SOLUTIONS OF HW2 MINGFENG ZHAO February 14, 2011 1. (Page 39, 28) Let F be increasing and right continuous, and let µF be the associated measure. Then µF ({a}) = F (a) − F (a−), µF ([a, b)) = F (b−) − F (a−), µF ([a, b]) = F (b) − F (a−), and µF ((a, b)) = F (b−) − F (a). Proof. Let F be increasing and right continuous, by 1.16 Theorem in Page 35, we know that we can find an associated Borel measure µF such that µF ((a, b]) = F (b) − F (a) for all a, b ∈ R. Since F is increasing, so F (a−) exists for all a ∈ R. We look at En = a − n1 , a , then En ⊃ En+1 for all n ≥ 1, and T∞ n=1 En = {a}, and µF (En ) = F (a) − F a − 1 n . And also we know µF (E1 ) = F (a)−F (a−1) < ∞. Since µF is a measure, then {a} is µF -measurable. By the continuity of measure from above, we get µF ({a}) = = = = lim µF (En ) n→∞ 1 lim F (a) − F a − n→∞ n 1 F (a) − lim F a − n→∞ n F (a) − F (a−). S For µF ([a, b]), and a < b, since [a, b] = (a, b] {a}, and (a, b], {a} are µF -measurable, then [a, b] is µF -measurable.And we have µF ([a, b]) = µF ((a, b] [ {a}) 1 2 MINGFENG ZHAO = µF ((a, b]) + µF ({a}) By additivity of µF = F (b) − F (a) + F (a) − F (a−) = F (b) − F (a−). S Since [a, b) {b} = [a, b], so [a, b) = [a, b]\{b}. Since [a, b], {b} are µF -measurable, then [a, b) is µF -measurable. And by additivity of µF , we have µF ([a, b]) = µF ([a, b)) + µF ({b}) That is, F (b) − F (a−) = µF ([a, b)) + F (b) − F (b−), so µF ([a, b)) = F (b) − F (a−) − (F (b) − F (b−)) = F (b) − F (a−) − F (b) + F (b−) = F (b−) − F (a−). S Since (a, b) {b} = (a, b], so (a, b) = (a, b]\{b}. Because (a, b], {b, } are µF -measurable, then (a, b) is µF -measurable. And by additivity of µF , we have µF ((a, b]) = µF ((a, b)) + µF ({b}). So we get µF ((a, b)) = µF ((a, b]) − µF ({b}) = F (b) − F (a) − (F (b) − F (b−)) = F (b) − F (a) − F (b) + F (b−) = F (b−) − F (a). SOLUTIONS OF HW2 3 2. (Page 39, 29) Let E be a Lebesgue measurable set. a. If E ⊂ N , where N is the nonmeasurbale set described in 1.1, then m(E) = 0. b. If m(E) > 0, then E contains a nonmeasurbale set. T Proof. a. If m(E) > 0, consider Er = r + E = {r + e : e ∈ E}, for any r ∈ Q [0, 1]. From the construction of N , we know that Er T T Es = φ if r 6= s, r, s ∈ Q [0, 1]. On the other hand, from T the definition of m, we know Er is also m-measurable for all r ∈ Q [0, 1]. And by the translation invariant of m, we get m(Er ) = m(E) > 0 for all r ∈. But since N ⊂ [0, 1], then Er ⊂ [0, 2] for all r∈Q T [0, 1] , which implies that S r∈Q T [0,1] Er ⊂ [0, 2], so m S r∈Q T [0,1] Er ≤ 2. T Since Q [0, 1] is countable, then by the additivity of measure, hence we have m [ r∈Q T Er X = [0,1] r∈Q r∈Q = ∞ S Since Er \ Es = φ if r 6= s, r, s ∈ Q m(E) T [0,1] Since m(E) > 0, and Q T r∈Q [0,1] \ [0, 1] is countable. Er ≤ 2. Therefore, m(E) = 0. b. For any E which is m-measurable, and m(E) > 0. For any n ∈ Z, let En = E En T \ [0, 1] T [0,1] X = This contradicts with m m(Er ) T [n, n + 1). Then El = φ if n 6= l, n, l ∈ Z, and En is m-measurable. Claim I: There exists some k ∈ Z such that m(Ek ) > 0. If m(En ) = 0 for all n ∈ Z.By the additivity of m, we have ! m(E) = m [ En n∈Z = X m(En ) n∈Z = X 0 n∈Z = 0, which contradicts with m(E) > 0. Therefore, there exists some k ∈ Z such that m(Ek ) > 0. 4 MINGFENG ZHAO Claim II: There exists some subset F ⊂ Ek ⊂ E which is not m-measurable. If all subsets of Ek are measurable. Since Ek ⊂ [k, k + 1), then Ek − k ⊂ [0, 1). For any r ∈ [−1, 1] T Q, we let Fr = (Ek − k) Since [0, 1] ⊂ S r∈[−1,1] T Q \ Nr = Ek Nr , so Ek − k = S \ r∈[−1,1] Since all subsets of Ek are m-measurable, then Ek m, we know Fr = Ek T Nr+k − k ⊂ [0, 1]. T T Q Fr . Nr+k is m-measurable. By the definition of Nr+k − k is m-measurable. So Fr − r ⊂ Nr − r = N is m-measurable, by the result of part a, we know m(Fr − r) = 0, so m(Fr ) = 0. Since [0, 1] ⊂ m(Ek − k) = X r∈[−1,1] Q Nr , then T Q m(Fr ) T Q T Q X r∈[−1,1] = T Fr r∈[−1,1] = r∈[−1,1] [ m = S 0 0, which contradicts with m(Ek − k) = m(Ek ) > 0. Therefore, There exists some subset F ⊂ Ek ⊂ E which is not m-measurable. Remark 1. By the same idea as book, we can get another proof of part b. For any n ∈ N, let En = E T [−n, n]. Then En is m-measurable, and En ⊂ En+1 for all n ∈ N. Since m(E) > 0, by the continuity of measure from below, we know for such that whenever n ≥ N , we have m(En ) > m(E) − m(E) 2 = m(E) 2 m(E) 2 > 0, there exists N ∈ N > 0. Let G = EN , then m(G) > 0. Now we will find a subset F ⊂ G = EN ⊂ E such that F is not m-measurable. (Follow the same idea before.) 1. Firstly, we define an equivalence relation ∼ on G: We say x ∼ y if x − y ∈ Q. SOLUTIONS OF HW2 5 Claim I: ∼ is an an equivalence relation ∼ on G. i. (Reflexivity) For any x ∈ G, we know x − x = 0 ∈ Q, so x ∼ x. ii. (Symmetry) For any x, y ∈ G, and x ∼ y, that is x − y ∈ Q. Since Q is a field, then y − x = −(x − y) ∈ Q. So y ∼ x. iii. (Transitivity) For any x, y, z ∈ G, and x ∼ y, y ∼ z, that is x − y ∈ Q, y − z ∈ Q. Since Q is a field, then x − z = (x − y) + (y − z) ∈ Q. So x ∼ z. By i, ii, iii, we know that ∼ is an an equivalence relation ∼ on G. Let [x] = {y ∈ G : x ∼ y}. T Claim II: For any x, y ∈ G, if [x] [y] 6= φ, then [x] = [y]. T T Since [x] [y] 6= φ, we take z ∈ [x] [y]. Then x ∼ z, and y ∼ z. By the transitivity of ∼, we know x ∼ y. For any t ∈ [x], then t ∼ x. Since x ∼ y, by the transitivity of ∼, then t ∼ y, which implies t ∈ [y]. On the other hand, for any t ∈ [y], then t ∼ y. Since x ∼ y, By the transitivity of ∼, we know t ∼ x, which implies t ∈ [x]. Therefore, we get [x] = [y]. Let P = {[x] : x ∈ G}. For any A ∈ P, we know A = [x] for some x ∈ G. Since x ∈ [x], so A 6= φ. Claim III: P is a partition of G. For any x ∈ G, we know x ∈ [x], and [x] ∈ P, so x ∈ For any A, B ∈ G, if A T S A∈P A. B 6= φ. From the definition of P, we know A = [x], B = [y] for some T x, y ∈ G, then [x] [y] 6= φ. By Claim II, we know [x] = [y], that is A = B. Therefore, P is a partition of G. Now we assume P = {Bα : α ∈ A}, where A is an index set such that α1 6= α2 , then Bα1 T Bα2 = φ. For each Bα , we take one point xα in Bα . Let F = {xα : α ∈ A}. Hence F ⊂ G. For any T r ∈ Q [−2R, 2R], let Fr = F + r = {f + r : f ∈ F }. Claim IV: Fr T T Fs = φ if r 6= s, r, s ∈ Q [−2R, 2R]. 6 MINGFENG ZHAO If r 6= s, r, s ∈ Q T [−2R, 2R], but Fr T Fs 6= φ. We say z ∈ Fr T Fs . So z − r ∈ T F , and z − s ∈ F , let z − r = f1 , z − s = f2 , which implies z = r + f1 = s + f2 . So r − s = f2 − f1 ∈ Q. By the definition of F , we know f1 = f2 , so r = s, contradiction. Therefore, we have Fr r 6= s, r, s ∈ Q T Claim V: G ⊂ S T Fs = φ if [−2R, 2R]. r∈Q T [−2R,2R] Fr ⊂ [−4R, 4R]. For any x ∈ G, by the definition of F , there exists y ∈ F such that x ∼ y, that is x − y ∈ Q. Since T x, y ∈ G ⊂ [−R, R], then −2R ≤ x − y ≤ 2R, so x − y ∈ Q [−2R, 2R], hence x ∈ Fx−y . By the arbitrary of x ∈ G, we conclude that G ⊂ S On the other hand, since for any x ∈ S r∈Q T r∈Q [−2R,2R] Fr . [−2R,2R] Fr , we have x = y + r for some y ∈ G, r ∈ T T Q [−2R, 2R]. But we know y, r ∈ [−2R, 2R], then x = y + r ∈ [−4R, 4R]. By the arbitrary of x∈ S r∈Q T [−2R,2R] , therefore, we get S r∈Q T [−2R,2R] Fr ⊂ [−4R, 4R]. Claim VI: F is not m-measurable. T If F is m-measurable. By the definition of F , we know for any r ∈ Q [−2R, 2R], Fr is measurable, and m(F ) = m(Fr ). By Claim V, we know 0 < m(E) 2 ≤ m(G) ≤ m(F ) ≤ 8R, and S r∈Q T [−2R,2R] Fr ⊂ [−4R, 4R]. Hence m([−4R, 4R]) ≥ [ m r∈Q T X = r∈Q T r∈Q T m(Fr ) By Claim IV [−2R,2R] X = Fr [−2R,2R] m(F ) [−2R,2R] T Since Q [−2R, 2R] is countable, and m([−4R, 4R]) < ∞, then m(F ) = 0, which contradicts with m(F ) ≥ m(G) ≥ m(E) 2 > 0. Therefore, F is not m-measurable subset of E. SOLUTIONS OF HW2 7 Department of Mathematics, University of Connecticut, 196 Auditorium Road, Unit 3009, Storrs, CT 06269-3009 E-mail address: mingfeng.zhao@uconn.edu