MSO205A PRACTICE PROBLEMS SET 8 SOLUTIONS

Question 1. Let X ∼ Binomial(n, p) for some integer n ≥ 3 and p ∈ (0, 1). Compute EX(X −
1)(X − 2), if it exists.

Answer: If E|X(X − 1)(X − 2)| < ∞, then EX(X − 1)(X − 2) exists. Now,
n
X n!
E|X(X − 1)(X − 2)| = |k(k − 1)(k − 2)| pk (1 − p)n−k
k!(n − k)!
k=0
n
X (n − 3)!
= n(n − 1)(n − 2)p3 pk−3 (1 − p)n−k
(k − 3)!(n − k)!
k=3

= n(n − 1)(n − 2)p (p + (1 − p))n−3

3

= n(n − 1)(n − 2)p3 < ∞.

Hence, EX(X − 1)(X − 2) exists and

n
X n!
EX(X − 1)(X − 2) = k(k − 1)(k − 2) pk (1 − p)n−k = n(n − 1)(n − 2)p3 .
k!(n − k)!
k=0

√
Question 2. Verify that Γ( 12 ) = π.
y2
R∞
Answer: We have Γ( 21 ) = 0 √1x e−x dx. First, we change variables x = 2 and hence,
√ Z ∞ − y2
 
1
Γ = 2 e 2 dy.
2 0
Squaring the above relation and going to polar co-ordinates, we have
  2 Z ∞  Z ∞ 
1 − x2
2 2
− y2
Γ =2 e dx e dy
2 0 0
Z ∞Z ∞
x2 +y 2
=2 e− 2 dx dy
0 0
π
Z ∞ Z 2 r2
=2 e− 2 rdr dθ
r=0 θ=0

=π
√
and hence Γ( 21 ) = π.

Question 3. Let X ∼ N (µ, σ 2 ) for µ ∈ R, σ > 0. Compute EX k for k = 2, 3, 4. [Hint: When

X ∼ N (0, 1), these moments has been computed in the lecture notes.]
1
2 MSO205A PRACTICE PROBLEMS SET 8 SOLUTIONS

X−µ
Answer: Consider the RV Y = σ ∼ N (0, 1). We have already seen that

EY = 0, EY 2 = 1, EY 3 = 0, EY 4 = 3.

Since, X = σY + µ, we have

EX 2 = E(σY + µ)2 = σ 2 EY 2 + 2σµ EY + µ2 = µ2 + σ 2 ,

EX 3 = E(σY + µ)3 = σ 3 EY 3 + 3σ 2 µ EY 2 + 3σµ2 EY + µ3 = µ3 + 3µσ 2 ,

and

EX 4 = E(σY + µ)4 = σ 4 EY 4 + 4σ 3 µ EY 3 + 6σ 2 µ2 EY 2 + 4σµ3 EY + µ4 = µ4 + 6µ2 σ 2 + 3σ 4 .

Note that we have established the existence of MGF of X and hence existence of all momemts EX k
follow.

Question 4. Fix α > 0, β > 0 and let X ∼ Beta(α, β). Compute the MGF of X, if it exists.

Answer: Recall that the p.d.f. of X is given by


 1 xα−1 (1 − x)β−1 , if x ∈ (0, 1)
fX (x) = B(α,β)
0, otherwise.

Since etX is a non-negative RV for all t ∈ R, to check the existence of EetX , we need to check
EetX < ∞. Now,
1 1
et
Z Z
1
EetX = etx xα−1 (1 − x)β−1 dx ≤ xα−1 (1 − x)β−1 dx = et < ∞, ∀t ∈ R.
B(α, β) 0 B(α, β) 0

Therefore, MX (t) = EetX exists for all t ∈ R. The MGF now can be computed by the Maclaurin’s
series expansion around the origin as
∞ k ∞ k
X t X t B(α + k, β)
MX (t) = EX k = .
k! k! B(α, β)
k=0 k=0

Question 5. Let X ∼ Beta(1, 1). Does the distribution of X match with any other distribution
discussed in the lecture notes?

Answer: The p.d.f. of X is given by


1

B(1,1) , if x ∈ (0, 1)
fX (x) =
0, otherwise
R1
with B(1, 1) = 0
dx = 1. Hence, X ∼ U nif orm(0, 1).
 MSO205A PRACTICE PROBLEMS SET 8 SOLUTIONS 3

Question 6. An RV X has the MGF given by the following expressions. Identify the distribution
of X.
(a) MX (t) = (1 − 2t )−3 , ∀t < 2.
(b) MX (t) = 31 e−t + 23 , ∀t ∈ R.

Answer: (a) Recall that an RV Y ∼ Gamma(α, β) with α > 0, β > 0 has the p.d.f.

 1 y α−1 β −α exp(− y ), if y > 0,
fY (y) = Γ(α) β
0, otherwise.

and the MGF

1
MY (t) = (1 − βt)−α , ∀t <
.
β
Since an MGF, if it exists, determines the distribution, we have X ∼ Γ(3, 21 ).
(b) Recall that for a discrete RV Y with support SY and p.m.f. fY , we have
X
MY (t) = ety fY (y).
y∈Sy

Comparing with the given expression for the MGF, we have SX = {−1, 0} and

1
 3 , if x = −1,



fX (x) = 32 , if x = 0,



0, otherwise.

Since fX above is a p.m.f. and an MGF, if it exists, determines the distribution, we have X is
discrete with support SX and p.m.f. fX .

Question 7. Let X be a continuous RV with P(X > 0) = 1 and such that µ01 = EX exists. Prove
that P(X > 2µ01 ) ≤ 12 .

Answer: We have µ01 > 0 (see Question 3, Problem set 5). Then, P(X > 2µ01 ) ≤ 1 0
2µ01 µ1 = 21 .

Question 8. Let x1 , x2 , · · · , xk > 0 be distinct real numbers and let n be a positive integer. Using
Jensen’s inequality discussed in the lecture notes, show that
n
xn + xn2 + · · · + xnk

x1 + x2 + · · · + xk
≤ 1
k k
Answer: Consider the convex function h(x) = xn on [0, ∞). Look at the discrete RV X with
support SX = {x1 , x2 , · · · , xk } and p.m.f.

 1 , if x ∈ SX ,
fX (x) = k
0, otherwise.
4 MSO205A PRACTICE PROBLEMS SET 8 SOLUTIONS

By Jensen’s inequality, we have h(EX) ≤ Eh(X) and hence,

n
xn + xn2 + · · · + xnk

x1 + x2 + · · · + xk
≤ 1
k k

Pk
Question 9. Let x1 , x2 , · · · , xk , p1 , p2 , · · · , pk > 0 be such that i=1 pi = 1. Prove the classical
AM-GM-HM inequality using the AM-GM-HM inequality for RVs discussed in the lecture notes,
k k
X Y 1
xi p i ≥ xpi i ≥ Pk pi
i=1 i=1 i=1 xi

Answer: Consider a discrete RV X with support SX = {x1 , x2 , · · · , xk } and p.m.f.


pi , if x = xi , i ∈ {1, 2, · · · , k},
fX (x) =
0, otherwise.
Pk Qk
Then AM of X is EX = i=1 xi pi , GM of X is eE ln X = i=1 xpi i and HM of X is 1
1
E[ X ]
= Pk 1 pi .
i=1 xi
From the AM-GM-HM inequality proved for an RV X in the lecture notes, we have the requied
inequality.

Question 10. Let (Ω, F, P) be a probability space. Let X = (X1 , X2 , X3 ) : Ω → R3 be a 3-

dimensional random vector. State and prove the non-decreasing property of the joint DF of X.

Answer: For all real numbers a1 < b1 , a2 < b2 , a3 < b3 , the required non-decreasing property is as
follows.
3
X X
(−1)k FX (x) = P(a1 < X1 ≤ b1 , a2 < X2 ≤ b2 , a3 < X3 ≤ b3 ) ≥ 0,
k=0 x∈∆3k
Q3
where ∆3k , k = 0, 1, 2, 3 denote the set of vertices of j=1 (aj , bj ] where exactly k many aj ’s appear.
To prove this, consider the following three sets

A1 := (−∞, a1 ] × (−∞, b2 ] × (−∞, b3 ],

A2 := (−∞, b1 ] × (−∞, a2 ] × (−∞, b3 ],

and
A3 := (−∞, b1 ] × (−∞, b2 ] × (−∞, a3 ].
Then,

P(a1 < X1 ≤ b1 , a2 < X2 ≤ P(A1 ∪A2 ∪A3 )b2 , a3 < X3 ≤ b3 ) = P(X1 ≤ b1 , X2 ≤ b2 , X3 ≤ b3 )−P(A1 ∪A2 ∪A3 ).

The result follows by applying the inclusion-exclusion principle on P(A1 ∪ A2 ∪ A3 ).