M15 Probability II Problems Sheet Gen.

Fns

Generating functions
The moment generating function MX (t) of a random variable X is defined via

def
X EX k
tX
MX (t) = E e ≡ tk .
k!
k≥0

If X takes values in {0, 1, 2, . . . }, one also considers the (probability) generating

function of X given by
def
X n
GX (s) = E sX ≡ s P(X = n) .
n≥0

Notice that GX (s) is always finite (and analytic) for |s| < 1, whereas MX (t)
might not be defined for t 6= 0.
GF 1. Find the generating functions, both ordinary G(s) and moment M (t), for
the following discrete probability distributions:
(a) the distribution describing a fair die;
(b) the distribution describing a die that always comes up 3;
(c) the uniform distribution on the set {n, n + 1, n + 2, . . . , n + k};
(d) the binomial distribution on {n, n + 1, n + 2, . . . , n + k};
GF 2. Let a random variable X with values in {0, 1, 2} have moments µ1 = 1 and
µ2 = 3/2.
(a) Find its probability generating function G(s) and the probabilities P(X = k)
for k ∈ {0, 1, 2};
(b) find the moment generating function MX (t) and the first six moments of X.
GF 3. Let X be a discrete random variable with values in {0, 1, 2, . . . , n} and
moment generating function M (t). Find, in terms of M (t), the generating functions
for:
a) − X , b) X + 1 , c) 3X , d) aX + b .
GF 4. Let X be a random variable with generating function GX (s). In terms of
GX (s),
a) find the generating functions of the random variables X + 1 and 2X;
P
b) find the generating function of the sequence P(X ≤ n) = k≤n P(X = k);
c) find the generating function of the sequence P(X = 2n).
1


Hint: Which sequence has generating function 2
GX (s) + GX (−s) ?

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF14

M15 Probability II Problems Sheet Gen. Fns

GF 5. Let the variables X and Y satisfy

1 2
P(X = 1) = P(Y = 5) = , P(X = 4) = P(Y = 2) = ;
3 3
(a) show that X and Y have the same first and second moments, but not the
same third and fourth moments;
(b) find the probability and moment generating functions for X and Y .
GF 6. Let X1 , X2 , . . . , Xn be an independent trials process, with values in {0, 1}
and mean µ = 1/3. Find the ordinary and moment generating functions for the
distribution of
a) S1 = X1 , b) S2 = X1 + X2 , Sn = X1 + · · · + Xn ,
c)
√
d) An = Sn /n , e) Sn∗ = (Sn − nµ)/ nσ 2 ,

where σ 2 is the variance of X.

GF 7. A biased coin is tossed n times, each outcome is independent of all the
others, and on each toss a head is shown with probability p. The total number of
heads shown is X. Use the probability generating function of X to find:
(a) the mean and the variance of X;
(b) the probability that X is even;
(c) the probability that X is divisible by 3.
GF 8. A random variable X has geometric distribution with parameter p, p ∈ (0, 1),
ie., P(X = k) = p(1 − p)k−1 , for k = 1, 2, . . . .
a) Find the generating function of X, its mean and variance.
b) Find the generating function of the sequence P(X > k).
GF 9. Fix 0 ≤ λ ≤ 1, and consider two probability generating functions G1 (s) and
G2 (s). Show that λG1 (s) + (1 − λ)G2 (s) is a probability generating function, and
interpret this result.
GF 10. Two magic dice are thrown independently, one showing a random number
from the set D1 = {1, 3, 4, 5, 6, 8} and another from D2 = {1, 2, 2, 3, 3, 4}.
a) Find the generating functions of the outcomes for both dice. Let T be the
sum of two results; find the generating function of T .
b) Do the same for a pair of standard dice, ie., compute the generating func-
tion of the sum of two independent outcomes taken uniformly in D0 =
{1, 2, 3, 4, 5, 6}. Compare your results.
GF 11. You have two fair tetrahedral dice whose faces are numbered 1, 2, 3, and 4.
Show how to renumber the faces so that the distribution of the sum is unchanged.

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF15

M15 Probability II Problems Sheet Gen. Fns

GF 12. Let Xn , n ≥ 1 be a sequence of iid (independent, identically distributed)

random variables with values in {0, 1, 2, . . . } and common generating function
GX (s). Let N ≥ 0 be an integer-valued random variable, independent of the
sequence Xn ; denote its generating function by GN (s). The sum
def
SN = X1 + X2 + · · · + XN

of random number of Xk ’s has a so-called compound distribution.

a) Use the partition theorem for expectations to find E(SN ) in terms of EX and
EN ;
b) Use the partition theorem for expectations to find Var(SN ) in terms of EX,
VarX, EN , and VarN ;


c) Show that the generating function GSN (s) of SN is GSN (s) ≡ GN GX (s) ;
d) Use the previous result to compute E(SN );
e) Use the result in c) to compute Var(SN ).
GF 13. Let p be the probability distribution on {0, 1, 2} with p(1) = 31 , p(2) = 23 ,
and let pn = p ? p ? · · · ? p be the n-fold convolution of p with itself.

(a) Find p2 by direct calculation;

(b) Find the generating functions G(s) and G2 (s) for p and p2 , and verify that

2
G2 (s) = G(s) .
(c) Find Gn (s) from G(s).
(d) Find the first two moments, and hence the mean and variance, of pn from
Gn (s). Verify that the mean of pn is n times the mean of p.

(e) Find those integers j for which pn (j) > 0 from Gn (s).
GF 14. Imagine a diligent janitor who replaces a light bulb the same day as it
burns out. Suppose the first bulb is put in on day 0 and let Xi be the lifetime of
the ith light bulb. For simplicity, we assume that the individual lifetimes Xi are iid
rv’s in {1, 2, . . . } having a common distribution with generating function GX (s).
In this interpretation Tk = X1 + X2 + · · · + Xk is the time the kth light bulb burns
out; we put T0 = 0. Let Rn be the event “a light bulb was replaced in n days”, so
that R0 ≡ {T0 = 0} ≡ Ω (a certain event).
Pn


a) Show that for every n ≥ 1 we have P(Rn ) = k=1 P X1 = k P Rn−k ;
b) Deduce that the characteristic function GR (s) of the sequence P(Rn ) equals

−1
GR (s) = 1 − GX (s) .

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF16

M15 Probability II Problems Sheet Gen. Fns

GF 15. A mature individual produces immature offspring according to the proba-

bility generating function F (s).
a) Suppose we start with a population of k immature individuals, each of which
grows to maturity with probability p and then reproduces, independently of
the other individuals. Find the probability generating function of the number
of immature individuals at the next generation.
b) Find the probability generating function of the number of mature individuals
at the next generation, given that there are k mature individuals in the parent
generation.

c) Show that the distributions in a) and b) above have the same mean, but not
necessarily the same variance.
[Hint: You might prefer to first consider the case k = 1, and then generalise.]

GF 16. In a sequence of Bernoulli trials, let X be the number of trials up to and

including ath success.

a) Show that  
r − 1 a r−a
P(X = r) = p q , r ≥ a.
a−1

b) Verify that the probability generating function for this sequence is

pa sa (1 − qs)−a

and deduce that EX = a/p and VarX = aq/p2 .

c) Show how X can be represented as a sum of a independent random variables,
all with the same distribution. Use this representation to derive again the
mean and the variance of X.

GF 17. Two players play a game called heads or tails. In this game, a coin coming
up heads with probability p is tossed consecutively. Each time a head comes up
Player I wins 1 pence, otherwise she loses 1 pence. Let Xk ∈ {−1, 1} denote the
outcome of the kth trial, and let Sn , n ≥ 0, be the total gain of Player I after n
trials, Sn = X1 + · · · + Xn , where different outcomes are assumed independent.
Let T be the moment when Player I is first in the lead, ie.,

Sk ≤ 0 for k < T , and ST = 1 .

(a) find the generating function GT (s) of T ;

(b) for which values of p is T a proper random variable, ie., when P(T < ∞) = 1?

(c) compute the expectation ET of T ;

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF17

M15 Probability II Problems Sheet Gen. Fns

GF 18. A slot machine operates so that at the first turn the probability for the
player to win is 1/2. Thereafter the probability for the player to win is 1/2 if she lost
at the last turn, but is p < 1/2 if she won at the last turn. If un is the probability
that the player wins at the nth turn, show that for n > 1
1  1
un + − p un−1 = .
2 2
Show that this equation also holds for n = 1, if u0 is suitably defined, and find un .
GF 19. A flea randomly jumps over non-negative integers according to the follow-
ing rule: a coin is flipped; if it shows “tails”, the flea jumps to the next integer (ie.,
k 7→ k + 1); if the coin shows “heads”, the flea jumps over the next integer (ie.,
k 7→ k + 2). Let un be the probability that, starting at the origin, the flea visits n
at some point.
a) if the coin is fair, show that un = (un−1 + un−2 )/2, compute the generating
function Gu (s) of the sequence un , and thus derive a formula for un ;
b) do the same for a biased coin showing “heads” with probability p ∈ [0, 1].
GF 20. A biased coin is tossed repeatedly; on each toss, it shows a head with
probability p. Let W be the number of tosses until the first occasion when r
consecutive tosses have shown heads. Find the probability generating function of
the random variable W .
GF 21. In a multiple-choice examination, a student chooses between one true and
one false answer to each question. Assume that the student answers at random,
and let N be the number of such answers until she first answers two successive
questions correctly. Show that E(sN ) = s2 (4 − 2s − s2 )−1 . Hence, find E(N ) and
P(N = k). Now find E(N ) directly.
GF 22. In a sequence of Bernoulli trials with success probability p, let un be the
probability that the first combination S(uccess)F (ailure) occurs (in that order) at
trials number n − 1 and n. Find the corresponding generating function, mean and
variance.
GF 23. A fair coin is tossed n times. Let un be the probability that the sequence
of tosses never has “head” followed by “head”. Show that
1 1
un = un−1 + un−2 .
2 4
Find un , using the condition u0 = u1 = 1. Compute u2 directly and check that
your formula gives the correct value for n = 2.
GF 24. A biased coin is tossed N times, where N is a Poisson random variable
with parameter λ. Let H be the number of heads and let T be the number of tails
observed in this experiment. Use generating functions to show that H and T are
Poisson random variables. Show that H and T are independent, and find the mean
and the variance of H − T .

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF18

M15 Probability II Problems Sheet Gen. Fns

5
Optional problems
GF 25.~ A loaded die may show different faces with different probabilities.
a) Is it possible to renumber two fair dice so that their sum is equally likely to
take any value between 2 and 12 inclusive?
b) Can one load two traditional cubic dice in such a way that the sum of their
scores is uniformly distributed on {2, 3, . . . , 12}?
c) Can the sum of three loaded dice have the same distribution as the sum of
three fair dice?
GF 26.~ Successive cars pass at instants Xi seconds apart, i ≥ 1. You require b
seconds to cross the road. If the random variables Xi (i ≥ 1) are independent and
identically distributed, find your expected waiting time until you cross. Evaluate
this when fX (k) = qpk−1 , k ≥ 1.
GF 27. ~ Let fn be the number of paths on the integer lattice Z2 of length n
starting from (0, 0), with steps of the type (1, 0), (−1, 0), or (0, 1), and never
intersecting themselves. For instance, f2 = 7, as shown in the figure below:

Show that X 1+s

fn sn =
1 − 2s − s2
n≥0

and deduce that

1h √
n+1 √
n+1 i
fn = 1+ 2 + 1− 2 .
2
Notice that without the restriction that the path doesn’t self-intersect, there are 3n
paths with n steps. With the restriction, the number has been reduced from 3n to
√
roughly (1 + 2)n = (2.414 . . . )n .
GF 28. ~ A biased coin is tossed N times, where N is a random variable with
finite mean. Show that if the numbers of heads and tails are independent, then N
is Poisson.
[Hint: You might want to use the fact that all continuous solutions of the equation
f (x + y) = f (x)f (y) take the form f (x) = eλx for some λ.]
GF 29.~ Show that for each integer n, the function
(s + n − 1)(s + n − 2) . . . s/n!
is the probability generating function of some random variable X. Show that
EX/ log n → 1 as n → ∞.
5 For fun, not for exam!

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF19

M15 Probability II Problems Sheet Gen. Fns

GF 30.~ Let G = (V, E) be a connected graph with uniformly bounded degree

of its vertices (ie., for some d ∈ N, every x ∈ V is connected to at most d other
vertices). Fix x0 ∈ V , n ∈ N, and consider
( )
0 0 0 0
def G = (V , E ) ⊂ G : |V | = n, x 0 ∈ V
Gx0 ,n = 0 ,
G − connected

the collection of all n-vertex connected subgraphs of G containing x0 . Let ϕK (s)

be the generating function of the sequence
 
Kn = Kn (x0 ) = maxGx0 ,n 

where the maximum is taken over all graphs G satisfying the conditions above.
Show that for all s ≥ 0 we have

d
ϕK (s) ≤ s 1 + ϕK (s) ;

thus deduce that the generating function ϕK (s) is finite for all complex s satisfying
|s| < s0 with s0 > 0 small enough, and derive the estimate

1  dd n
Kn ≤ .
d − 1 (d − 1)d−1

GF 31.~ For n ∈ N, let gn denote the number of k-element subsets of the set
(k)

{1, 2, . . . , n − 1, n} not containing consecutive integers. E.g., for n = 4 and k = 2

(2)
there are only three such subsets: {1, 3}, {1, 4} and {2, 4}, so that g4 = 3. It is
(k)
immediate that gn = 0 if n < 2k − 1. Let Gk (s) be the generating function of
(k) P∞
(k)
the sequence gn , Gk (s) = gn sn .
n=1

(a) Show that for all n, k,

n−2
X n
X
gn−j 1{j≥2} ,
(k) (k)
gn(k+1) = gn−j−1 =
j=1 j=0

ie., g (k+1) is the convolution of g (k) and 1{j≥2} = (0, 0, 1, 1, 1, . . . ); deduce

that Gk (s) = s2k−1 (1 − s)−k−1 .
(k)
(b) Expand Gk (s) to get a closed expression for gn ; can you find a direct proof
of your result?

(c) Use your answer in (b) to compute the probability of not seeing a pair of
consecutive numbers in a (random) draw of the lottery “6 out of 49”. Does
it correspond to your experience? You might wish to compare your answer
with the actual results, see e.g., http://lottery.merseyworld.com/.

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF20

M15 Probability II Problems Sheet Gen. Fns

GF 32.~ For n ≥ 1 let Xk , k = 1, . . . , n, be Bernoulli random variables with

(n)

(n) (n) (n)

P(Xk = 1) = 1 − P(Xk = 0) = pk .
def (n)
Assume that δ (n) = max1≤k≤n pk → 0 as n → ∞ and that
n
X n
X
(n) (n)
pk ≡E Xk → λ ∈ (0, ∞) .
k=1 k=1

P
n
(n)
Using generating functions, show that the distribution of Xk converges to that
k=1
of a Poisson(λ) random variable. This result is known as the law of rare events.
[Hint: You might wish to check that −y 2 ≤ log(1 − y) + y ≤ 0 for all y ∈ [0, 1/2].]

GF 33.~ In n tosses of a biased coin, let Ln be the length of the longest run of
heads, and set πn,r = P(Ln < r). Show that
∞
X
n 1 − pr sr
1+ s πn,r = .
n=1
1 − s + qpr sr+1

GF 34.~ Let the generating function of the family size in an ordinary branching
process be G(s) = 1 − p(1 − s)β , where 0 < p, β < 1. Show that if Z0 = 1, then
n−1 n
E(sZn ) = 1 − p1+β+···+β (1 − s)β .

GF 35.~ A series of objects passes a checkpoint. Each object has (independently)

probability p of being defective, and probability α of being subjected to a check,
which infallibly detects a defect if it is present. Let N be the number of objects
passing the checkpoint before the first defective is detected, and let D be the
number of these passed objects that were defective (but undetected). Find:
(a) the joint p.g.f of D and N ; (b) E(D/N ) .

If the check is not infallible, but errs with probability δ, find the above two quantities
in this case.

O.H. http://maths.dur.ac.uk/stats/courses/ProbMC2H/Probability2H.html GF21