Instructions: You may only consult the textbook and lecture notes: do not consult any
other materials or other persons. Reference any results you use from the textbook or lecture
notes. Solutions must be produced using LaTeX and submitted as hardcopy. Handwritten
solutions will not be accepted.
1) Let S and T be independent random variables that each have the uniform distribution on
the interval [−1, 1].
Show that conditional on the event {R ≤ 1} the two random variables
are independent standard normal random variables.
[10 points]
2) Let X and Y be independent nonnegative random variables with density functions f and
g that are continuous on (0, ∞). Suppose for any u > 0 that the conditional distribution of
X given X + Y = u is uniform on the interval [0, u].
Show that X and Y are identically distributed and that their common distribution
is exponential.
[10 points]
Hint: Write h for the density function of X + Y . Explain why the density function of
(X, X +Y ) is given by fX,X+Y (x, u) = 1
u
h(u) for 0 ≤ x ≤ u. Then explain why 1
x+y
h(x+y) =
f(x)g(y) for x, y ≥ 0. By considering what happens when x = 0 and y = 0, show that f = g.
Conclude that f(x+y)f(0) = f(x)f(y) for x, y ≥ 0. Define a function ¯f by ¯f(z) = f(z)/f(0).
Show that ¯f(x + y) = ¯f(x)
¯f(y) for x, y ≥ 0 . Conclude that ¯f(z) = e
−λz when z ≥ 0 for
some λ > 0. Lastly, observe that f(x) = λe−λx when x ≥ 0.
3) Given two n×n matrices A = (aij ) and B = (bij ), their Schur product is the n×n matrix
C = (cij ) with i, j entry cij = aij bij for 1 ≤ i, j ≤ n. Suppose that Σ0 and Σ00 are two n × n
nonnegative definite matrices with Schur product Σ.
Show that Σ is also nonnegative definite.[10 points]
Hint: Explain why we may assume that there are independent random vectors X0 and X00
such that X0 and X00 both have mean vector 0, X0 has variance-covariance matrix Σ
0, and
X00 has variance-covariance matrix Σ
00. Then consider the variance-covariance matrix of the
random vector X, where Xi = X0
X00ifor 1 ≤ i ≤ n.
4) Alice and Bob are having a snowball fight. The times at which Alice throws her snowballs
are the arrival times of Poisson process with (constant) intensity λ. The times at which Bob
throws his snowballs are the arrival times of Poisson process with (constant) intensity µ.
Assume that these two Poisson processes are independent. Consider the event that in the
time interval [0, T] Alice and Bob throw the same non-zero number of snowballs and that
they alternate throws (for example, one way this could happen is that they each throw two
snowballs and the sequence of throws is Alice, Bob, Alice, Bob).
What is the conditional probability given this event that Alice throws the first
snowball?
[10 points]
Hint: Use the coloring theorem to build the Poisson processes for Alice and Bob from a single
Poisson process with intensity λ + µ.
5) Suppose that the random variable U has the uniform distribution on [0, 1] and that
conditional on U = u the random variable X has the binomial distribution with number of
trials n and success probability u.
Show using generating functions or otherwise that P{X = x} =1n+1 for x ∈{0, 1,. . . , n}.
[10 points]
6) Let 0 = S0, S1, S2, . . . be a simple random walk on the integers with p = q=12. Recall
that p0(n), n ≥ 0, is the probability that the walk is at position 0 at time n and f0(n),
n ≥ 1, is the probability that the walk first returns to position 0 at time n.
Put a(0) = 1, a(2k) = P{S1 6= 0, S2 6= 0, . . . , S2k 6= 0} for k ≥ 1.
Show that a(2k) = p0(2k) =2kk(12)2k for k ≥ 0.
7) Let 0 = S0, S1, S2, . . . be a simple random walk on the integers with 0 < p < q < 1.
Say that the time n ≥ 0 is an upcrossing time for the random walk if Sn < Sn+1 (that is,
Sn+1 = Sn + 1). Let T−1 = min{n ≥ 0 : Sn = −1} be the first time that the random walk
visits −1.
Find the probability generating function for the number of upcrossing times that
occur before time T−1.
[10 points]
Hint: Put Z0 = 1 and define Zm for m ≥ 1 by
Zm = |{0 ≤ n < T−1 : Sn = m − 1, Sn+1 = m}|;
that is, Zm for m ≥ 1 is the number of times the random walk makes an upcrossing from
state m − 1 to state m prior to the time T−1. Show that Z0, Z1, Z2, . . . is a branching process
with P{# of offspring = k} = pkq for k ≥ 0.