Professional Documents
Culture Documents
p1-Introduction-p2-Combinatorial Probability
p1-Introduction-p2-Combinatorial Probability
p1-Introduction-p2-Combinatorial Probability
§1 Introduction
1.1 Many real-life problems deal with variability or random phenomena, e.g.
1.2 Probability and statistics — science or art of understanding and analysing variability and
uncertainty.
1. Subjective —
a measure of “chance” as perceived by a person, i.e. how likely that person “feels” that a
designated event may occur. As more observed data are made available, one might adjust
his/her perception from time to time to give the measure a more objective content.
2. Objective —
a measure of chance from a frequentist viewpoint, i.e. long-term proportion of times that a
designated event occurs if the observation process “were” replicated independently many
many times.
1
§2 Combinatorial probability
§2.1 Selection of distinguishable objects
2.1.1 A simple rule:
n 1 ways to do job 1
n2 ways to do job 2
Total k jobs → .. ⇒ n1 × n2 × · · · × nk ways to do the k jobs
.
n ways to do job k
k
2.1.2 Definition. For any integer n ≥ 0, define the factorial of n, written n!, by
(
1, n = 0,
n! =
n(n − 1) · · · 1, n ≥ 1.
2
Equivalent problem: allocate r distinguishable objects into n distinguishable cells, with no
cell containing > 1 objects.
Example: Beauty Contest, 15 candidates, how many ways to award Winner, 1st and 2nd runners-
up?
15 × |{z}
Answer: |{z} 14 × |{z}
13 = 2730
Winner 1st 2nd
Special case: If r = n, each possible arrangement is known as a permutation.
3
Equivalent problem: allocate r distinguishable objects into n distinguishable cells, with no
restriction on no. of objects allocated in each cell. [ e.g. allocating 1st and 2nd objects into jth cell
is equivalent to selecting jth object at 1st and 2nd times. ]
Example: Number lock with 4 digits, each digit may be set to 0, 1, . . . , 9. There are 104 = 10000
different settings.
(1 + x + x2 + · · · )(1 + x + x2 + · · · ) · · · (1 + x + x2 + · · · ) = (1 + x + x2 + · · · )n = (1 − x)−n .
Picking jth object s times in our selection is equivalent to picking the term xs in the jth
bracket in above expansion of (1 − x)−n .
◦ ◦ | ◦ | | ◦ ◦ | ··· | ◦ ◦ ◦
Each such sequence may be regarded as an allocation of r objects into n cells (2 in 1st cell, 1
in 2nd, 0 in 3rd etc.) Thus,
4
Example: 4 distinct food items (burger, fries, cake, ice cream), how many different ways to order
a pack of 3 items?
Answer: 20
2.1.8 Summary
Putting ith object into group j is equivalent to picking pj from the ith bracket.
The answer is
n n!
= ,
n1 , . . . , n k n1 ! n2 ! · · · nk !
which is known as a multinomial coefficient.
Example: consider word “PEPPER”. How many ways to rearrange letters P, E, P, P, E, R into a
6-letter word?
Problem equivalent to partitioning the n = 6 letter positions into k = 3 groups, labelled by ‘P’, ‘E’ and
‘R’ respectively. The 3 groups have sizes n1 = 3, n2 = 2, n1 = 1 respectively. Answer: 60.
(a) In a peace talk on the Middle East, 4 delegates come from the UN, 3 from USA, 5 from Russia
and 2 from Syria. They are to be seated in a row. Americans, Russians and Syrians must be
separated by the UN delegates, and delegates representing the same country must be seated
next to each other. How many ways to seat them?
Answer: 4! 3! 5! 2! × (5 · 4 · 3) = 2, 073, 600.
5
∗(b) Similar to (a), but with the 14 delegates seated around a circular table. How many ways to
seat them?
Answer: 4! 3! 5! 2! × (3 × 2) × 14 = 2, 903, 040.
(c) A lift can reach 5 different floors. How many exit patterns are there for 8 persons inside the
lift?
5+8−1 12
Answer: 58 = 390, 625, or = = 495 if the persons are identical robots!
8 8
∗(d) Mary has 5 dresses, 3 skirts, 4 blouses, 3 pairs of shoes, and 2 hats. She always wears shoes
and either a dress or a blouse and a skirt. She may or may not wear a hat.
(i) How many different combinations can she wear?
(ii) Suppose Mary can afford to buy either an extra dress or an extra hat (but not both).
Which should she buy if she decides to maximise the number of different combinations
that she can wear?
(iii) Suppose Mary’s yellow shoes do not match her black or green dress, and that the black
dress does not match her two hats. How many matching combinations can she wear?
§2.2 Definitions
2.2.1 Definition. Consider an experiment with a finite number of different possible outcomes. The
set Ω of all possible outcomes is called the sample space.
2.2.2 An event A is a subset of Ω. It occurs if and only if the outcome of the experiment belongs to
the subset A.
2.2.3 Definition. Assume that each outcome in Ω is equally likely, i.e. has equal chance of occur-
ring. The probability of event A is defined by
no. of outcomes in A
P(A) = .
no. of outcomes in Ω
2.2.4 Example.
If we roll a dice, possible outcomes are Ω = {1, 2, 3, 4, 5, 6}.
The event A = {2, 4, 6} refers to that “an even score turns up”.
The probability of getting an even score is
3 1
P(A) = = .
6 2
6
2.2.5 In the above definition, it is important that all outcomes in Ω must be equally likely, such that
for each ω ∈ Ω,
1
P({ω}) = .
no. of outcomes in Ω
Counter-example. When we toss a coin, it can land on a head, a tail OR on its edge. We
may define Ω = {H, T, Edge}. The above definition instructs that
§2.3 Examples
2.3.1 If we roll 2 dice, what is the probability that the total is 6?
Here Ω = {(1, 1), (1, 2), . . . , (1, 6), (2, 1), . . . , (6, 6)} contains 6 × 6 = 36 outcomes. The event of
interest is
A = {(1, 5), (2, 4), (3, 3), (4, 2), (5, 1)},
A problem of statistical inference: N unknown, k known and r observed; pick N such that
P(A) is maximum; this is an example of maximum likelihood estimation of the total number
of fish in the lake.
7
2.3.3 In a game of bridge, what is the probability that North and South each gets 2 aces?
Take
Ω =collection of partitions of 52 cards into 3 sets of sizes 13, 13, 26. Then Ω contains
52
outcomes.
13, 13, 26
Take A = collection of partitions of 52cards
into 3 sets
of sizes 13, 13, 26 such that the first 2 sets
4 48
contain 2 aces each. Then A contains outcomes. Thus,
2 11, 11, 26
4 48 52
P(A) = .
2 11, 11, 26 13, 13, 26
An alternative game is suggested as follows. de Méré wins if at least one double six appears
in 24 throws of a pair of dice. Who has the better chance to win?
Now Ω contains 3624 outcomes. The event that you win has 3524 outcomes. Thus
Remark:
de Méré argued that in the first game he would experience 4 out of 6 (i.e. 2/3) of all possible outcomes,
one of which (“6”) is favourable. In the second game, again one of the 36 possible outcomes is favourable
and with 24 throws he would experience 24 out of 36 possible outcomes, the proportion is again 2/3.
He thus concluded that the second game is equally favourable to him. What’s wrong with de Méré’s
reasoning?
2.3.5 Toss 2 coins; we can get either 2 heads, 2 tails, or 1 head and 1 tail. So,
Correct?
8
§2.4 ∗ ∗ ∗ More challenges ∗ ∗ ∗
2.4.1 (a) One fork, two identical chopsticks and three identical spoons are to be put into two baskets,
made of bamboo and plastic respectively, such that each basket contains exactly three objects.
Show that there are altogether six different ways of distributing the objects.
(b) One Communist, two Socialists and three Democrats are forming a coalition government in
the form of a legislative council and an executive council, such that each council consists of
three members. Suppose that the assignment of members to the two councils is completely
random, so that each member has the same chance of sitting in either council. An observer
wants to calculate the probability that no Communists and Democrats sit in the same council.
He argues as follows.
“According to result (a), there are 6 different ways of distributing government people
to two councils. The Communist and the Democrats can be separated in exactly 2
ways. Therefore the probability required is given by the proportion 2/6 ... ”
2.4.2 If n people are seated randomly in a row of 2n seats, what is the probability that no two persons
will sit on adjacent seats?