Homework 4 Math 501 Due September 26, 2014 Exercise 1 Let A1 = (0, 1), A2 = {(x, 0) ∈ R2 : x ∈ (0, 1)}. (a) Note that A1 = B1/2 (1/2) = {x ∈ R : |x − 1/2| < 1/2} is the open ball in R of radius 1/2 about the point 1/2. Hence A1 is open because open balls are open. (b) Note that for any x = (x1 , 0) ∈ A2 , there does not exist r > 0 such that Br (x) ⊂ A2 . This is the case because for every r > 0, the set Br (x) = {y ∈ R2 : |x − y| < r} contains for example the point (x1 , r/2) 6∈ A2 . It follows that A2 is not open in R2 . Exercise 2 Let M be a metric space and suppose that S ⊂ T ⊂ M . Prove that (a) S ⊂ T , (b) int S ⊂ int T . (a) Let x ∈ S = lim S. Then there exists (xn ) ⊂ S such that xn → x. Note that (xn ) is a sequence in T as well. Therefore, x ∈ lim T = T . (b) Let x ∈ int S. Then x ∈ U for some open set contained in S. Note that U is also an open set contained in T . Therefore x is contained in the union of all open sets contained in T . In other words, x ∈ int T . 1 Exercise 3 Let M be a metric space and let A ⊂ B ⊂ C ⊂ M . Suppose that A is dense in B and B is dense in C. To show that A is dense in C it suffices to show that for every c ∈ C, there exists a sequence (an ) ⊂ A such that an → c. Let c ∈ C and > 0. Since B is dense in C, there exists a sequence (bn ) ⊂ B such that bn → c. That is, there exists N 0 such that d(bn , c) < /2 for all n ≥ N 0 . Note that since A is dense in B, for each n ≥ 1, there exists a (n) (n) sequence (am ) ⊂ A such that am → bn as m → ∞. That is, for each n ≥ 1, there exists Nn such that d(a(n) m , bn ) < /2 for all m ≥ Nn . Let N = max{N 0 , NN 0 }. Then for n ≥ N , we have that (n) d(a(n) n , c) ≤ d(an , bn ) + d(bn , c) < . (n) Therefore the sequence an = an is a sequence in A that converges to c. Exercise 4 (a) We first claim that any singleton set {n} ⊂ N is open. Indeed, observe that for any r ∈ (0, 1), the ball Br (n) = {m ∈ N : |n − m| < r} = {n}. Next consider any subset S ⊂ N. Note that S is the union of open sets– namely the union of singletons– and hence open. Thus any subset of N is open. In particular, S c is open as well, which implies S is closed. (b) Let M be a metric space and let f : N → M be a function. Note that if U ⊂ M is open, then f −1 (U ) is open in N since every subset of N is open. Hence f is continuous. Exercise 5 Let M be a metric space, let S ⊂ M be a non-empty subset, and let p ∈ M . Define dist(p, S) = inf{d(p, s) : s ∈ S}. Suppose that p ∈ lim S. Then there exists a sequence (pn ) ⊂ S such that pn → p. Thus for any > 0, there exists N such that d(pn , p) < . Therefore 0 ≤ d(p, S) ≤ d(pn , p) < , 2 which implies dist(p, S) = 0. Conversely suppose that dist(p, S) = 0. Note that for every > 0, there exists q ∈ S such that 0 = dist(p, S) ≤ d(p, q) ≤ dist(p, S) + = . Therefore, for every n ≥ 1, there exists qn ∈ S such that 0 ≤ d(p, qn ) ≤ 1 . n It follows that the sequence (qn ) ⊂ S converges to p. Hence p ∈ lim S. Exercise 6 Let (xn ) ⊂ R be a sequence that converges to say x ∈ R. It is straightforward to check that the function f : R → R given by f (x) = |x| is continuous. From this it follows that f (xn ) → f (x). In other words, the sequence (|xn |) converges. Exercise 7 (a) Let f : R → R2 be given by f (x) = (x, 0). It is straightforward to see that f is a homeomorphism onto its image. To see that f (R) is closed, let (xn , yn ) ∈ f (R) be a sequence that converges to (x, y) ∈ R2 . To see that f (R) is closed, we must show that (x, y) ∈ f (R). Note first that it must be the case that yn = 0 for all n ≥ 1, from which it follows that y = 0. Therefore (x, 0) = f (x) ∈ f (R). (b) Let N = (0, 1) ∈ R2 denote the so-called “north-pole” of S 1 . Consider the map g : R → S 1 \ {N } which is defined as follows. For each x ∈ R, consider the line `x ⊂ R2 which connects (x, 0) and N . Define g(x) ∈ R2 to be the point of intersection of `x with S 1 \ {N }. (i) Let p ∈ R be nonzero. Note that the line `p has Cartesian equation y = −x/p + 1. To find the point of intersection of `p with S 1 \ {N } we solve the system y = − xp + 1 x2 + y 2 = 1. A straightforward computation shows that (2p/(p2 + 1), (p2 − 1)/(p2 + 1)) solves the system. It is easy to see that 2p p2 − 1 g(p) = , p2 + 1 p2 + 1 is continuous. 3 (ii) To see that g is a homeomorphism, it suffices to show that g has an inverse and that inverse is continuous. To find the inverse of g, first suppose (u, v) ∈ S 1 \ {N }. The line which passes through N and (u, v) is given by the Cartesian equation y= v−1 x + 1, u which clearly passes through the x-axis at u/(1 − v). It follows that g −1 (u, v) = u , 1−v which again is clearly continuous. (c) There cannot exist an embedding which maps R into a closed and bounded subset of R2 because this would imply that R is homeomorphic to a compact set, which in turn would imply R is compact, which is not the case. 4