2.13 (a) If A and B are disjoint closed sets in some metric space X, prove that they are separated. Rudin’s Ex. 19 (b) Prove the same for disjoint open sets. (c) Fix p ∈ X, δ > 0, define A to be the set of all q ∈ X for which d(p, q) < δ, define B similarly, with > in place of <. Prove that A and B are separated. (d) Prove that every connected metric space with at least two points is uncountable. Hint: Use (c). Proof (a) If A and B are closed, then A = A and B = B. By A ∩ B = ∅, we have A ∩ B = A ∩ B = ∅ and A ∩ B = A ∩ B = ∅. Hence A and B are separated. (b) Since A ∩ B = ∅, we know that B ⊂ Ac . The complement Ac is closed, since A is open. Hence B ⊂ Ac = Ac . Thus A ∩ B ⊂ A ∩ Ac = ∅, which implies A ∩ B = ∅. Similarly, we have A ∩ B = ∅. Therefore, A and B are separated. (c) The sets A and B are A = {x ∈ X : d(x, p) < δ}, B = {x ∈ X : d(x, p) > δ}. The set A is open, by Theorem 2.19. To see that B is open, we let x ∈ B be any point. By d(x, p) > δ, we take any number r such that 0 < r ≤ d(x, p) − δ. We claim that Nr (x) ⊂ B. In fact, if z ∈ Nr (x), then d(z, p) ≥ d(x, p) − d(z, x) > d(x, p) − r ≥ δ, which implies z ∈ B. Thus B is open. It is clear that A and B are disjoint. Hence, by (b), A and B are separated. (d) Suppose p and q are two distinct points of a connected metric space X. Denote L = d(p, q) > 0. For each number r, with 0 < r < L, put Ar = {x ∈ X : d(x, p) < r}, Br = {x ∈ X : d(x, p) > r}. Since p ∈ Ar and q ∈ Br , we know that both sets are non-empty. By (c), we further know that A and B are separated. By the condition, X is connected, so X 6= Ar ∪ Br . Since Ar ∪ Br ⊂ X, there exists a point zr such that zr ∈ X − (Ar ∪ Br ). By the definitions of Ar and Br , we know that d(zr , p) = r. If r 6= s, then d(zr , p) 6= d(zs , p), so zr 6= zs . Thus, r 7→ zr defines an injective map from (0, L) to X. Since (0, L) is uncountable, so is X. 2.14 Let A and B be separated subsets of some Rk , suppose a ∈ A, b ∈ B, and define p(t) = (1 − t)a + tb for t ∈ R. Put A0 = p−1 (A), B0 = p−1 (B). [Thus t ∈ A0 if and only if p(t) ∈ A.] (a) Prove that A0 and B0 are separated sets of R. (b) Prove that there exists t0 ∈ (0, 1) such that p(t0 ) ∈ / A ∪ B. (c) Prove that every convex subset of Rk is connected. 1 Rudin’s Ex. 21 Proof (a) We shall prove A0 ∩ B0 = ∅. In fact, if A0 ∩ B0 6= ∅, we let x ∈ A0 ∩ B0 . Then x ∈ A0 , and x ∈ B0 ∪ B00 . If x ∈ B0 , then x ∈ A0 ∩ B0 , which implies p(x) ∈ A ∩ B ⊂ A ∩ B. This contradicts to the assumption that A and B are separated. Thus, x is a limit point of B0 . We shall prove that p(x) is a limit point of B. For any neighborhood Nr (p(x)), we need to prove that Nr (p(x)) contains a point p(y) in B, with p(y) 6= p(x). Indeed, notice that since A and B be separated, hence, for a ∈ A and b ∈ B, we have a 6= b. Since x a limit point of B0 , in the neighborhood of x: (x − r/|b − a|, x + r/|b − a|) , there exists a point y 6= x such that y ∈ B0 , so p(y) ∈ B. From |p(y) − p(x)| = |[(1 − y)a + yb] − [(1 − x)a + xb]| = |y − x| · |b − a|, we know that p(y) 6= p(x), and p(y) ∈ Nr (p(x)). This concludes that p(x) is a limit point of B, or p(x) ∈ B 0 . Thus, again we have p(x) ∈ A ∩ B, which contradicts to the assumption that A and B are separated. Therefore, A0 ∩ B0 = ∅. It is similar to prove that A0 ∩ B0 = ∅. Hence, A0 and B0 are separated. (b) Suppose for every t ∈ (0, 1), we have p(t) ∈ A ∪ B. Since a = p(0) ∈ A and b = p(1) ∈ B, we have p(t) ∈ A∪B for every t ∈ [0, 1]. This implies that t ∈ A0 ∪B0 . Hence, [0, 1] ⊂ A0 ∪ B0 . It follows that [0, 1] = {A0 ∩ [0, 1]} ∪ {B0 ∩ [0, 1]} . It is clear that 0 ∈ A0 ∩ [0, 1], and 1 ∈ B0 ∩ [0, 1]. By (a), A0 and B0 are separated, so we have {A0 ∩ [0, 1]} ∩ {B0 ∩ [0, 1]} ⊂ A0 ∩ B0 = ∅, and [B0 ∩ (0, 1)] ∩ [A0 ∩ (0, 1)] ⊂ B0 ∩ A0 = ∅. Hence, the segment [0, 1] is a union of two non-empty separated sets, but this contradicts to the fact that [0, 1] is connected, by Theorem 2.47. Hence, there is a point t0 ∈ [0, 1] such that p(t0 ) ∈ / A ∪ B. Apparently, t0 can be neither 0 nor 1, since 0 ∈ A and 1 ∈ B. Therefore, there is a point t0 ∈ (0, 1) such that p(t0 ) ∈ / A ∪ B. This completes the proof of (b). (c) Let S be a convex subset of Rk . If S is not connected, by the definition, there are two non-empty separated sets A and B such that S = A ∪ B. By (b), there exists t0 ∈ (0, 1) such that p(t0 ) ∈ / A∪B. This contradicts to the fact that p(t) ∈ S = A∪B for every t ∈ (0, 1) for the convex set S. Hence, S is connected. 2