1
Department of Statistical Sciences
STA3045F Tutorial 8 March 2, 2022
More on σ-algebras; random variables
Notation: IB is the indicator function of B.
1. For each of the following measurable spaces (Ω, F) and functions X : Ω → R,
i. Describe (or find) F.
ii. Decide whether or not X is a random variable.
iii. Find the sigma algebra generated by X, σ(X).
Also try and give a description of all random variables on (Ω, F) if possible.
(a) Ω = {ω = ω1 ω2 ω3 : ωi = H or T for i = 1, 2, 3} (i.e. the experiment of three coin tosses).
Let F = σ({The first toss is a head} , {All three tosses have the same outcome}) and
X(ω) = number of heads in ω.
(b) Ω = R, F = σ({{q} : q ∈ Q}) and X(ω) = 2ω 3 .
(c) Ω = N, F = σ({e} : e is even) and X(ω) = (mod(ω, 2) − 1)ω. Here mod(x, y) is the
remainder when x is divided by y. For instance, mod(3, 2) = 1, while mod(42, 2) = 0.
(d) Ω = [−2, 2], F = B([−2, 2]) and X(ω) = ωIA (ω) + 4IB (ω), where A = [0, 1] and
B = [−1, 1].
(e) Let Ω = R and F = σ ({finite subsets of R}) and X(ω) = ω.
2. Let (Ω, F) be a measurable space and X : Ω → R be a mapping. Show that if F is generated
by a countable collection of blocks B = {Bn : n ∈ N+ }, then X is a random variable (i.e.,
F/B(R)-measurable) if and only if X is constant on the blocks. That is, if there exist real
numbers {an : n ∈ N+ } such that
∞
X
X(ω) = an IBn (ω), ω ∈ Ω.
n=1
3. (a) Let Ω be a set and X : Ω → R be a function. Show that
σ(X) := X −1 (B) : B ∈ B(R)
is a sigma algebra on Ω.
(b) Show that if X has a countable range {an : n ∈ N+ }, then σ(X) = σ({X −1 ({an }) : n ∈ N+ }).
(c) Show that if F is a sigma algebra on Ω and X : Ω → S is a function to a set S, then
A ⊆ S : X −1 (A) ∈ F
is a sigma algebra on S.
4. In this question we prove The Monotone Class Theorem (MCT).
Theorem 0.0.1 (Monotone Class Theorem). If A is a π-system, then σ(A) = λ(A).
Note that you can’t use the π − λ Theorem since we used MCT to prove it.
(a) First show that λ(A) ⊆ σ(A) using Lemma 2.2.1 in the notes.
2
(b) We now show the reverse inclusion σ(A) ⊆ λ(A).
i. Explain why it is enough to show that λ(A) is a π-system.
ii. To this end, let A ∈ A and define
DA := {B ∈ λ(A) : A ∩ B ∈ λ(A)} .
Show that DA is a λ-system that contains A. Conclude that DA = λ(A).
iii. Now pick A ∈ λ(A) and define
DA := {B ∈ λ(A) : A ∩ B ∈ λ(A)} .
Again show that DA is a λ-system contain A and conclude that DA = λ(A).
iv. Conclude that σ(A) ⊆ λ(A).
5. We prove the following theorem:
Theorem 0.0.2. Let Ω be a set, A ⊆ 2Ω be a π-system on Ω and F := σ(A) be the sigma
algebra generated by A. Let µ1 , µ2 : F → R̄ be two measures on F with µ1 (Ω) = µ2 (Ω) < ∞.
If µ1 (A) = µ2 (A) for every A ∈ A, then µ1 (A) = µ2 (A) for every A ∈ F. That is, if µ1 and
µ2 are finite measures that agree on a π-system generating F, then they agree everywhere
on F.
(a) Let
D := {A ∈ F : µ1 (A) = µ2 (A)} .
Explain why it is enough to show that D = F. This technique is often called the
“Principle of Appropriate Sets”.
(b) Show that D is a λ-system containing A.
(c) Conclude that D = F.
