6 Coursework Exercises 2

  • C2.1

    Consider a duel. Players A and B take it in turns to shoot at each other. Player A shoots first, and has probability a of hitting player B. At his turn, player B has probability b of shooting player A. The winner is the first player to hit the other.

    Let N be the number of shots in the duel. Show that the pgf of N is given by

    G(z)=az+b(1-a)z21-z2(1-a)(1-b).

    Hence calculate the mean of N.

    [3]

  • C2.2

    The pgf of the time to the first success in a Bernoulli process with success probability p is

    G(z)=pz1-(1-p)z.

    Let X be the time until the first success in a sequence of Bernoulli trials with success probability 1/3, and Y the time until the first success in a sequence of Bernoulli trials with success probability 2/3. Write down the pgf for T=X+Y, assuming the two sequences of Bernoulli trials are independent.

    Using the same approach as in 3.2.2 in the notes, show that if pi=(T=i) then for i>2:

    9pi-9pi-1+2pi-2=0

    Solve this difference equation (using the correct initial conditions) to obtain pi.

    [5]

  • C2.3

    Consider the pgf of N in question 1 of this sheet. Denote this pgf by G(z).

    Calculate 12(G(1)-G(-1)). How does this relate to the probability that player A wins the duel (see quiz sheet 1).

    Let pi=(N=i). By expanding the probability generating function:

    G(1)=p0+p1+p2+, and G(-1)=p0-p1+p2-,

    explain why the relationship between 12(G(1)-G(-1)) and the probability that player A wins holds.

    [2]

Challenge question

Full marks may be obtained for correct solutions to the previous questions. A further 5 marks, up to maximum total mark of 10, can be gained by successfully answering the following.

  • C2.4

    For the rv T, the time to ruin starting from X0=1 in a random walk with p=12, show using the pgf of T that

    P(T=2m+1)=(2m)!m!(m+1)!(12)(2m+1).