Exercises

  1. Does there exist an infinite σ\sigma -algebra which has only countably many members?

  2. Prove an analogue of Theorem 1.8 for nn functions.

  3. Prove that if ff is a real function on a measurable space XX such that {x:f(x)r}\{x: f(x) \ge r\} is measurable for every rational rr , then ff is measurable.

  4. Let {an}\{a_n\} and {bn}\{b_n\} be sequences in [,][-\infty, \infty] , and prove the following assertions:

(a)lim sup(an)=lim infan.(a) \quad \limsup (-a_n) = -\liminf a_n.(b)lim sup(an+bn)lim supan+lim supbn(b) \quad \limsup (a_n + b_n) \le \limsup a_n + \limsup b_n

provided none of the sums is of the form \infty - \infty .

(c) If anbna_n \le b_n for all nn , then

lim infanlim infbn.\liminf a_n \le \liminf b_n.

Show by an example that strict inequality can hold in (b).

  1. (a) Suppose f:X[,]f: X \to [-\infty, \infty] and g:X[,]g: X \to [-\infty, \infty] are measurable. Prove that the sets
{x:f(x)<g(x)},{x:f(x)=g(x)}\{x: f(x) < g(x)\}, \{x: f(x) = g(x)\}

are measurable.

(b) Prove that the set of points at which a sequence of measurable real-valued functions converges (to a finite limit) is measurable.

  1. Let XX be an uncountable set, let M\mathcal{M} be the collection of all sets EXE \subset X such that either EE or EcE^c is at most countable, and define μ(E)=0\mu(E) = 0 in the first case, μ(E)=1\mu(E) = 1 in the second. Prove that M\mathcal{M} is a σ\sigma -algebra in XX and that μ\mu is a measure on M\mathcal{M} . Describe the corresponding measurable functions and their integrals.

  2. Suppose fn:X[0,]f_n: X \to [0, \infty] is measurable for n=1,2,3,n = 1, 2, 3, \dots , f1f2f30f_1 \ge f_2 \ge f_3 \ge \dots \ge 0 , fn(x)f(x)f_n(x) \to f(x) as nn \to \infty , for every xXx \in X , and f1L1(μ)f_1 \in L^1(\mu) . Prove that then

limnXfndμ=Xfdμ\lim_{n \to \infty} \int_X f_n d\mu = \int_X f d\mu

and show that this conclusion does not follow if the condition " f1L1(μ)f_1 \in L^1(\mu) " is omitted.

  1. Put fn=χEf_n = \chi_E if nn is odd, fn=1χEf_n = 1 - \chi_E if nn is even. What is the relevance of this example to Fatou’s lemma?

  2. Suppose μ\mu is a positive measure on XX , f:X[0,]f: X \to [0, \infty] is measurable, Xfdμ=c\int_X f d\mu = c , where 0<c<0 < c < \infty , and α\alpha is a constant. Prove that

limnXnlog[1+(f/n)α]dμ={if 0<α<1,cif α=1,0if 1<α<.\lim_{n \to \infty} \int_X n \log [1 + (f/n)^\alpha] d\mu = \begin{cases} \infty & \text{if } 0 < \alpha < 1, \\ c & \text{if } \alpha = 1, \\ 0 & \text{if } 1 < \alpha < \infty. \end{cases}

Hint: If α1\alpha \ge 1 , the integrands are dominated by αf\alpha f . If α<1\alpha < 1 , Fatou’s lemma can be applied.

  1. Suppose μ(X)<\mu(X) < \infty , {fn}\{f_n\} is a sequence of bounded complex measurable functions on XX , and fnff_n \to f uniformly on XX . Prove that
limnXfndμ=Xfdμ,\lim_{n \to \infty} \int_X f_n d\mu = \int_X f d\mu,

and show that the hypothesis " μ(X)<\mu(X) < \infty " cannot be omitted.

  1. Show that
A=n=1k=nEkA = \bigcap_{n=1}^{\infty} \bigcup_{k=n}^{\infty} E_k

in Theorem 1.41, and hence prove the theorem without any reference to integration.

  1. Suppose fL1(μ)f \in L^1(\mu) . Prove that to each ϵ>0\epsilon > 0 there exists a δ>0\delta > 0 such that Efdμ<ϵ\int_E |f| d\mu < \epsilon whenever μ(E)<δ\mu(E) < \delta .

  2. Show that proposition 1.24(c) is also true when c=c = \infty .