1

PROBLEM - SET 8
Exercise 1. Let X1 ∼ Pois(λ1 ), X2 ∼ Pois(λ2 ) be independent random variables.
Compute E(X1 |X1 + X2 = n).
Solution 1. First we recall that X1 + X2 ∼ Pois(λ1 + λ2 ). Moreover

λ1k −λ2 λ2n−k

pX1 ,X1 +X2 (k, n) = P(X1 = k, X1 +X2 = n) = P(X1 = k, X2 = n−k) = e−λ1 e .
k! (n − k)!

It follows that
  m  n−m
n λ1 λ2
pX1 |X1 +X2 (m|n) = .
m λ1 + λ2 λ1 + λ2
 
Note that pX1 |X1 +X2 (m|n), as function of m, is a density of a Bin n, λ λ+λ
1
. Thus
1 2

λ1
∑ mpX1 |X1 +X2 (m|n) = n λ1 + λ2 ,
m

and therefore
λ1
E(X1 |X1 + X2 = n) = n .
λ1 + λ2
Exercise 2. Let (Xn )n≥1 be independent Be(p) random variables, and N be a Pois(λ )
random variable independent of all the Xn . Set
N
Y = ∑ Xi .
i=1

(a) Determine P(Y = k|N = n) for all 0 ≤ k ≤ n.

(b) Compute the joint density of Y and N.
(c) Show that Y ∼ Pois(pλ ).
(d) Determine the joint density N −Y and Y .
(e) Compute E(N −Y |Y = k) and deduce E(N|Y = k).
Solution 2. (a)

P(Y = k, N = n P(∑ni=1 Xi = k, N = n)
P(Y = k|N = n) = =
P(N = n) P(N = n)
n n
P(∑i=1 Xi = k) P(N = n)
= = P( ∑ Xi = k)
P(N = n) i=1
 
n k
= p (1 − p)n−k ,
k

where we have used the facts that ∑ni=1 Xi ∼ Bin(n, p) and it is independent of
N.
2

(b)
λn
 
n k
pY,N (k, n) = P(Y = k|N = n) P(N = n) = p (1 − p)n−k e−λ .
k n!
(c)

pk k +∞ (1 − p)n−k λ n−k
pY (k) = ∑ pY,N (k, n) = e−λ λ ∑
n k! n=k (n − k)!
pk k λ (1−p) λ k pk
= e−λ λ e = e−λ p ,
k! k!
which is the density of a Pois(λ p).
(d)

(λ p)k (λ (1 − p))m
P(N −Y = m,Y = k) = P(N = m+k,Y = k) = pY,N (k, m+k) = e−λ .
k! m!
Note that this implies that N −Y and Y are independent, respectively Pois(λ (1−
p)) and Pois(λ p).
(e) Thus E(N −Y |Y = k) = E(N −Y ) = λ (1 − p). Finally

E(N|Y = k) = E(N −Y |Y = k) + E(Y |Y = k) = λ (1 − p) + k.

Exercise 3. X and Y form a continuous random vector with joint density

( −y
e
f (x, y) = y for 0 < x < y < +∞
0 otherwise.

Compute E(X 3 |Y = y).

Solution 3. For y > 0
Z y −y
e
fY (y) = dx = e−y ,
0 y
so 1
y for 0 < x < y
fX|Y (x|y) =
0 otherwise.
Finally Z y Z y
1 1
E(X 3 |Y = y) = x3 fX|Y (x|y)dx = x3 dx = y3 .
0 y 0 4
Exercise 4. X and Y form a continuous random vector with joint density
1
fX,Y (x, y) = e−|x−y| e−|y| .
4
a. Compute the conditional density fX|Y and E(X|Y = y).
b. Compute E(X) without computing the density fX .
 3

Solution 4. a.
1 1
Z Z
fY (y) = fX,Y (x, y)dx = e−|y| e−|x−y| dx = e−|y| .
4 2
Thus
1
Z
fX|Y (x|y) = e−|x−y| .
2
By symmetry E(X|Y ) = Y .
b.
E(X) = E[E(X|Y )] = E(Y ) = 0,
as it follows from the fact that the density of Y is an even function.
Exercise 5. X and Y form a normal random vector with mean (0, 1) and covariance
matrix  
41
Σ=
11
a. Compute the distribution of the random variable X + 3Y .
b. Compute the conditional density fX|Y .
c. Compute E(X|Y = y). Hint: note that for y fixed, the conditional density
fX|Y (x|y) is the density of a normal random variable.
Solution 5. a. X + 3Y is normal, being a linear function of a normal random
vector. Moreover
E(X + 3Y ) = E(X) + 3 E(Y ) = 3
Var(X + 3Y ) = Var(X) + 9Var(Y ) + 6Cov(X,Y ) = 4 + 9 + 6 = 19.
So X + 3Y ∼ N(3, 19).
b. Note that
 
1 1
fX,Y (x, y) = p exp − (x, y − 1)Σ −1 (x, y − 1)T .
2π det(Σ ) 2

Moreover  
1 1 −1
det(Σ ) = 3, Σ −1 = ,
3 −1 4
so  
1 1
√ exp − x2 + 4(y − 1)2 − 2x(y − 1) .


fX,Y (x, y) =
2π 3 6
Then, using the fact that Y ∼ N(1, 1)
1√
exp − 16 x2 + 4(y − 1)2 − 2x(y − 1)


fX,Y (x, y) 2π 3
fX|Y (x|y) = = 1 2
fY (y) √1 e− 2 (y−1)
2π
 
1 1 2
=√ exp − [(x − (y − 1)]
2π · 3 6
4

that, as function of x, is the density of a normal of mean y − 1 and variance 3.

So:
c. Z +∞
E(X|Y = y) = x fX|Y (x|y)dx = y − 1.
−∞