Homework 2 Solutions, Real Analysis I, Fall, 2010. (6) Rudin, chapter 1, problem 5. (a) Solution: To prove A = {x : f (x) < g(x)} is measurable, we would like to say that A = (f − g)−1 ([−∞, 0)), which would be measurable since [−∞, 0) is open and f − g is measurable. Unfortunately, this doesn’t avoid the possibility that f − g is of ∞ − ∞ type, which must be handled separately. To be precise, the remarks in 1.9 show that if f, g are Rvalued functions, we may consider them to be C-valued, and then f − g = f + (−1)g is measurable. This leaves out the case of infinite values, however. The worst case happens when ∞ − ∞ occurs, which is on E = [f −1 (∞) ∩ g −1 (∞)] ∪ [f −1 (−∞) ∩ g −1 (−∞)]. E is measurable since {∞} and {−∞}, as closed points, are Borel sets. Also let F = f −1 ({∞, −∞}) ∪ g −1 ({∞, −∞}) be the set in which either f or g is infinite. Then E ⊂ F and F is measurable also. On F c , f − g is measurable, since f, g are finite and measurable there. Then we may define measurable functions f˜ = f χF c , g̃ = gχF c , h = f˜ − g̃. Note f˜ is measurable since −1 f ((α, ∞]) − F for α ≥ 0, −1 f˜ ((α, ∞]) = f −1 ((α, ∞]) ∪ F for α < 0. Now we may check A = [h−1 ((−∞, 0)) ∪ f −1 (−∞) ∪ g −1 (∞)] − E, which is clearly measurable. The other set {x : f (x) = g(x)} is also measurable, since it is equal to h−1 (0) ∪ E, and {0} is a Borel set. (b) Solution: Let fn : X → R be a sequence of measurable functions. Then we know by Theorem 1.14 that k(x) = lim sup fn (x) and `(x) = lim inf fn (x) are measurable. The set of points where {fn (x)} converge to a finite limit are 1 2 then the set of points in which k(x) = `(x) and k(x) is finite. This is equal to {x : k(x) = `(x)} − k −1 ({−∞, ∞}). This is measurable by part (a) and since {−∞, ∞} is a Borel set. (7) Rudin, chapter 1, problem 10. Solution: Let µ(X) < ∞ and let fn be a sequence of bounded complex measurable functions on X. Assume fn → f uniformly on X. Assume |fn | ≤ Cn . The uniform convergence fn → f means for all > 0, there is an N so that n ≥ N implies |fn (x) − f (x)| < . Note is independent of x ∈ X. So choose and find N as above. Then |f (x)| ≤ |fN (x)| + |fN (x) − f (x)| ≤ CN + for all x ∈ X. Then for n ≥ N , |fn (x)| ≤ |f (x)| + |fn (x) − f (x)| ≤ CN + 2 for all x ∈ X. So for all n = 1, 2, . . . and x ∈ X, we have |fn (x)| ≤ C for C = max{C1 , C2 , . . . , CN −1 , CN + 2}. Now we may apply the Dominated Convergence Theorem, since Z C dµ = Cµ(X) < ∞ =⇒ C ∈ L1 (µ). X This proves lim n→∞ Z X fn dµ = Z f dµ. X To provide a counterexample when µ(X) = ∞, let fn (x) = n1 . Then fn converges uniformly to f (x) = 0, but Z 1 fn dµ = µ(X) = ∞, n X R while X f dµ = 0. (8) Rudin, chapter 1, problem 12. Solution: Let f ∈ L1R(µ). We want to show that for all > 0, there is δ > 0 so that E |f | dµ < whenever µ(E) < δ. We prove the statement by contradiction. If this is not true, then there R is an > 0 and measurable sets En with µ(En ) < −n 2 , but En |f | dµ ≥ . 3 (1) We will apply Fatou’s Lemma to the sequence of positive functions {|f |(1 − χEn ) = |f |χEnc }. Fatou’s Lemma states Z Z (lim inf |f |χEnc ) dµ ≤ lim inf |f |χEnc dµ. X n→∞ n→∞ X First of all, analyze the right hand side. For each n, Z Z Z Z |f |χEnc dµ = |f | dµ − |f | dµ − . |f | dµ ≤ X X En X R Therefore, the right-hand side of (1) is ≤ X |f | dµ − . Now analyze the integrand of the left-hand side of (1). lim inf |f |χEnc = |f | lim inf (1 − χEn ) = |f |(1 − lim sup χEn ). n→∞ n→∞ n→∞ Recall the definition of lim sup to compute lim sup χEn (x) = inf sup χEi (x) k≥1 i≥k n→∞ = inf χAk (x), Ak = k≥1 ∞ [ Ei i=k = χB (x), B= ∞ \ k=1 Ak = ∞ ∞ [ \ Ei . k=1 i=k The Ak and B are measurable, and µ(B) ≤ µ(Ak ) for all k, by set inclusion. Now the measure of Ak is given by µ(Ak ) = µ( ∞ [ i=k Ei ) ≤ ∞ X µ(Ei ) < i=k ∞ X 2−i = 2−k+1 . i=k (This is by assumption on the measure of Ei and Lemma 1 below.) Thus for all k, µ(B) ≤ µ(Ak ) < 2−k+1 . This implies µ(B) = 0. Putting this together, we have the left-hand side of (1) satisfies Z Z c (lim inf |f |χEn )dµ = |f |(1 − lim sup χEn )dµ X n→∞ ZX Zn→∞ = |f | dµ − |f |χB dµ X X Z Z = |f | dµ − |f | dµ X B Z = |f | dµ, X 4 since B has measure 0. Putting this together with the righthand side of (1) then shows Z Z |f | dµ ≤ |f | dµ − , X X R which is a contradiction since we assume X |f | dµ < ∞. Finally, we need the following Lemma: Lemma 1. Let Ai be measurable sets. Then ∞ ∞ [ X µ( Ai ) ≤ µ(Ai ). i=1 i=1 Proof. Of course, this is an equality if the Ai are disjoint sets. Define B1 = A1 , B2 = A2 − A1 , B3 = A3 − (A1 ∪ A2 ), . . . . Then Bi ⊂ Ai implies µ(Bi ) ≤ µ(Ai ), and ∞ ∞ [ [ Ai = Bi . i=1 i=1 So compute ∞ ∞ ∞ ∞ [ [ X X µ( Ai ) = µ( Bi ) = µ(Bi ) ≤ µ(Ai ). i=1 i=1 i=1 i=1