E_EOR1_IDS VU Amsterdam
2024-2025 Yicong Lin
P-I, W-4 Sep 26, 2024
Exercise 1
Prove that the following functions, F, G, H, I, J, are CDFs.
1. F (x) = 1/2 + 1/π tan−1 (x), x ∈ R;
−1
2. G(x) = 1 + e−x , x ∈ R;
3. H(x) = exp − exp(−x) , x ∈ R;
4. I(x) = 1 − e−x for x ∈ (0, ∞) and I(x) = 0 otherwise;
5. For some ϵ ∈ (0, 1),
1−ϵ
, x < 0,
−x
1 + e
J(x) =
ϵ + 1 − ϵ ,
x ≥ 0.
1 + e−x
Proof Three conditions to be verified:
(a) F is non-decreasing;
(b) F (x) ∈ [0, 1], ∀x ∈ R, limx→−∞ F (x) = 0, limx→+∞ F (x) = 1;
(c) limx→x0 F (x) = F (x0 ), ∀x0 ∈ R.
1. Clearly, F is strictly increasing and continuous. Moreover,
1 1 −1
lim F (x) = lim + tan (x)
x→−∞ x→−∞ 2 π
1 1
= + lim tan−1 (x)
2 π x→−∞
1 1 π
= + − = 0,
2 π 2
and
1 1 −1
lim F (x) = lim + tan (x)
x→+∞ x→+∞ 2 π
1 1
= + lim tan−1 (x)
2 π x→+∞
1 1 π
= + = 1.
2 π 2
Since F is strictly increasing, F must be bounded by 0 and 1.
1
2. G is strictly increasing and continuous. Clearly,
lim G(x) = 0, lim G(x) = 1.
x→−∞ x→+∞
3. H is strictly increasing and continuous. Moreover,
−x −x
lim H(x) = lim e−e = lim e−(limx→−∞ e ) = 0.
x→−∞ x→−∞ x→−∞
Similarly, limx→+∞ H(x) = 1.
4. I is non-decreasing, bounded, and continuous.
5. J is strictly increasing, bounded, and right-continuous.
The formal proof of the results relies on the definitions of the three properties. ■
Exercise 2
Suppose the CDF FX of an r.v. X is given by
0, x < 0,
FX (x) = c0 x, 0 ≤ x < 1,
2
1, x ≥ 1,
where c0 is some constant and FX is continuous. Find the value of c0 and compute P(0.3 < X ≤ 0.7).
Solution: Since FX is continuous, we have
lim FX (x) = FX (1) ⇐⇒ c0 /2 = 1 ⇐⇒ c0 = 2.
x↑1
Therefore, P(0.3 < X ≤ 0.7) = FX (0.7) − FX (0.3) = 0.4. ■
Exercise 3
Suppose that the CDF FX of a discrete r.v. X is given by
0, x < 0,
1/3, x ∈ [0, 1/2),
FX (x) =
x ∈ [1/2, 3/4),
1/2,
1, x ≥ 3/4.
Plot the CDF and determine the PMF of X.
2
Solution: We have the PMF
1/3, x = 0,
1/6, x = 1/2,
pX (x) =
1/2, x = 3/4,
0,
otherwise.
The CDF is plotted as follows.
FX (x)
1
1/2
1/3
x
1/2 3/4 1
Exercise 4
If p1 and p2 are PMFs, show that p = λp1 + (1 − λ)p2 is also a PMF, where p(x) = λp1 (x) + (1 − λ)p2 (x) for
x ∈ R, and λ ∈ [0, 1].
Proof Note that p(x) = λp1 (x) + (1 − λ)p2 (x) ≥ 0, ∀x ∈ R. Moreover,
X X X X
p(x) = [λp1 (x) + (1 − λ)p2 (x)] = λ p1 (x) + (1 − λ) p2 (x) = 1.
x x x x
Therefore, p is a PMF.
P P P
Note: Strictly speaking, an infinite sum (an + bn ) cannot always be split into an + bn . However, if
P P P P
both an and bn converge, then this holds. Since p1 (x) = 1 and p2 (x) = 1, we can therefore split
the sum. ■
Exercise 5
An appliance store receives a shipment of 30 microwave ovens, 5 of which are defective. The store manager
randomly selects 4 ovens, without replacement, to test for defects. Let X represent the number of defective
ovens found. Compute the PMF and CDF of X, and plot the CDF.
Solution: We have
5 25
k 4−k
pX (k) = 30
, k = 0, 1, . . . , 4.
4
P
By definition, FX (x) = k≤⌊x⌋ pX (k). We omit the plot here. ■
3
Exercise 6
There are 3 boxes. The first box contains 1 white ball and 4 black balls, the second box contains 2 white
balls and 3 black balls, and the third box contains 3 white balls and 2 black balls. A box is selected at
random, and then 3 balls are drawn from it. Let X represent the number of white balls drawn.
(a) Find the PMF pX of X;
(b) Compute the probability that the number of white balls is at least 2.
Solution: (a) Let Bi = { the i-th box is selected }, i = 1, 2, 3. By the law of total prob.,
3
!
4
X 1 3 1 1
P(X = 0) = P (X = 0 | Bi ) P (Bi ) = 5
+ 5 + 0 = ,
3 3 3
6
i=1
3
!
4 2 3 3
X 1 1× ×1 1
P(X = 1) = P (X = 1 | Bi ) P (Bi ) = 5
2 + 1
5
2 + 1
5
= ,
3 3 3 3
2
i=1
3 3
3
2 !
X 1 1× 1 2 3
P(X = 2) = P (X = 2 | Bi ) P (Bi ) = 0+ 5
+ 5
1 = ,
3 3 3
10
i=1
3
!
X 1 1 1
P(X = 3) = P (X = 3 | Bi ) P (Bi ) = 0+0+ 5
= .
3 3
30
i=1
k 0 1 2 3
pX (k) 1/6 1/2 3/10 1/30
(b) P(X ≥ 2) = pX (2) + pX (3) = 1/3. ■
Exercise 7
Suppose there are 10 identical electrical components, 2 of which are defective. During assembly, a component
is selected at random. If it is defective, it is discarded and another component is selected. This process
continues until a non-defective component is found. Compute the expected number of defective components
drawn before obtaining a non-defective one.
Solution: Let X be the number of defective components drawn before obtaining a non-defective one. Let
Ai = {the ith component is non-defective}. Note that X takes values in {0, 1, 2}.
8 4
P(X = 0) = P (A1 ) = = ,
10 5
2 8 8
P(X = 1) = P A∁1 ∩ A2 = · = ,
10 9 45
∁ ∁
2 1 1
P(X = 2) = P A1 ∩ A2 ∩ A3 = · = .
10 9 45
4
k 0 1 2
pX (k) 4/5 8/45 1/45
Therefore,
2
X 4 8 1 2
E(X) = k · pX (k) = 0 · +1· +2· = .
5 45 45 9
k=0
■
Exercise 8
For which values of the constant C do the following define valid PMFs on Z+ ?
2k
(a) pX (k) = C ;
k!
(b) pX (k) = C 2−k .
P+∞ P+∞ 2k
Solution: (a) k=1 pX (k) = C k=1 = C(e2 − 1) = 1 =⇒ C = (e2 − 1)−1 ;
k!
P+∞ P+∞ −k 1/2
(b) k=1 pX (k) =C k=1 2 =C· = C =⇒ C = 1. ■
1 − (1/2)
Exercise 9
Is it generally true that E(1/X) = 1/E(X)? Is it ever true that E(1/X) = 1/E(X)? Provide a brief
explanation or counterexample to support your answer.
Solution: No. it is generally not true that E(1/X) = 1/E(X). For instance, consider
k −1 1
pX (k) 1/2 1/2
1 −1
1 1 1
Then, E(1/X) = (−1)−1 · + 1−1 · = 0. But (EX)−1 = −1 · + 1 · = +∞. Therefore,
2 2 2 2
E X −1 ̸= (EX)−1 .
However, in some special cases, E X −1 = (EX)−1 . For instance, consider
k −1 1/2 2
pX (k) 1/9 4/9 4/9
Then
1 4 4 1 8 2
E X −1 = (−1)−1 · + (1/2)−1 · + (2)−1 · = − + + = 1.
9 9 9 9 9 9
5
Moreover,
1 1 4 4
E(X) = (−1) · + · + 2 · = 1,
9 2 9 9
implying that (EX)−1 = 1 = 1/E(X). ■
Exercise 10
An urn contains n balls numbered 1, 2, . . . , n. We randomly remove k ≤ n balls without replacement and
sum their numbers. Find the expected value of the total sum.
Solution: Let Xi be the number shown on the ith ball. Therefore, the expected value of the total sum is
k k
!
X X 1 1 1 1 n(n + 1) k(n + 1)
E Xi = EXi = k · · 1 + · 2 + ··· + · n = k · · = .
n n n n 2 2
i=1 i=1
Exercise 11
Suppose E(X) = Var(X) = λ and E (X − 1)(X − 2) = 1. What is the value of λ?
Solution: Note that
E[(X − 1)(X − 2)] =E X 2 − 3X + 2
=E X 2 − 3E(X) + 2
= Var(X) + (EX)2 − 3E(X) + 2
= λ + λ2 − 3λ + 2
=λ2 − 2λ + 2.
Since E[(X − 1)(X − 2)] = 1 and λ > 0, we obtain λ2 − 2λ + 1 = 0 ⇐⇒ λ = 1. ■
Exercise 12
Suppose E(X) = −2 and E(X 2 ) = 5. Compute Var(1 − 3X).
Solution: Var(1 − 3X) = Var(−3X) = 9 Var(X) = 9 E(X 2 ) − (E(X)2 = 9 5 − (−2)2 = 9.
■
Exercise 13
For any constant c ̸= E(X). Prove that Var(X) < E (X − c)2 .
6
Proof Note that
Var(X) = E(X − EX)2
= E(X − c + c − EX)2
= E (X − c)2 + 2(X − c)(c − EX) + (c − EX)2
= E(X − c)2 − (c − EX)2
< E(X − c)2 ,
because c ̸= E(X). ■
Exercise 14
Suppose X ∼ Bern(p) where p ∈ (0, 1). If E(X) = 3 Var(X), what is the value of P(X = 0)?
Solution: Since X ∼ Bern(p), we know EX = p and Var X = p(1 − p). By E(X) = 3 Var(X), we have
2
p = 3p(1 − p) ⇐⇒ p = ,
3
because p > 0. Therefore,
1
P(X = 0) = 1 − p = .
3
■
Exercise 15
It is known that 20% of customers who reserve seats for 7 p.m. at the restaurant do not show up. The
restaurant has 50 seats at 7 p.m., but 52 reservations have been made. What is the probability that a
customer arrives and no seats are available?
Solution: Let X be the total number of customers, out of the 52 with reservations, who do not show up.
Then, X ∼ Bin(52, 0.2). Therefore,
P(a customer arrives with no seats are available) = P(X ≤ 1)
= P(X = 0) + P(X = 1)
52
= 0.852 + · 0.2 · 0.851 .
1
Exercise 16
Suppose X ∼ Poisson(λ). If P(X = 1) = P(X = 2), what is the value of P(X = 4)?
7
Solution: Since X ∼ Poisson(λ), we have
P(X = 1) = λ −λ
1! e ,
P(X = 2) = λ2 −λ
2! e .
Since P(X = 1) = P(X = 2), we obtain
1 λ>0
λ = λ2 ⇐⇒ λ = 2.
2
Therefore,
24 −2
P(X = 4) = e ≈ 0.0902.
4!
■
