Homework 3 Solutions Math 501 Due September 19, 2014 Exercise 1 (a) This is obvious. (b) Let f (x) = 2x. Let > 0, let x ∈ R, and let δ = /2. Observe that for any t ∈ R such that |x − t| < δ, we have |f (x) − f (t)| = |2x − 2t| = 2|x − t| < 2δ = . Since δ is independent of x, it follows that f is uniformly continuous. (c) Let g(x) = x2 . We must show that for any δ > 0 there exist > 0 and x, t ∈ R with |x − t| < δ such that |g(x) − g(t)| ≥ . Let δ > 0 and let = 1. Let x = δ −1 and let t = x + δ/2. Then |x − t| = δ/2 < δ. However, observe that |f (x) − f (t)| = |x2 − t2 | = |xδ + δ 2 /4| = 1 + δ 2 /4 ≥ 1 = . (d) Let h(x) = sin(1/x). Let δ > 0 and let = 1. For each n ≥ 1, let xn = π 2 1 , + 2πn tn = 1 . − π2 + 2πn It’s easy to see that there exist N1 , N2 ∈ N such that |xn | < δ/2 for n ≥ N1 and |tn | < δ/2 for n ≥ N2 . Let N = max{N1 , N2 } and observe that |xN − tN | ≤ |xN | + |tN | < δ. Moreover, observe that |h(xN ) − h(tN )| = | sin(π/2 + 2πN ) − sin(−π/2 + 2πN )| = 2 > . 1 Exercise 2 Let xn = sup{ak : k ≥ n}, yn = sup{bk : k ≥ n}, zn = sup{ak + bk : k ≥ n}, and let x = lim xn , n→∞ y = lim yn , n→∞ z = lim zn . n→∞ [Note that it was shown in a previous assignment that these limits do in fact exist and the method by which this was proved showed that the limits were the infimums shown above.] We claim that the inequality zn ≤ xn + yn holds for all n ≥ 1. We shall prove this fact below, but in the meantime note that this implies z ≤ x + y, the desired inequality. To see that the claimed inequality holds, let n ≥ 1, let k ≥ n, and observe that ak + bk ≤ sup{ak : k ≥ n} + sup{bk : k ≥ n}. It follows that sup{ak : k ≥ n} + sup{bk : k ≥ n} is an upper bound for the set {ak + bk : k ≥ n}. Since sup{ak + bk : k ≥ n} is the least upper bound for this set the desired inequality follows. Exercise 3 (a) Let (V, h·, ·i) and define a norm k·k : V → R by kvk = that for u, v ∈ V 2 p hv, vi. Observe 2 ku + vk + ku − vk = hu + v, u + vi + hu − v, u − vi = hu, ui + 2hu, vi + hv, vi + hu, ui − 2hu, vi + hv, vi = 2hu, ui + 2hv, vi 2 2 = 2 kuk + kvk . (b) Skip. (c) Let V = C[0, 1] be the space of continuous functions defined on the unit interval [0, 1]. Define a map k·k∞ : C[0, 1] → R by kf k∞ = sup |f (x)| = sup{|f (x)| : x ∈ [0, 1]}. x∈[0,1] (i) The boundedness theorem tells us that functions in C[0, 1] are bounded, and hence have finite sup-norm. 2 (ii) Before proving that k·k∞ is a norm, we prove the following facts about supremum. Lemma 1. Let S ⊂ R be a nonempty set, let c > 0, and let f, g : S → R be functions. The following hold: i. supx∈S cx = c sup S. ii. supx∈S |f (x) + g(x)| ≤ supx∈S |f (x)| + supx∈S |g(x)| Proof of Lemma. i. Let x ∈ S, let α = sup S, and let β = supx∈S cx. Then x ≤ α and hence cx ≤ cα. Since x was arbitrary it follows that cα is an upper bound of the set {cx : x ∈ S} which implies β ≤ cα. On the other hand, note that β ≥ cx for all x ∈ S, which implies c−1 β ≥ x for all x ∈ S. Therefore, c−1 β is an upper bound for S, which implies c−1 β ≥ α. Hence β ≥ cα. It follows that β = cα. ii. Let x ∈ S and let α = supx∈S |f (x)| and β = supx∈S |g(x)|. |f (x) + g(x)| ≤ |f (x)| + |g(x)| ≤ α + β which implies α + β is an upper bound for {|f (x) + g(x)| : x ∈ S}. Therefore, sup |f (x) + g(x)| ≤ α + β. x∈S Now, we proceed to show that k·k∞ is a norm. Let f, g ∈ C[0, 1] and let c ∈ R. By the first part of the lemma, we have that kcf k∞ = sup |cf (x)| = |c| sup |f (x)| = |c| kf k∞ . x∈[0,1] x∈[0,1] Next observe that kf + gk∞ ≤ kf k∞ + kgk∞ directly by the second part of the lemma. Finally, it is clear that |f (x)| ≥ 0 for all x ∈ [0, 1], and hence kf k∞ ≥ 0. As for equality, if f ≡ 0 then |f (x)| = 0 for all x ∈ [0, 1] and hence kf k∞ = 0. Conversely, if kf k∞ = 0, then 0 ≤ |f (x)| ≤ kf k∞ = 0 for all x ∈ [0, 1], which implies f ≡ 0. (d) We claim that the sup-norm does not satisfy the parallelogram law. Consider f (x) = 1 and g(x) = x − 1. Then observe that (f + g)(x) = x and (f − g)(x) = 2 − x and kf k∞ = 1, Hence kgk∞ = 1, kf + gk∞ = 1, kf − gk∞ = 2 2 2 2 2 kf + gk∞ + kf − gk∞ 6= 2 kf k∞ + kgk∞ . 3 Exercise 4 Let p ∈ R and suppose that f : R → R is not continuous at p. That is, suppose that there exists > 0 such that for every δ > 0 there exists x such that |x − p| < δ and |f (x) − f (p)| ≥ . This implies that for every n ≥, there exists xn such that |xn − p| < 1/n and |f (xn ) − f (p)| ≥ . It’s clear that the sequence xn converges to p. Indeed, for any α > 0, if we let N = dα−1 e then |xn − p| < α for all n ≥ N . Moreover, since |f (xn ) − f (p)| ≥ for all n ≥ 1, it cannot be that f (xn ) converges to f (p). Exercise 5 Consider the space M = C[0, 1] of continuous, real-valued functions defined on the unit interval [0, 1]. Define a map d on M × M by d(f, g) = sup |f (x) − g(x)| = sup {|f (x) − g(x)| : x ∈ [0, 1]} . x∈[0,1] (a) That d(f, g) ∈ R for all f, g ∈ C[0, 1] is again a consequence of the boundedness (or extreme value) theorem. (b) Note that d(f, g) = kf − gk∞ , where k·k∞ denotes the sup-norm on C[0, 1]. Since kf − gk∞ = kg − f k∞ is easy to see, it clearly follows that d(f, g) = d(g, f ). Moreover, note that because k·k∞ is a norm, it follows that khk∞ ≥ 0 for all h ∈ C[0, 1] with equality holding if and only if h ≡ 0. This holds in particular for h = f − g. This implies, d(f, g) = kf − gk∞ ≥ 0 with equality if and only if f − g ≡ 0; that is, if and only if f = g. Finally note that d(f, g) = kf − gk∞ = kf − h + h − gk∞ ≤ kf − hk∞ + kh − gk∞ = d(f, h) + d(h, g). It follows that d is a metric. (c) The above arguments show that any norm on a vector space naturally induce a metric structure on the vector space. 4