Math 3215 Intro.
Probability & Statistics Summer ’14
Practice Exam 3: Solutions
1. Consider the function f (x, y) = α(x + 2y) if 0 < x, y < 1 and f (x, y) = 0 otherwise. Find a
value α such that f is a joint p.d.f. Is this choice unique? Find the marginal p.d.f.’s of X and
Y . Find the covariance of X and Y . Is it possible to determine whether or not X and Y are
independent only using what you just found out about the covariance?
Solution: The space of X and Y is S = {(x, y) : 0 < x < 1, 0 < y < 1} and we need
RR
S
f (x, y) dydx = 1. Integrating, we have
ZZ Z 1 Z 1
f (x, y) dydx = α(x + 2y) dydx
S 0 0
Z 1 1
=α (xy + y 2 ) 0 dx
Z0 1
1
=α (x + 1) dx = α (x2 /2 + x) 0 = 3α/2.
0
So, α = 2/3. The marginal p.d.f.’s of X and Y are found via integrating:
Z 1 Z 1
1
f1 (x) = f (x, y) dy = .67(x + 2y) dy = .67xy + .67y 2 0 = .67x + .67,
Z0 1 Z0 1
1
f2 (y) = f (x, y) dx = .67(x + 2y) dx = .67x2 + 1.33xy 0 = .67 + 1.33y.
0 0
Recall Cov(X1 , X2 ) = E((X1 − µ1 )(X2 − µ2 )) = E(X1 X2 ) − µ1 µ2 , so to find the covariance
Cov(X, Y ) we will first compute E(XY ), µ1 , µ2 . We have
Z 1Z 1 Z 1Z 1
E(X1 X2 ) = .67xy(x + 2y) dydx = .67x2 y + 1.33xy 2 dydx = .333
Z0 1 0 0 0
µ1 = .67x2 + .67x dx = .55833
Z0 1
µ2 = .67y + 1.33y 2 dy = .77833
0
So, Cov(X, Y ) = (1/3) − (5/9)(11/18) = −1/162 = −.0062. We know that if X1 and X2
are independent, then Cov(X1 , X2 ) = 0. Thus, X and Y must not be independent, since
Cov(X1 , X2 ) 6= 0. That is, Cov(X, Y ) 6= 0 =⇒ X and Y are independent.
2
2
2. State the Central Limit Theorem. A soda company has installed a new machine to fill its
bottles. The machine was poorly installed, so the amount the machine dispenses has a vari-
ance (changed from standard deviation) of 0.1 liters, and the mean is 2 liters. Approximate
the probability that after 40 bottles are filled that the the sample mean is at least 1.9 liters.
Solution: The Central Limit Theorem says that if X̄ is the mean of a random sample of size
√
n coming from a distribution with mean µ and variance σ 2 , and Z = (X̄ − µ)/(σ/ n), then
Z is approximately standard normal N (0, 1), and that the approximation is better for large
values of n.
p
We set Z = (X̄ − 2)/( .1/40) and then P (X̄ ≥ 1.9) = P (Z ≥ 2) ≈ 1 − .9772 = .0228 .
3. The cumulative distribution function of a continuous random variable is given by F (x) =
1 − e−3x , x > 0. Find the p.d.f. of this random variable.
Solution: The p.d.f. is F 0 (x) = f (x) = 3e−3x , x > 0.
4. Cars arrive at a toll booth at a rate of 4 calls every 6 minutes. Assume these cars arrive as a
Poisson process. What is the probability that the 5th car arrives at exactly 6 minutes and 45
seconds?
Solution: The probability that the cars arrive at exactly any time is zero.
2
3
5. Let X and Y be random variables on the space S = {(0, 0), (1, 1), (1, −1), (2, 0)} with joint
p.m.f. f (x, y) = 1/4. Compute the covariance and correlation coefficient. Are X and Y
independent? Can you tell if they are independent only from the correlation coefficient?
Solution: We compute Cov(X, Y ) = E(XY ) − µ1 µ2 and ρ = Cov(X, Y )/(σ1 σ2 ). We have,
X
µ1 = xf (x, y) = (0 + 1 + 1 + 2)/4 = 1,
S
X
µ2 = yf (x, y) = (0 + 1 − 1 + 0)/4 = 0,
S
X
E(XY ) = xyf (x, y) = (0 + 1 − 1 + 0)/4 = 0.
S
So, Cov(X, Y ) = 0 − 1 · 0 = 0 and ρ = 0/(σ1 σ2 ) = 0. It is not possible to tell if X and Y are
independent simply by knowing that Cov(X, Y ) = 0. In fact, by observing that X = 0 implies
Y = 0, we see that X and Y are clearly not independent (but from class if Cov(X, Y ) = 0
implies that X and Y are independent).
6. Let X1 , . . . , X8 be a random sample from a distribution having p.m.f. f (x) = (x + 1)/6,
x = 0, 1, 2. What is the p.m.f. of Y1 = X1 + X2 ? What is the p.m.f. of Y2 = X3 + X4 ? What
about Y = X1 + X2 + X3 + X4 and W = X1 + . . . + X8 ?
Solution: We have that the p.m.f. of Y1 = X1 + X2 is g(y) = 4i=0 f (i)f (y − i) = f (0)f (y) +
P
f (1)f (y −1)+. . . f (4)f (y −4), and note that many of these terms are often zero. In particular,
g(0) = f (0)f (0) = 1/36
g(1) = f (0)f (1) + f (1)f (0) = 4/36
g(2) = f (0)f (2) + f (1)f (1) + f (2)f (0) = (3 + 4 + 3)/36 = 10/36
g(3) = f (1)f (2) + f (2)f (1) = 12/36
g(4) = f (2)f (2) = 9/36.
Clearly, Y2 = X3 + X4 has the same p.m.f. as that of Y1 and they are independent random
variables. So Y = X1 + X2 + X3 + X4 = Y1 + Y2 has the p.m.f. h(y) = 8i=1 g(i)g(y − i), where
P
4
again many of the terms might be zero depending on the choice of y ∈ {0, . . . , 8}. We have
h(0) = g(0)g(0) = 1/64
h(1) = g(0)g(1) + g(1)g(0) = 8/64
h(2) = g(0)g(2) + g(1)g(1) + g(2)g(0) = (10 + 16 + 10)/64 = 36/64
h(3) = g(0)g(3) + g(1)g(2) + g(2)g(1) + g(3)g(0) = (2 · 12 + 2 · 40)/64 = 104/64
h(4) = g(0)g(4) + g(1)g(3) + g(2)g(2) + g(3)g(1) + g(4)g(0) = (2 · 9 + 2 · 48 + 10)/64 = 124/64
h(5) = g(1)g(4) + g(2)g(3) + g(3)g(2) + g(4)g(1) = (2 · 36 + 2 · 120) = 312/64
h(6) = g(2)g(4) + g(3)g(3) + g(4)g(2) = (2 · 90 + 12)/64 = 192/64
h(7) = g(3)g(4) + g(4)g(3) = 216/64
h(8) = g(4)g(4) = 81/64
Clearly, the random variable Y 0 = X5 + X6 + X7 + X8 has the same p.m.f. as that of
Y , and Y, Y 0 are independent. So, the p.m.f. of W = X1 + X2 + . . . + X8 = Y + Y 0 is
p(w) = 16
P
i=1 h(i)h(w − i).
We can also unwrap all the definitions to write the p.m.f. of W as follows.
16
X
p(w) = h(i)h(w − i)
i=1
16 8
! 8
!
X X X
= g(j)g(i − j) g(k)g(w − i − k)
i=1 j=1 k=1
16 X
X 8 X
8
= g(j)g(i − j)g(k)g(w − i − k)
i=1 j=1 k=1
16 X
X 8 4
X
= f (l)f (j − l)f (m)f (i − j − m)f (n)f (k − n)f (o)f (w − i − k − o)
i=1 j,k=1 l,m,n,o=1
