λ = m 1 1 m 1 + y 0 m 2 1 m 2 x 1 + y 1. By Lemma 4.1.1, λ 1 > 0. Therefore, λ 0, λ 1, λ < 1 because λ 0 +λ 1 +λ = 1 and λ 0, λ 1, λ 0. Hence, p(x, y) = λ 0 p t0 (x, y) + λ 1 p t1 (x, y) + λ p 1 (x, y)) for (x, y) B p. Induction Step: Assume that p co{p t : 0 t 1} for n 1. Let p be a norm in C G with piecewise linear boundary with n + 1 extreme points not lying on the x-y axis, say (x 1, y 1 ), (x 2, y 2 ),..., (x n+1, y n+1 ). Assume that 0 < y 1 < y 2 <... < y n+1 < x n+1 < x n <... < x 1 < 1. Let y 0 = max {y : p(1, y) = 1} and x > 0 such that p(x, x ) = 1. Thus, 0 y 0 < y i < x < x i for all i = 1, 2,..., n + 1. Let t 0 = y 0, t i = y i x i for i = 1, 2,..., n + 1. Let p be a norm with piecewise linear boundary with n extreme points (x 1, y 1 ), (x 2, y 2 ),..., (x n, y n ) not lying on the x-y axis and let x > 0 such that p (x, x ) = 1. By the induction hypothesis, where Claim: p = λ i p ti + λ p 1 λ i + λ = 1 and 0 < λ i, λ < 1, i = 0, 1,..., n. n+1 p = λ i p ti + λ p 1

35 27 n+1 where λ i + λ = 1 and 0 < λ i, λ < 1, i = 0, 1,..., n, n + 1. Then it is enough to show that each λ i exists for i = 0, 1,..., n + 1,. Note that the boundaries of B p and B p intersect on 0 y x t n. Then p(x, y) = p (x, y) for 0 y x t 0 and t k < y x t k+1, k = 0, 1,..., n 1. On 0 y x t 0. Then λ i x + λ x = n+1 λ i x + λ x. (4.14) and on t 0 < y x t 1, λ 0 x + y 1 + t 0 + n+1 λ i x + λ x + y x = λ 0 + λ i x + λ x. (4.15) 1 + t 0 By subtracting (4.15) from (4.14), λ 0 ( x x + y ) 1 + t 0 ( = λ 0 x x + y ) 1 + t 0 Therefore, λ 0 = λ 0. Next, fix k = 1,..., n 1. Similarly, we have that k λ i x + y + i=k+1 λ i x + λ x = k λ i x + y + n+1 i=k+1 λ i x + λ x. (4.16) and k 1 λ i x + y + λ i x + λ x = i=k By subtracting (4.17) from (4.16), λ k ( ) x + y x 1 + t k k 1 n+1 x + y λ i + λ i x + λ x. (4.17) i=k ( ) x + y = λ k x 1 + t k Thus, λ k = λ k for k = 1,..., n 1. Hence, by (4.14), we have that λ n + λ = λ n + λ n+1 + λ. Then, we will show that λ n, λ n+1, λ 0. Let L n 1, L n, L n+1 be the linear equations for the boundary of p on t n 1 < y x t n, t n < y x t n+1 and t n+1 < y x 1, respectively. Then m n 1 = y n y n 1 x n x n 1, m n = y n+1 y n x n+1 x n and m n+1 = x y n+1 x x n+1

36 28 are slopes of L n 1, L n and L n+1, respectively. Note that m n 1 < m n < m n+1 1. Therefore, we can write L n 1 : m n 1 x + y = m n 1 x n 1 + y n 1, if t n 1 < y x t n ; (4.18) L n : m n x + y = m n x n + y n, if t n < y x t n+1 ; (4.19) L n+1 : m n+1 x + y = m n+1 x n+1 + y n+1, if t n+1 < y x 1. (4.20) Then these linear equations are the same as p on the boundary on each interval, namely ( n 1 ) ( n 1 ) λ i λ i + λ n + λ n+1 + λ x + y = 1, if t n 1 < y x t n ; (4.21) ( ( n+1 λ i + λ n+1 + λ ) λ i + λ x + ) x + ( n+1 ( λ i λ i ) ) y = 1, if t n < y x t n+1 ; (4.22) y = 1, if t n+1 < y 1. (4.23) x For t n 1 < y x t n, multiply both sides of (4.21) by m n 1 x n 1 + y n 1, then ( m n 1 x n 1 + y n 1 ) ( n 1 ) λ i + λ n + λ n+1 + λ = m n 1 x n 1 + y n 1. x + ( m n 1 x n 1 + y n 1 ) ( n 1 λ i ) y By comparing coefficients with (4.18), we have that ( m n 1 x n 1 + y n 1 ) ( n 1 ) λ i + λ n + λ n+1 + λ = m n 1 (4.24) and By subtracting (4.25) from (4.24), ( m n 1 x n 1 + y n 1 ) ( n 1 λ i ) = 1. (4.25) λ n + λ n+1 + λ = m n 1 1 m n 1 x n 1 + y n 1. (4.26) For t n < y x t n+1, multiply both sides of (4.22) by m n x n + y n, then ( m n x n + y n ) ( λ i + λ n+1 + λ ) x + ( m n x n + y n ) ( λ i ) y = m n x n + y n.

37 29 By comparing coefficients with (4.19), we have that ( ) λ i ( m n x n + y n ) + λ n+1 + λ = m n (4.27) and ( ) λ i ( m n x n + y n ) By subtracting (4.28) from (4.27), = 1 (4.28) λ n+1 + λ = m n 1 m n x n + y n. (4.29) Then, by (4.26), λ n = m n 1 λ n+1 + λ = m n 1 m n 1 1. m n x n + y n m n x n + y n m n 1 x n 1 + y n 1 By Lemma 4.1.1, λ n > 0. For t n+1 < y x 1, multiply both sides of (4.23) by m n+1x n+1 + y n+1 ), then ( m n+1 x n+1 +y n+1 ) ( n+1 ) λ i + λ x+( m n+1 x n+1 +y n+1 ) ( n+1 λ i ) y = m n+1 x n+1 +y n+1. By comparing coefficients with (4.20), we have that and ( m n+1 x n+1 + y n+1 ) By subtracting (4.31) from (4.30), ( n+1 ) λ i + λ ( m n+1 x n+1 + y n+1 ) ( n+1 λ i ) = m n+1 (4.30) = 1 (4.31) λ = m n+1 1 m n+1 x n+1 + y n+1 0. By (4.29), λ n+1 = m n 1 λ = m n 1 m n+1 1. m n x n + y n m n x n + y n m n+1 x n+1 + y n+1 By Lemma 4.1.1, λ n+1 > 0. Hence, we complete the proof.

38 Theorem Any p C G can be approximated by norms in C G with piecewise linear boundaries. Proof. Let p C G. To prove the statement of the theorem, we have to show that there is a sequence of norms p n in C G with piecewise linear boundaries converging to p in the norm space H G. That is, given ɛ > 0, there exists an N 0 N such that all n N 0. Claim that it is enough to show that sup p(x,y)=1 Let ɛ > 0. Assume that there is an N 1 N such that sup p(x,y)=1 for all n N. With this N 1, for any 0 < c 1 and n N, ɛ c sup p n (x, y) p(x, y) < ɛ for l 2 (x,y) 1 p n (x, y) p(x, y) < ɛ for all n N 0. sup p(x,y)=1 p n (x, y) p(x, y) < ɛ p n (x, y) p(x, y) <. Then, sup p n (c x, c y) p(c x, c y) < ɛ. Since this is true for arbitrary 0 < c 1, p(x,y)=1 p n (x, y) p(x, y) < ɛ for all n N 1. sup p(x,y) 1 Now, by Theorem 1.2.6, there exists m, M > 0 such that m l 2 (x, y) p(x, y) M l 2 (x, y). Choose N 2 N such that p(m x, M y) < ɛ. Then, sup p n (x, y) p(x, y) < ɛ p(x,y) 1 M. For n N 2, sup p n (M x, M y) p(x,y) 1 p n (x, y) p(x, y) < ɛ. Notice that if l 2 (x, y) 1, then sup p(x,y) M p(x, y) M l 2 (x, y) M. Thus, sup l 2 (x,y) 1 Then, let N 0 = min{n 1, N 2 }. We have that p n (x, y) p(x, y) sup p(x,y) M p n (x, y) p(x, y). p n (x, y) p(x, y) < ɛ for all n N 0. sup l 2 (x,y) 1 Therefore, we have the claim. Hence, it suffices to show that there is a sequence of norms p n with piecewise linear boundaries in C G which converges to 1 on the boundary of B p. Without loss of generality, we will consider only (x, y) R 2 such that x, y 0. Let 0 = x 0 < x 1 < x 2 <... < x n = 1 such that x k+1 x k = 1 n for k = 0, 1, 2,..., n 1. Let y 0 = max {y : p(1, y) = 1} and y k such that p(x k, y k ) = 1 for k = 1,..., n. Note that y 0 y 1... y n since B p is convex. Let B n be a subset of R 2 with piecewise linear boundary where (x 0, y 0 ), (x 1, y 1 ),..., (x n, y n ) are its extreme points. Note that B n is the unit ball of a norm p n for some p n C G. Let (x, y) B p. Then x [x k, x k+1 ] for some k = 0, 1, 2,..., n 1. Let ɛ > 0 be given. Since the boundary of p is a continuous function on x, y 0, there exists δ > 0 such that if x x < δ, then y y < ɛ 2 for y R such that p(x, y ) = 1. Choose

39 31 N = max { 1 δ, 2 ɛ } + 1. Then for any n N, p n (x, y) 1 = p n (x, y) p n (x k, y k ) p n ((x, y) (x k, y k )) = p n (x x k, y y k ) p n (x x k, 0) + p n (0, y y k ) < x x k + y y k 1 n + y y k < ɛ 2 + ɛ 2 = ɛ.

40 32 CHAPTER 5. SUMMARY AND DISCUSSION First, recall that C is a set of norms on R 2 with for any norm p C, p(0, 1) = p(1, 0) = 1. We know that C is bounded and equicontinuous but it is not closed. By the Arzela-Ascoli Theorem, this set is not a compact set. Then C does not satisfy the sufficient conditions of the Krein-Milman Theorem. Thus, we cannot conclude whether its extreme points exist. However, we have found that the minimum norm is an extreme point of C. Next, we consider the maximum norm as a candidate for another extreme point. Unfortunately, the result appears to be unlike expected since we can write it in term of a convex combination of two distinct norms in C. Furthermore, the set C G is defined to be a set of norms on R 2 with for any p C G, p(1, 0) = p(0, 1) = 1 and p is invariant under the dihedral group. This set is closed, bounded and equicontinuous. Thus, by the Arzela-Ascoli Theorem, C G is compact. In addition, this set is also a convex set. Therefore, the Krein-Milman Theorem guarantees that an extreme point of C G exists and the closed convex hull of the set of all extreme points of C G is the set C G. The same as the set C, the minimum norm is an extreme point of C G. Moreover, we have that the maximum norm is another extreme point of C G. These two norms lead us to an idea of finding other extreme points of C G. More specifically, recall that for 0 t 1, x if 0 y x p t (x, y) = t ; x+y 1+t if t < y x 1. We have that p t is a norm in C G and is also an extreme point of C G. Note that p 0 is the minimum norm and p 1 is the maximum norm. In addition, the set {p t : 0 t 1} is closed. To show that {p t : 0 t 1} is the only extreme points of C G, we raised the following known theorem.

41 33 Theorem Let X be a norm space and K a compact convex subset of X. If F is a closed subset of extreme points of K such that co(f ) = K, then F = ext(k). Comparing with the set C G, it can be seen that the closed convex hull of {p t : 0 t 1} is a subset of C G since C G is a closed convex set containing each p t. Then, the next theorems are our main results which show that the converse of the previous statement is true. Theorem The convex hull of {p t : 0 t 1} is the set of all norms in C G with piecewise linear boundaries. Theorem Any p C G can be approximated by norms in C G with piecewise linear boundaries. These two theorems imply that for any norm p in C G, there is a sequence of p n such that p n converges to p where each p n is a convex combination of {p t : 0 t 1}. In other words, C G is contained in the closed convex hull of the set {p t : 0 t 1}. Hence, the closed convex hull of the set {p t : 0 t 1} is equal to C G. Therefore, we can conclude that {p t : 0 t 1} is the set of all extreme points of C G. From this result, one could expand an idea to a set of symmetric norms on R 2 without the restriction on the points (1, 0) and (0, 1) lying on the boundaries of the unit balls. Alternatively, instead of norms on R 2, one could consider the set C and C G by the norms on R n.

42 34 APPENDIX. KNOWN THEOREMS In this section, we provide proofs of some known theorems in the functional analysis that are referred to in this paper. The first theorem is about equivalence of norms on R n. Next,the Arzela-Ascoli Theorem and the Krien-Milman Theorem are provided. Based on our interest, these two theorems are restated to consider only on a metric space. The last one is the theorem that we use to conclude our desirable result. Theorem Any two norms on R n are equivalent. Proof. To proof the statement of this theorem, it is suffices to show that any norm on R n is equivalent to a Euclidean norm l 2. Let p be a norm on R n. For any u = (u 1,..., u n ) R n, ( p(u) = p = u i e i ) p(u i e i ) u i p(e i ) ( ) 1 2 (u i ) 2 p(e i ) = l 2 (u) p(e i ). where e i = (0,..., 0, 1, 0,..., 0), only the i th component is equal to 1, i = 1, 2,..., n. Choose M = p(e i ) > 0. Then p(u) M l 2 (u) for u R n. Now, let S = {v R n : l 2 (v) = 1}. Then S is closed and bounded with respect to the Euclidean norm l 2. By the Heine-Borel theorem,

43 35 S is compact. Since p is continuous, p(s) attains a minimum, says m. Then m = p(v 0 ) for some v 0 S. Since l 2 (v 0 ) = 1, v 0 0. Therefore, m = p(v 0 ) > 0. For any u R n, ) = p m, i.e. p(u) m l 2 (u). Hence, m l 2 (u) p(u) M l 2 (u) for any p(u) l 2 (u) u R n. ( u l 2 (u) Arzela-Ascoli Theorem Let (X, d) be a compact metric space. Let B(X, R) be the set of all bounded continuous functions from X to R. For any subset C of B(X, R), C is compact if and only if C is closed, bounded, and equicontinuous. Proof. Let C be a subset of B(X, R). Note that B(X, R) is a metric space induced by a supremum norm. Recall that a compact subset of a metric space is closed. Then, if C is compact, then C is closed. Note that B(X, R) is a complete metric space. Recall that a closed subspace of a complete space is complete. Thus, if C is closed, then C is complete. By the Heine Borel Theorem, it suffices to show that C is totally bounded if and only if C is bounded and equicontinuous. First, assume that C is totally bounded. Let ɛ > 0 be given. Then there exist f i C n such that C {f B(X, R) : f i f < ɛ } for some n and i = 1,..., n. Let x X. By 3 continuity of each f i, there exists δ i > 0 such that if d(x, y) < δ i, then f i (x) f i (y) < ɛ 3 for all i = 1,..., n. For any f C, we have that f k f < ɛ 3 for some k = 1,..., n. Choose δ = min δ i > 0. If d(x, y) < δ, then,...,n f(x) f(y) = f(x) f k (x) + f k (x) f k (y) + f k (y) f(y) f(x) f k (x) + f k (x) f k (y) + f k (y) f(y) < ɛ 3 + ɛ 3 + ɛ 3 = ɛ. Conversely, Assume that C is bounded and equicontinuous. Let x X. Then there exists δ x > 0 such that if d(x, y) < δ x, then f(x) f(y) < ɛ 3 for all f F. Let U x = {y X :

44 36 d(x, y) < δ x }. Thus, {U x : x X} is an open cover of X. Since X is compact, X some x i X, i = 1,..., n. n U xi for Since C is bounded, there is an M > 0 such that f M for all f C. Let K = [ M, M]. Note that K is compact. Note that {a R : a α < ɛ } is an open cover for K. Then 2 there exist α j K such that K α K m j=1 {a R : a α j < ɛ } for some m and i = 1, 2,..., m. 2 Let B = {(β 1,..., β n ) : β i = α j for i = 1, 2,..., n, j = 1, 2,..., m}. Thus, B = m n. Let {φ i } be a partition of unity subordinate to the cover {U xi } for i = 1, 2,..., n, which is to say φ i = 1 and φ i (x) = 0 if x / U xi. For b = (β 1,..., β n ) B, let g b = Claim: C {f B(X, R) : f g b < ɛ}. b B β i φ i. Let f C, then {f(x 1 ),..., f(x n )} K and so f(x i ) α ji < ɛ 2 for some j i = 1, 2,..., m and i = 1, 2,..., m. Pick b = (β 1,..., β n ) B such that β i = α ji for each i = 1, 2,..., n. For any x X, we have that x U xk for some k = 1, 2,..., n and so f(x) g b (x) = f(x) β i φ i (x) = f(x) φ i (x) β i φ i (x) = (f(x) β i ) φ i (x) f(x) β i φ i (x) = f(x) β k f(x) f(x k ) + f(x k ) β j < ɛ. Hence, f(x) g b (x) < ɛ for any x X, i.e. f g b < ɛ.

45 37 Krein-Milman Theorem Let (X, d) be a norm space and let K be any nonempty compact convex subset of X. Then ext(k) and K is the closed convex hull of its extreme points. Proof. Note that x ext(k) if and only if K {x} is convex. Let F be the set of all proper relatively open convex subsets of K. If K is a singleton, then we are done. Assume a, b are distinct elements in K. Let r = d(a,b) 2, then B(a, r) is an open set in X. Thus, B(a, r) K is a relatively open convex subset of K. Since b / B(a, r), F is nonempty. Let C be a chain in F and let U = {E : E C }. Clearly, U is open and convex. If U = K, by compactness, n U = E i for some E i C. Since C is a chain, we can assume that E i E i+1 for all i = 1,..., n 1, then U = E n K, a contradiction. Thus, U F. By Zorn s lemma, F has a maximal element, say M. For x M, 0 λ < 1, let T x,λ : K K be defined by T x,λ (y) = λ y + (1 λ) x. Let ɛ > 0 and let δ = ɛ λ > 0. If d(y 1, y 2 ) < δ, then T x,λ (y 1 ) T x,λ (y 2 ) = λ y 1 y 2 < ɛ. Therefore, T x,λ is continuous. Claim 1: T x,λ (K) M. Note that T x,λ (M) M, i.e. M T 1 1 x,λ (M). Let s, t Tx,λ (M). For any 0 < α < 1, consider T x,λ (α s + (1 α) t) = λ (α s + (1 α) t) + (1 λ) x = α (λ s + (1 λ) x) + (1 α) (λ t + (1 λ) x) = α T x,λ (s) + (1 α) T x,λ (t). Since T x,λ (s), T x,λ (t) M, T x,λ (α s + (1 α) t) M. Thus, α s + (1 α) t T 1 x,λ (M). That is, T 1 1 x,λ (M) is convex. Notice that Tx,λ (M) is open. For y cl(m) M, T x,λ(y) M. Therefore, cl(m) T 1 1 x,λ (M). By the maximality of M, Tx,λ (M) = K. Hence, we have Claim 1. Claim 2: For any open convex subset V of K, if V M, then V M = K. Let V be an open convex subset of K such that V M. Note that V M is open. Let x, y be distinct elements in V M. For 0 < λ < 1, if both x and y are either in V or M, then

46 38 λ y + (1 λ) x V M. Suppose that x M and y V M. Then T x,λ (y) M V M, i.e. V M is convex. By the maximality of M, V M = K. Thus, we have Claim 2. Next, we want to show that K M is a singleton. Suppose that there exist a, b K M such that a b. Let V a, V b be disjoint open convex subset of K such that a V a and b V b. By Claim 2, V a M = K. Since b / M, b / V a M, a contradiction. Therefore, K M is a singleton and so ext(k). Claim 3: For any open convex subset V of X, if ext(k) V, then K V. Let V be an open convex subset of X such that ext(k) V. Suppose that V K K. Then V K is a proper open convex subset of K and so V K F. Thus, V K is contained in a maximal element M of F. Note that K M is a singleton of an extreme point of K. This is a contradiction, since V contains all extreme points of K. Then, V K = K, i.e. K V. Therefore, we have Claim 3. Let C be the closed convex hull of the extreme points of K. Suppose C K, then let x K C. Note that {x} is compact. By the Hahn-Banach Seperation Theorem, there is a function f : X R such that for all c C, f(c) α < α + ɛ f(x) for some α C. Let V = {y X : f(y) < α + ɛ 2 }. Then C V and by Claim 3, K V. This is a contradiction since x K and f(x) α, i.e. x / V. Hence, C = K. Theorem Let X be a norm space and K a compact convex subset of X. If F is a closed subset of extreme points of K such that co(f ) = K, then F = ext(k). Proof. Suppose that there is an extreme point x 0 of K such that x 0 / F. Then we can find a norm p on X such that F {x X : p(x x 0 ) < 1} =. Let U = {x X : p(x) < 1 3 }. Then U is an open convex set and 0 U. Assume that there is an s (x 0 + U) (F + U). Then s = x 0 + u and s = y + v for some u, v U and y F. That is, y x 0 = u v and so p(y x 0 ) = p(u v) p(u) + p(v) < < 1. Then, y / F, a contradiction. Thus, (x 0 + U) (F + U) =. Therefore, x 0 / cl(f + U) since x 0 + U is a neighborhood of x 0. Note

Numerical computation of an optimal control problem with homogenization in one-dimensional case

Numerical computation of an optimal control problem with homogenization in one-dimensional case Retrospective Theses and Dissertations Iowa State University Capstones, Theses and Dissertations 28 Numerical computation of an optimal control problem with homogenization in one-dimensional case Zhen

More information

Extreme points of compact convex sets

Extreme points of compact convex sets Extreme points of compact convex sets In this chapter, we are going to show that compact convex sets are determined by a proper subset, the set of its extreme points. Let us start with the main definition.

More information

B. Appendix B. Topological vector spaces

B. Appendix B. Topological vector spaces B.1 B. Appendix B. Topological vector spaces B.1. Fréchet spaces. In this appendix we go through the definition of Fréchet spaces and their inductive limits, such as they are used for definitions of function

More information

Locally convex spaces, the hyperplane separation theorem, and the Krein-Milman theorem

Locally convex spaces, the hyperplane separation theorem, and the Krein-Milman theorem 56 Chapter 7 Locally convex spaces, the hyperplane separation theorem, and the Krein-Milman theorem Recall that C(X) is not a normed linear space when X is not compact. On the other hand we could use semi

More information

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

1 Topology Definition of a topology Basis (Base) of a topology The subspace topology & the product topology on X Y 3 Index Page 1 Topology 2 1.1 Definition of a topology 2 1.2 Basis (Base) of a topology 2 1.3 The subspace topology & the product topology on X Y 3 1.4 Basic topology concepts: limit points, closed sets,

More information

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

converges as well if x < 1. 1 x n x n 1 1 = 2 a nx n Solve the following 6 problems. 1. Prove that if series n=1 a nx n converges for all x such that x < 1, then the series n=1 a n xn 1 x converges as well if x < 1. n For x < 1, x n 0 as n, so there exists

More information

7 Complete metric spaces and function spaces

7 Complete metric spaces and function spaces 7 Complete metric spaces and function spaces 7.1 Completeness Let (X, d) be a metric space. Definition 7.1. A sequence (x n ) n N in X is a Cauchy sequence if for any ɛ > 0, there is N N such that n, m

More information

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

MAT 570 REAL ANALYSIS LECTURE NOTES. Contents. 1. Sets Functions Countability Axiom of choice Equivalence relations 9 MAT 570 REAL ANALYSIS LECTURE NOTES PROFESSOR: JOHN QUIGG SEMESTER: FALL 204 Contents. Sets 2 2. Functions 5 3. Countability 7 4. Axiom of choice 8 5. Equivalence relations 9 6. Real numbers 9 7. Extended

More information

Introduction to Real Analysis Alternative Chapter 1

Introduction to Real Analysis Alternative Chapter 1 Christopher Heil Introduction to Real Analysis Alternative Chapter 1 A Primer on Norms and Banach Spaces Last Updated: March 10, 2018 c 2018 by Christopher Heil Chapter 1 A Primer on Norms and Banach Spaces

More information

Compact operators on Banach spaces

Compact operators on Banach spaces Compact operators on Banach spaces Jordan Bell jordan.bell@gmail.com Department of Mathematics, University of Toronto November 12, 2017 1 Introduction In this note I prove several things about compact

More information

SOLUTIONS TO SOME PROBLEMS

SOLUTIONS TO SOME PROBLEMS 23 FUNCTIONAL ANALYSIS Spring 23 SOLUTIONS TO SOME PROBLEMS Warning:These solutions may contain errors!! PREPARED BY SULEYMAN ULUSOY PROBLEM 1. Prove that a necessary and sufficient condition that the

More information

Chapter 2 Metric Spaces

Chapter 2 Metric Spaces Chapter 2 Metric Spaces The purpose of this chapter is to present a summary of some basic properties of metric and topological spaces that play an important role in the main body of the book. 2.1 Metrics

More information

MATH 54 - TOPOLOGY SUMMER 2015 FINAL EXAMINATION. Problem 1

MATH 54 - TOPOLOGY SUMMER 2015 FINAL EXAMINATION. Problem 1 MATH 54 - TOPOLOGY SUMMER 2015 FINAL EXAMINATION ELEMENTS OF SOLUTION Problem 1 1. Let X be a Hausdorff space and K 1, K 2 disjoint compact subsets of X. Prove that there exist disjoint open sets U 1 and

More information

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

Functional Analysis. Franck Sueur Metric spaces Definitions Completeness Compactness Separability... Functional Analysis Franck Sueur 2018-2019 Contents 1 Metric spaces 1 1.1 Definitions........................................ 1 1.2 Completeness...................................... 3 1.3 Compactness......................................

More information

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

Lecture Notes in Advanced Calculus 1 (80315) Raz Kupferman Institute of Mathematics The Hebrew University Lecture Notes in Advanced Calculus 1 (80315) Raz Kupferman Institute of Mathematics The Hebrew University February 7, 2007 2 Contents 1 Metric Spaces 1 1.1 Basic definitions...........................

More information

Topology. Xiaolong Han. Department of Mathematics, California State University, Northridge, CA 91330, USA address:

Topology. Xiaolong Han. Department of Mathematics, California State University, Northridge, CA 91330, USA  address: Topology Xiaolong Han Department of Mathematics, California State University, Northridge, CA 91330, USA E-mail address: Xiaolong.Han@csun.edu Remark. You are entitled to a reward of 1 point toward a homework

More information

The Heine-Borel and Arzela-Ascoli Theorems

The Heine-Borel and Arzela-Ascoli Theorems The Heine-Borel and Arzela-Ascoli Theorems David Jekel October 29, 2016 This paper explains two important results about compactness, the Heine- Borel theorem and the Arzela-Ascoli theorem. We prove them

More information

MATH 202B - Problem Set 5

MATH 202B - Problem Set 5 MATH 202B - Problem Set 5 Walid Krichene (23265217) March 6, 2013 (5.1) Show that there exists a continuous function F : [0, 1] R which is monotonic on no interval of positive length. proof We know there

More information

A NICE PROOF OF FARKAS LEMMA

A NICE PROOF OF FARKAS LEMMA A NICE PROOF OF FARKAS LEMMA DANIEL VICTOR TAUSK Abstract. The goal of this short note is to present a nice proof of Farkas Lemma which states that if C is the convex cone spanned by a finite set and if

More information

Set, functions and Euclidean space. Seungjin Han

Set, functions and Euclidean space. Seungjin Han Set, functions and Euclidean space Seungjin Han September, 2018 1 Some Basics LOGIC A is necessary for B : If B holds, then A holds. B A A B is the contraposition of B A. A is sufficient for B: If A holds,

More information

Continuous Functions on Metric Spaces

Continuous Functions on Metric Spaces Continuous Functions on Metric Spaces Math 201A, Fall 2016 1 Continuous functions Definition 1. Let (X, d X ) and (Y, d Y ) be metric spaces. A function f : X Y is continuous at a X if for every ɛ > 0

More information

Helly's Theorem and its Equivalences via Convex Analysis

Helly's Theorem and its Equivalences via Convex Analysis Portland State University PDXScholar University Honors Theses University Honors College 2014 Helly's Theorem and its Equivalences via Convex Analysis Adam Robinson Portland State University Let us know

More information

Topological properties

Topological properties CHAPTER 4 Topological properties 1. Connectedness Definitions and examples Basic properties Connected components Connected versus path connected, again 2. Compactness Definition and first examples Topological

More information

Problem Set 2: Solutions Math 201A: Fall 2016

Problem Set 2: Solutions Math 201A: Fall 2016 Problem Set 2: s Math 201A: Fall 2016 Problem 1. (a) Prove that a closed subset of a complete metric space is complete. (b) Prove that a closed subset of a compact metric space is compact. (c) Prove that

More information

Analysis Finite and Infinite Sets The Real Numbers The Cantor Set

Analysis Finite and Infinite Sets The Real Numbers The Cantor Set Analysis Finite and Infinite Sets Definition. An initial segment is {n N n n 0 }. Definition. A finite set can be put into one-to-one correspondence with an initial segment. The empty set is also considered

More information

Notes on Distributions

Notes on Distributions Notes on Distributions Functional Analysis 1 Locally Convex Spaces Definition 1. A vector space (over R or C) is said to be a topological vector space (TVS) if it is a Hausdorff topological space and the

More information

FIXED POINT METHODS IN NONLINEAR ANALYSIS

FIXED POINT METHODS IN NONLINEAR ANALYSIS FIXED POINT METHODS IN NONLINEAR ANALYSIS ZACHARY SMITH Abstract. In this paper we present a selection of fixed point theorems with applications in nonlinear analysis. We begin with the Banach fixed point

More information

Spaces of continuous functions

Spaces of continuous functions Chapter 2 Spaces of continuous functions 2.8 Baire s Category Theorem Recall that a subset A of a metric space (X, d) is dense if for all x X there is a sequence from A converging to x. An equivalent definition

More information

Continuity. Matt Rosenzweig

Continuity. Matt Rosenzweig Continuity Matt Rosenzweig Contents 1 Continuity 1 1.1 Rudin Chapter 4 Exercises........................................ 1 1.1.1 Exercise 1............................................. 1 1.1.2 Exercise

More information

Convex Geometry. Carsten Schütt

Convex Geometry. Carsten Schütt Convex Geometry Carsten Schütt November 25, 2006 2 Contents 0.1 Convex sets... 4 0.2 Separation.... 9 0.3 Extreme points..... 15 0.4 Blaschke selection principle... 18 0.5 Polytopes and polyhedra.... 23

More information

S. Mrówka introduced a topological space ψ whose underlying set is the. natural numbers together with an infinite maximal almost disjoint family(madf)

S. Mrówka introduced a topological space ψ whose underlying set is the. natural numbers together with an infinite maximal almost disjoint family(madf) PAYNE, CATHERINE ANN, M.A. On ψ (κ, M) spaces with κ = ω 1. (2010) Directed by Dr. Jerry Vaughan. 30pp. S. Mrówka introduced a topological space ψ whose underlying set is the natural numbers together with

More information

Notes for Functional Analysis

Notes for Functional Analysis Notes for Functional Analysis Wang Zuoqin (typed by Xiyu Zhai) November 6, 2015 1 Lecture 18 1.1 The convex hull Let X be any vector space, and E X a subset. Definition 1.1. The convex hull of E is the

More information

Analysis III. Exam 1

Analysis III. Exam 1 Analysis III Math 414 Spring 27 Professor Ben Richert Exam 1 Solutions Problem 1 Let X be the set of all continuous real valued functions on [, 1], and let ρ : X X R be the function ρ(f, g) = sup f g (1)

More information

Chapter 2 Convex Analysis

Chapter 2 Convex Analysis Chapter 2 Convex Analysis The theory of nonsmooth analysis is based on convex analysis. Thus, we start this chapter by giving basic concepts and results of convexity (for further readings see also [202,

More information

TOPOLOGICAL VECTOR SPACES

TOPOLOGICAL VECTOR SPACES TOPOLOGICAL VECTOR SPACES PRADIPTA BANDYOPADHYAY 1. Topological Vector Spaces Let X be a linear space over R or C. We denote the scalar field by K. Definition 1.1. A topological vector space (tvs for short)

More information

McGill University Math 354: Honors Analysis 3

McGill University Math 354: Honors Analysis 3 Practice problems McGill University Math 354: Honors Analysis 3 not for credit Problem 1. Determine whether the family of F = {f n } functions f n (x) = x n is uniformly equicontinuous. 1st Solution: The

More information

Math 5052 Measure Theory and Functional Analysis II Homework Assignment 7

Math 5052 Measure Theory and Functional Analysis II Homework Assignment 7 Math 5052 Measure Theory and Functional Analysis II Homework Assignment 7 Prof. Wickerhauser Due Friday, February 5th, 2016 Please do Exercises 3, 6, 14, 16*, 17, 18, 21*, 23*, 24, 27*. Exercises marked

More information

THE INVERSE FUNCTION THEOREM

THE INVERSE FUNCTION THEOREM THE INVERSE FUNCTION THEOREM W. PATRICK HOOPER The implicit function theorem is the following result: Theorem 1. Let f be a C 1 function from a neighborhood of a point a R n into R n. Suppose A = Df(a)

More information

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

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 Math 8B Solutions Charles Martin March 6, Homework Problems. Let (X i, d i ), i n, be finitely many metric spaces. Construct a metric on the product space X = X X n. Proof. Denote points in X as x = (x,

More information

Maths 212: Homework Solutions

Maths 212: Homework Solutions Maths 212: Homework Solutions 1. The definition of A ensures that x π for all x A, so π is an upper bound of A. To show it is the least upper bound, suppose x < π and consider two cases. If x < 1, then

More information

CHAPTER 7. Connectedness

CHAPTER 7. Connectedness CHAPTER 7 Connectedness 7.1. Connected topological spaces Definition 7.1. A topological space (X, T X ) is said to be connected if there is no continuous surjection f : X {0, 1} where the two point set

More information

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

Real Analysis Math 131AH Rudin, Chapter #1. Dominique Abdi Real Analysis Math 3AH Rudin, Chapter # Dominique Abdi.. If r is rational (r 0) and x is irrational, prove that r + x and rx are irrational. Solution. Assume the contrary, that r+x and rx are rational.

More information

Problem Set 5: Solutions Math 201A: Fall 2016

Problem Set 5: Solutions Math 201A: Fall 2016 Problem Set 5: s Math 21A: Fall 216 Problem 1. Define f : [1, ) [1, ) by f(x) = x + 1/x. Show that f(x) f(y) < x y for all x, y [1, ) with x y, but f has no fixed point. Why doesn t this example contradict

More information

Course 212: Academic Year Section 1: Metric Spaces

Course 212: Academic Year Section 1: Metric Spaces Course 212: Academic Year 1991-2 Section 1: Metric Spaces D. R. Wilkins Contents 1 Metric Spaces 3 1.1 Distance Functions and Metric Spaces............. 3 1.2 Convergence and Continuity in Metric Spaces.........

More information

2. Dual space is essential for the concept of gradient which, in turn, leads to the variational analysis of Lagrange multipliers.

2. Dual space is essential for the concept of gradient which, in turn, leads to the variational analysis of Lagrange multipliers. Chapter 3 Duality in Banach Space Modern optimization theory largely centers around the interplay of a normed vector space and its corresponding dual. The notion of duality is important for the following

More information

ANALYSIS WORKSHEET II: METRIC SPACES

ANALYSIS WORKSHEET II: METRIC SPACES ANALYSIS WORKSHEET II: METRIC SPACES Definition 1. A metric space (X, d) is a space X of objects (called points), together with a distance function or metric d : X X [0, ), which associates to each pair

More information

Convex Analysis and Optimization Chapter 2 Solutions

Convex Analysis and Optimization Chapter 2 Solutions Convex Analysis and Optimization Chapter 2 Solutions Dimitri P. Bertsekas with Angelia Nedić and Asuman E. Ozdaglar Massachusetts Institute of Technology Athena Scientific, Belmont, Massachusetts http://www.athenasc.com

More information

Math 421, Homework #9 Solutions

Math 421, Homework #9 Solutions Math 41, Homework #9 Solutions (1) (a) A set E R n is said to be path connected if for any pair of points x E and y E there exists a continuous function γ : [0, 1] R n satisfying γ(0) = x, γ(1) = y, and

More information

Real Analysis Chapter 4 Solutions Jonathan Conder

Real Analysis Chapter 4 Solutions Jonathan Conder 2. Let x, y X and suppose that x y. Then {x} c is open in the cofinite topology and contains y but not x. The cofinite topology on X is therefore T 1. Since X is infinite it contains two distinct points

More information

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

MATH 51H Section 4. October 16, Recall what it means for a function between metric spaces to be continuous: MATH 51H Section 4 October 16, 2015 1 Continuity Recall what it means for a function between metric spaces to be continuous: Definition. Let (X, d X ), (Y, d Y ) be metric spaces. A function f : X Y is

More information

USING FUNCTIONAL ANALYSIS AND SOBOLEV SPACES TO SOLVE POISSON S EQUATION

USING FUNCTIONAL ANALYSIS AND SOBOLEV SPACES TO SOLVE POISSON S EQUATION USING FUNCTIONAL ANALYSIS AND SOBOLEV SPACES TO SOLVE POISSON S EQUATION YI WANG Abstract. We study Banach and Hilbert spaces with an eye towards defining weak solutions to elliptic PDE. Using Lax-Milgram

More information

1. Let A R be a nonempty set that is bounded from above, and let a be the least upper bound of A. Show that there exists a sequence {a n } n N

1. Let A R be a nonempty set that is bounded from above, and let a be the least upper bound of A. Show that there exists a sequence {a n } n N Applied Analysis prelim July 15, 216, with solutions Solve 4 of the problems 1-5 and 2 of the problems 6-8. We will only grade the first 4 problems attempted from1-5 and the first 2 attempted from problems

More information

Geometry and topology of continuous best and near best approximations

Geometry and topology of continuous best and near best approximations Journal of Approximation Theory 105: 252 262, Geometry and topology of continuous best and near best approximations Paul C. Kainen Dept. of Mathematics Georgetown University Washington, D.C. 20057 Věra

More information

I. The space C(K) Let K be a compact metric space, with metric d K. Let B(K) be the space of real valued bounded functions on K with the sup-norm

I. The space C(K) Let K be a compact metric space, with metric d K. Let B(K) be the space of real valued bounded functions on K with the sup-norm I. The space C(K) Let K be a compact metric space, with metric d K. Let B(K) be the space of real valued bounded functions on K with the sup-norm Proposition : B(K) is complete. f = sup f(x) x K Proof.

More information

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

l(y j ) = 0 for all y j (1) Problem 1. The closed linear span of a subset {y j } of a normed vector space is defined as the intersection of all closed subspaces containing all y j and thus the smallest such subspace. 1 Show that

More information

Bootcamp. Christoph Thiele. Summer As in the case of separability we have the following two observations: Lemma 1 Finite sets are compact.

Bootcamp. Christoph Thiele. Summer As in the case of separability we have the following two observations: Lemma 1 Finite sets are compact. Bootcamp Christoph Thiele Summer 212.1 Compactness Definition 1 A metric space is called compact, if every cover of the space has a finite subcover. As in the case of separability we have the following

More information

Functional Analysis

Functional Analysis The Hahn Banach Theorem : Functional Analysis 1-9-06 Extensions of Linear Forms and Separation of Convex Sets Let E be a vector space over R and F E be a subspace. A function f : F R is linear if f(αx

More information

consists of two disjoint copies of X n, each scaled down by 1,

consists of two disjoint copies of X n, each scaled down by 1, Homework 4 Solutions, Real Analysis I, Fall, 200. (4) Let be a topological space and M be a σ-algebra on which contains all Borel sets. Let m, µ be two positive measures on M. Assume there is a constant

More information

MAT 544 Problem Set 2 Solutions

MAT 544 Problem Set 2 Solutions MAT 544 Problem Set 2 Solutions Problems. Problem 1 A metric space is separable if it contains a dense subset which is finite or countably infinite. Prove that every totally bounded metric space X is separable.

More information

Analysis III Theorems, Propositions & Lemmas... Oh My!

Analysis III Theorems, Propositions & Lemmas... Oh My! Analysis III Theorems, Propositions & Lemmas... Oh My! Rob Gibson October 25, 2010 Proposition 1. If x = (x 1, x 2,...), y = (y 1, y 2,...), then is a distance. ( d(x, y) = x k y k p Proposition 2. In

More information

Spectral theory for compact operators on Banach spaces

Spectral theory for compact operators on Banach spaces 68 Chapter 9 Spectral theory for compact operators on Banach spaces Recall that a subset S of a metric space X is precompact if its closure is compact, or equivalently every sequence contains a Cauchy

More information

Lecture Notes on Metric Spaces

Lecture Notes on Metric Spaces Lecture Notes on Metric Spaces Math 117: Summer 2007 John Douglas Moore Our goal of these notes is to explain a few facts regarding metric spaces not included in the first few chapters of the text [1],

More information

Math 209B Homework 2

Math 209B Homework 2 Math 29B Homework 2 Edward Burkard Note: All vector spaces are over the field F = R or C 4.6. Two Compactness Theorems. 4. Point Set Topology Exercise 6 The product of countably many sequentally compact

More information

Most Continuous Functions are Nowhere Differentiable

Most Continuous Functions are Nowhere Differentiable Most Continuous Functions are Nowhere Differentiable Spring 2004 The Space of Continuous Functions Let K = [0, 1] and let C(K) be the set of all continuous functions f : K R. Definition 1 For f C(K) we

More information

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

Problem Set 6: Solutions Math 201A: Fall a n x n, Problem Set 6: Solutions Math 201A: Fall 2016 Problem 1. Is (x n ) n=0 a Schauder basis of C([0, 1])? No. If f(x) = a n x n, n=0 where the series converges uniformly on [0, 1], then f has a power series

More information

Theorems. Theorem 1.11: Greatest-Lower-Bound Property. Theorem 1.20: The Archimedean property of. Theorem 1.21: -th Root of Real Numbers

Theorems. Theorem 1.11: Greatest-Lower-Bound Property. Theorem 1.20: The Archimedean property of. Theorem 1.21: -th Root of Real Numbers Page 1 Theorems Wednesday, May 9, 2018 12:53 AM Theorem 1.11: Greatest-Lower-Bound Property Suppose is an ordered set with the least-upper-bound property Suppose, and is bounded below be the set of lower

More information

Problem Set 5. 2 n k. Then a nk (x) = 1+( 1)k

Problem Set 5. 2 n k. Then a nk (x) = 1+( 1)k Problem Set 5 1. (Folland 2.43) For x [, 1), let 1 a n (x)2 n (a n (x) = or 1) be the base-2 expansion of x. (If x is a dyadic rational, choose the expansion such that a n (x) = for large n.) Then the

More information

Math 5210, Definitions and Theorems on Metric Spaces

Math 5210, Definitions and Theorems on Metric Spaces Math 5210, Definitions and Theorems on Metric Spaces Let (X, d) be a metric space. We will use the following definitions (see Rudin, chap 2, particularly 2.18) 1. Let p X and r R, r > 0, The ball of radius

More information

Appendix B Convex analysis

Appendix B Convex analysis This version: 28/02/2014 Appendix B Convex analysis In this appendix we review a few basic notions of convexity and related notions that will be important for us at various times. B.1 The Hausdorff distance

More information

APPLICATIONS IN FIXED POINT THEORY. Matthew Ray Farmer. Thesis Prepared for the Degree of MASTER OF ARTS UNIVERSITY OF NORTH TEXAS.

APPLICATIONS IN FIXED POINT THEORY. Matthew Ray Farmer. Thesis Prepared for the Degree of MASTER OF ARTS UNIVERSITY OF NORTH TEXAS. APPLICATIONS IN FIXED POINT THEORY Matthew Ray Farmer Thesis Prepared for the Degree of MASTER OF ARTS UNIVERSITY OF NORTH TEXAS December 2005 APPROVED: Elizabeth M. Bator, Major Professor Paul Lewis,

More information

Aliprantis, Border: Infinite-dimensional Analysis A Hitchhiker s Guide

Aliprantis, Border: Infinite-dimensional Analysis A Hitchhiker s Guide aliprantis.tex May 10, 2011 Aliprantis, Border: Infinite-dimensional Analysis A Hitchhiker s Guide Notes from [AB2]. 1 Odds and Ends 2 Topology 2.1 Topological spaces Example. (2.2) A semimetric = triangle

More information

Real Analysis Problems

Real Analysis Problems Real Analysis Problems Cristian E. Gutiérrez September 14, 29 1 1 CONTINUITY 1 Continuity Problem 1.1 Let r n be the sequence of rational numbers and Prove that f(x) = 1. f is continuous on the irrationals.

More information

EC 521 MATHEMATICAL METHODS FOR ECONOMICS. Lecture 1: Preliminaries

EC 521 MATHEMATICAL METHODS FOR ECONOMICS. Lecture 1: Preliminaries EC 521 MATHEMATICAL METHODS FOR ECONOMICS Lecture 1: Preliminaries Murat YILMAZ Boğaziçi University In this lecture we provide some basic facts from both Linear Algebra and Real Analysis, which are going

More information

Rose-Hulman Undergraduate Mathematics Journal

Rose-Hulman Undergraduate Mathematics Journal Rose-Hulman Undergraduate Mathematics Journal Volume 17 Issue 1 Article 5 Reversing A Doodle Bryan A. Curtis Metropolitan State University of Denver Follow this and additional works at: http://scholar.rose-hulman.edu/rhumj

More information

Week 5 Lectures 13-15

Week 5 Lectures 13-15 Week 5 Lectures 13-15 Lecture 13 Definition 29 Let Y be a subset X. A subset A Y is open in Y if there exists an open set U in X such that A = U Y. It is not difficult to show that the collection of all

More information

Part III. 10 Topological Space Basics. Topological Spaces

Part III. 10 Topological Space Basics. Topological Spaces Part III 10 Topological Space Basics Topological Spaces Using the metric space results above as motivation we will axiomatize the notion of being an open set to more general settings. Definition 10.1.

More information

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

g 2 (x) (1/3)M 1 = (1/3)(2/3)M. COMPACTNESS If C R n is closed and bounded, then by B-W it is sequentially compact: any sequence of points in C has a subsequence converging to a point in C Conversely, any sequentially compact C R n is

More information

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

d(x n, x) d(x n, x nk ) + d(x nk, x) where we chose any fixed k > N Problem 1. Let f : A R R have the property that for every x A, there exists ɛ > 0 such that f(t) > ɛ if t (x ɛ, x + ɛ) A. If the set A is compact, prove there exists c > 0 such that f(x) > c for all x

More information

Lecture 8: Basic convex analysis

Lecture 8: Basic convex analysis Lecture 8: Basic convex analysis 1 Convex sets Both convex sets and functions have general importance in economic theory, not only in optimization. Given two points x; y 2 R n and 2 [0; 1]; the weighted

More information

The Arzelà-Ascoli Theorem

The Arzelà-Ascoli Theorem John Nachbar Washington University March 27, 2016 The Arzelà-Ascoli Theorem The Arzelà-Ascoli Theorem gives sufficient conditions for compactness in certain function spaces. Among other things, it helps

More information

Honours Analysis III

Honours Analysis III Honours Analysis III Math 354 Prof. Dmitry Jacobson Notes Taken By: R. Gibson Fall 2010 1 Contents 1 Overview 3 1.1 p-adic Distance............................................ 4 2 Introduction 5 2.1 Normed

More information

A New Family of Topological Invariants

A New Family of Topological Invariants Brigham Young University BYU ScholarsArchive All Theses and Dissertations 2018-04-01 A New Family of Topological Invariants Nicholas Guy Larsen Brigham Young University Follow this and additional works

More information

On the Countable Dense Homogeneity of Euclidean Spaces. Randall Gay

On the Countable Dense Homogeneity of Euclidean Spaces. Randall Gay On the Countable Dense Homogeneity of Euclidean Spaces by Randall Gay A thesis submitted to the Graduate Faculty of Auburn University in partial fulfillment of the requirements for the Degree of Master

More information

REVIEW OF ESSENTIAL MATH 346 TOPICS

REVIEW OF ESSENTIAL MATH 346 TOPICS REVIEW OF ESSENTIAL MATH 346 TOPICS 1. AXIOMATIC STRUCTURE OF R Doğan Çömez The real number system is a complete ordered field, i.e., it is a set R which is endowed with addition and multiplication operations

More information

MAA6617 COURSE NOTES SPRING 2014

MAA6617 COURSE NOTES SPRING 2014 MAA6617 COURSE NOTES SPRING 2014 19. Normed vector spaces Let X be a vector space over a field K (in this course we always have either K = R or K = C). Definition 19.1. A norm on X is a function : X K

More information

Topological properties of Z p and Q p and Euclidean models

Topological properties of Z p and Q p and Euclidean models Topological properties of Z p and Q p and Euclidean models Samuel Trautwein, Esther Röder, Giorgio Barozzi November 3, 20 Topology of Q p vs Topology of R Both R and Q p are normed fields and complete

More information

NOTES ON VECTOR-VALUED INTEGRATION MATH 581, SPRING 2017

NOTES ON VECTOR-VALUED INTEGRATION MATH 581, SPRING 2017 NOTES ON VECTOR-VALUED INTEGRATION MATH 58, SPRING 207 Throughout, X will denote a Banach space. Definition 0.. Let ϕ(s) : X be a continuous function from a compact Jordan region R n to a Banach space

More information

Mathematics for Economists

Mathematics for Economists Mathematics for Economists Victor Filipe Sao Paulo School of Economics FGV Metric Spaces: Basic Definitions Victor Filipe (EESP/FGV) Mathematics for Economists Jan.-Feb. 2017 1 / 34 Definitions and Examples

More information

Experimental designs for multiple responses with different models

Experimental designs for multiple responses with different models Graduate Theses and Dissertations Graduate College 2015 Experimental designs for multiple responses with different models Wilmina Mary Marget Iowa State University Follow this and additional works at:

More information

Topology Part of the Qualify Exams of Department of Mathematics, Texas A&M University Prepared by Zhang, Zecheng

Topology Part of the Qualify Exams of Department of Mathematics, Texas A&M University Prepared by Zhang, Zecheng Topology Part of the Qualify Exams of Department of Mathematics, Texas A&M University Prepared by Zhang, Zecheng Remark 0.1. This is a solution Manuel to the topology questions of the Topology Geometry

More information

Austin Mohr Math 730 Homework. f(x) = y for some x λ Λ

Austin Mohr Math 730 Homework. f(x) = y for some x λ Λ Austin Mohr Math 730 Homework In the following problems, let Λ be an indexing set and let A and B λ for λ Λ be arbitrary sets. Problem 1B1 ( ) Show A B λ = (A B λ ). λ Λ λ Λ Proof. ( ) x A B λ λ Λ x A

More information

Lectures on Analysis John Roe

Lectures on Analysis John Roe Lectures on Analysis John Roe 2005 2009 1 Lecture 1 About Functional Analysis The key objects of study in functional analysis are various kinds of topological vector spaces. The simplest of these are the

More information

Metric Spaces and Topology

Metric Spaces and Topology Chapter 2 Metric Spaces and Topology From an engineering perspective, the most important way to construct a topology on a set is to define the topology in terms of a metric on the set. This approach underlies

More information

Overview of normed linear spaces

Overview of normed linear spaces 20 Chapter 2 Overview of normed linear spaces Starting from this chapter, we begin examining linear spaces with at least one extra structure (topology or geometry). We assume linearity; this is a natural

More information

REAL AND COMPLEX ANALYSIS

REAL AND COMPLEX ANALYSIS REAL AND COMPLE ANALYSIS Third Edition Walter Rudin Professor of Mathematics University of Wisconsin, Madison Version 1.1 No rights reserved. Any part of this work can be reproduced or transmitted in any

More information

Immerse Metric Space Homework

Immerse Metric Space Homework Immerse Metric Space Homework (Exercises -2). In R n, define d(x, y) = x y +... + x n y n. Show that d is a metric that induces the usual topology. Sketch the basis elements when n = 2. Solution: Steps

More information

Convex Analysis and Economic Theory Winter 2018

Convex Analysis and Economic Theory Winter 2018 Division of the Humanities and Social Sciences Ec 181 KC Border Convex Analysis and Economic Theory Winter 2018 Supplement A: Mathematical background A.1 Extended real numbers The extended real number

More information

MATH 722, COMPLEX ANALYSIS, SPRING 2009 PART 5

MATH 722, COMPLEX ANALYSIS, SPRING 2009 PART 5 MATH 722, COMPLEX ANALYSIS, SPRING 2009 PART 5.. The Arzela-Ascoli Theorem.. The Riemann mapping theorem Let X be a metric space, and let F be a family of continuous complex-valued functions on X. We have

More information

Definition 1.1. A partially ordered set (M, ) is a set M equipped with a binary relation, called a partial ordering, satisfying

Definition 1.1. A partially ordered set (M, ) is a set M equipped with a binary relation, called a partial ordering, satisfying 1 Zorn s Lemma Definition 1.1. A partially ordered set (M, ) is a set M equipped with a binary relation, called a partial ordering, satisfying (a) x x for every x M. (b) If x y and y x, then x = y. (c)

More information

Chapter 1 Preliminaries

Chapter 1 Preliminaries Chapter 1 Preliminaries 1.1 Conventions and Notations Throughout the book we use the following notations for standard sets of numbers: N the set {1, 2,...} of natural numbers Z the set of integers Q the

More information