MTH309A PROBLEM COLLECTION

APRIL 5, 2019

1. Classes of sets, measures and measurable functions

Question 1. Suppose that a collection F of subsets of Ω has the following properties.
(1) ∅ ∈ F.
(2) If A, B ∈ F, then A ∩ B ∈ F.
(3) If A ∈ F, then Ac is a finite disjoint union of members of F.
Consider another collection C consisting of finite disjoint union of members of F. Show that C is a
field of subsets of Ω.
Remarks: (not included in the exercise) The collection of left-open right-closed intervals in R is an
example of such an F. Recall that we consider intervals of the form (a, ∞), a ∈ R to be left-open
right-closed. This collection C was mentioned in the class as an example of a field, but not a σ-field.
Question 2. Let f : Ω → Ω0 be a function and let F be a σ-field in Ω0 . Show that f −1 (F) :=
{f −1 (A) : A ∈ F} is a σ-field in Ω.
Question 3. Let f : Ω → Ω0 be a function and let F be a collection of subsets of Ω0 . Show that
σ(f −1 (F)) = f −1 (σ(F)),
where f −1 (A) denotes the pre-image of A ⊂ Ω0 , f −1 (F) := {f −1 (A) : A ∈ F } and f −1 (σ(F)) :=
{f −1 (A) : A ∈ σ(F)}. (Hint: Use the principle of good sets.)
Question 4. Say that a σ-field is countably generated (or separable) if it can be generated by some
countable collection of sets. Show that the Borel σ-field BR on R is countably generated.
Question 5. Let (Ω, F, P) be a Probability space. Fix A ∈ F. Show that
C := {B ∈ F : P(A ∩ B) = P(A)P(B)}
is a Monotone class.
Question 6. Suppose that F is a collection of subsets of Ω with the following properties: Ω ∈ F
and A, B ∈ F implies A \ B := A ∩ B c ∈ F. Show that F is a field.
Question 7. Let (Ωi , Fi ), i = 1, 2 be measurable spaces and let A be a collection of subsets of Ω2
which generates F2 . Assume that f : Ω1 → Ω2 satisfies f −1 (A) ∈ F1 , ∀A ∈ A. Show that f is
measurable.
Question 8. Show that all continuous maps f : Rn → Rm are Borel measurable. (Hint: We have
discussed the case f : R2 → R. Generalize the argument.)
Question 9. Let {fn } be a sequence of R valued Borel measurable functions defined on Ω.
(a) Show that the functions ω 7→ supn fn (ω) and ω 7→ inf n fn (ω) are R̄ valued Borel measurable
functions.
1
2 MTH309A PROBLEM COLLECTION APRIL 5, 2019

(b) Show that the functions ω 7→ lim supn fn (ω) and ω 7→ lim inf n fn (ω) are R̄ valued Borel
measurable functions.
Question 10. Let λ denote the Lebesgue measure on (R, BR ). For any Borel set A and x ∈ R, define
A + x := {a + x : a ∈ A}. Show that A + x is again a Borel set and that λ is translation invariant,
i.e.
λ(A) = λ(A + x), ∀x ∈ R, A ∈ BR .
Question 11. Let {An } be a sequence of events in (Ω, F, P). Prove Boole’s inequality, i.e.
!
[ X
P An ≤ P(An ).
n n

Question 12. Let µ and ν be two probability measures on (R, BR ) with the same distribution
function. Show that µ = ν.
Question 13. Let f and g be R valued Borel measurable functions on (Ω, F). Fix A ∈ F and define
the function h : Ω → R by (
f (ω), ω ∈ A,
h(ω) =
g(ω), ω ∈ Ac .
Show that h is also Borel measurable.
Question 14. Suppose that a collection F of subsets of Ω is both a π system and a λ system. Show
that F is a σ-field.
Question 15. Let X, Y : Ω → R be Borel measurable. Assume that Y is σ(X)/BR measurable.
Show that there exists a Borel measurable function f : R → R such that Y = f ◦ X.
Remark: (not included in the exercise) If X : Ω → R and f : R → R are measurable, then so is
f ◦ X. This is the converse.
Question 16. Let {An } be a sequence of subsets of Ω. For all ω ∈ Ω, prove the following statements.
lim inf 1An (ω) = 1lim inf n→∞ An (ω),
n→∞
and
lim sup 1An (ω) = 1lim supn→∞ An (ω).
n→∞

2. Integration of measurable functions, related inequalities

Question 17. Let f, g be Borel measurable, integrable functions such that
Z Z
f dµ ≤ g dµ, ∀A ∈ F.
A A
Show that f ≤ g, µ-a.e..
Question 18. Let {fn } be a sequence of real valued measurable functions on a measure space
(Ω, F, µ) such that Z
sup |fn |1+ dµ < ∞
n Ω
for some  > 0. Show that the sequence is uniformly integrable, i.e. the following holds
Z
lim sup |fn | dµ = 0.
x→∞ n (|fn |>x)
 MTH309A PROBLEM COLLECTION APRIL 5, 2019 3

Question
R 19. LetR f : R → R be Borel measurable and fix a ∈ R. Consider the two integrals
‘ R f (x) dx’ and ‘ R f (x − a) dx’ with respect to the Lebesgue measure. If one integral exists, show
the existence of the other. In this case, show that the two integrals are actually equal.
Question 20. Let {Xn } be a sequence of non-negative, integrable random variables on (Ω, F, P).
Show that !
X X
E Xn = EXn .
n n
Does the result hold if EXn = ∞ for some n?
Question 21. Prove the following version of Markov inequality. Let f : Ω → R̄ be Borel measurable.
Then for any A ∈ F and c > 0 show that
Z
1
µ({|f | ≥ c} ∩ A) ≤ |f | dµ.
c A
Question 22. Construct a probability space (Ω, F, µ) and a real valued integrable function f on
this space such that f is not bounded.
Question 23. Fix a, b ∈ R with a < b. Suppose that there exist functions f : (a, b) × Ω → R and
g : Ω → R such that
(1) ω ∈ Ω 7→ f (t, ω) is µ-integrable for every fixed t ∈ (a, b),
(2) t ∈ (a, b) 7→ f (t, ω) is continuous for every fixed ω ∈ Ω,
(3) g ≥ 0 and µ-integrable,
(4) |f (t, ω)| ≤ g(ω), ∀t, ω.
Then show that the function h : (a, b) → R defined by
Z
h(t) := f (t, ω) dµ(ω)
Ω

is continuous.
Question 24. Fix a, b ∈ R with a < b. Suppose that there exist functions f : (a, b) × Ω → R and
g : Ω → R such that
(1) ω ∈ Ω 7→ f (t, ω) is µ-integrable for every fixed t ∈ (a, b),
(2) t ∈ (a, b) 7→ f (t, ω) is differentiable (in t) for every fixed ω ∈ Ω,
(3) g ≥ 0 and µ-integrable,
∂
(4) | ∂t f (t, ω)| ≤ g(ω), ∀t, ω.
Then show that the function h : (a, b) → R defined by
Z
h(t) := f (t, ω) dµ(ω)
Ω
0 ∂
R
is differentiable with h (t) = Ω ∂t
f (t, ω) dµ(ω).
Question 25. In class, we have seen the correspondence between the set of Lebesgue-Stieltjes mea-
sures on (R, BR ) and the set of distributions functions on R. Later, we started proving the same
correspondence in R2 . Complete the argument (you may follow the arguments as in dimension
one).
4 MTH309A PROBLEM COLLECTION APRIL 5, 2019

(a) (from measures to distributions functions) Let µ be a Lebesgue-Stieltjes measure on (R2 , BR2 ).
Define F : R2 → R by
F ((x1 , x2 )) := µ({(a, b) ∈ R2 : −∞ < a ≤ x1 , −∞ < b ≤ x2 }).
Show that F is a distribution function on R2 , i.e. it is ‘right-continuous’ and ‘non-decreasing’
(as per the definition given in class).
(b) (from distributions functions to measures) Let F be a distribution function on R2 . Con-
struct a Lebesgue-Stieltjes measure on (R2 , BR2 ), by extending the set function µ such that
µ((x1 , y1 ] × (x2 , y2 ]) := F (y1 , y2 ) − F (x1 , y2 ) − F (y1 , x2 ) + F (x1 , x2 ).
Remark: (not included in the exercise) Compare the previous relation with the inclusion-exclusion
principle.
Question 26. Let f, g : (Ω, F) → (R, BR ) be measurable functions. Fix real numbers p, q ∈ (1, ∞)
such that p−1 + q −1 = 1. Then, using Young’s inequality, prove the following version of Hölder’s
inequality, viz.
Z Z 1/p Z 1/q
|f g| dµ ≤ |f |p dµ |g|q dµ .
Ω Ω Ω

Question 27. Let (Ω, F, P) be a measurable space. Suppose P1 and P2 are two probability measures
on this space. Show that that the following conditions are equivalent.
(1) For any A ∈ F, P1 (A) = 0 implies P2 (A) = 0.
(2) For every  > 0, there exists δ > 0 such that if A ∈ F with P1 (A) < δ, then P2 (A) < .

3. Various notions of convergence of random variables and probability measures

Question 28. Find examples of random variables X, X1 , X2 , · · · on some probability space (Ω, F, P)
such that the indicated properties hold.
L1 (P)
(1) Xn −−−−→ X, but Xn 9 X a.s.
n→∞
a.s.
(2) Xn −−−−→ X, but Xn 9 X in L1 (P).
n→∞
in probability
(3) Xn −−−−−−−−→ X, but Xn 9 X in L1 (P).
n→∞
in probability
(4) Xn −−−−−−−−→ X, but Xn 9 X a.s.
n→∞
n→∞
Question 29. Let {xn } be a sequence of real numbers. Prove that δxn ⇒ δ0 if and only if xn −−−−→ 0.
Question 30. Let X, X1 , X2 , · · · be random variables on some probability space (Ω, F, P). Prove
the following equivalences.
in probability |Xn −X| n→∞
(a) Xn −−−−−−−−→ X ⇐⇒ E 1+|X n −X|
−−−−→ 0.
n→∞
in probability n→∞
(b) Xn −−−−−−−−→ X ⇐⇒ E (|Xn − X| ∧ 1) −−−−→ 0.
n→∞
in probability n→∞
(c) Xn −−−−−−−−→ X ⇐⇒ inf{ > 0 : P(|Xn − X| > ) < } −−−−→ 0.
n→∞
Note: One can now formulate metrics on the set of random variables, the convergence in which
are equivalent to the convergence in probability. These metrics are sometimes referred to as the Ky
Fan metrics.
 MTH309A PROBLEM COLLECTION APRIL 5, 2019 5

Question 31. Let X1 , X2 , · · · be random variables on some probability space (Ω, F, P). Suppose
in probability
Xn ⇒ c, where c ∈ R is some constant. Show that Xn −−−−−−−−→ c.
n→∞

Question 32. Let {Xn } be a sequence of real valued random variables on a probability space (Ω, F),
such that the sequence converges in p-th mean to a random variable X. Show that there exists a
subsequence {Xnk } converging a.s. to X.
R R
Question 33. Let ν and µ be measures on (Ω, F) with ν  µ. Show that gdν = gf dµ for all
dν
real valued measurable functions g, where f = dµ .
Question 34. Let {Xn } be a sequence of real valued random variables on a probability space (Ω, F),
such that every sequence has a further subsequence converging a.s. to X. Show that {Xn } converges
in probability to X.
Question 35. Let {Xn } be a sequence of real valued random variables on a probability space (Ω, F),
such that every sequence has a further subsequence converges in probability to X. Show that this
fact is equivalent to the fact that {Xn } converges in probability to X.
Question 36. Let F, F1 , F2 , · · · be the distribution functions corresponding to the probability mea-
sures µ, µ1 , µ2 , · · · on (R, BR ). Prove the equivalence between the following two statements. (These
statements were mentioned in class as a part of equivalences (12 & 13) for weak convergence)
(1) limn Fn (x) = F (x), for all continuity points x of F , where Fn (y) := µn ((−∞, y]), F (y) :=
µ((−∞, y]), ∀y ∈ R (corresponding distribution functions).
(2) There exists a dense set D in R such that limn Fn (x) = F (x), ∀x ∈ D.

4. Indenpendence of events and random variables, Characteristic functions

Question 37. Let X and Y be independent random variables defined on a probability space (Ω, F, P).
Further assume that X and Y are square integrable. Show that X, Y, X 2 , Y 2 , XY are integrable
and that
E(XY ) = (EX)(EY ), V ar(X + Y ) = V ar(X) + V ar(Y ).
Question 38. Let X and Y be two real valued random variables defined on the probability space
(Ω, F, P). Prove the following two statements.
(1) X and Y are independent if and only if σ(X) and σ(Y ) are independent.
(2) Given any measurable functions f, g : R → R, independence of X and Y implies the
independence of f (X) and g(Y ).
Question 39. Let X and Y be two real valued, independent random variables defined on the prob-
ability space (Ω, F, P). Suppose that P(X + Y = a) = 1 some real number a. Show that both X
and Y are constant random variables.
Question 40. Let X and Y be two real valued, independent, integrable random variables defined
on the probability space (Ω, F, P). Prove that XY is also integrable.

5. Convergence of series of independent random variables, Law of Large numbers

and Central Limit Theorem