CM32OX: Topics in Mathematics

Game Theory

Andrew Pressley

At first sight. the Theorv of Games might appear to be a trivial pursuit. However. game theory aims to model any situation involving conflic t where each party (or player) has only partial control over the outcome. This includes not only games in the usual sense (e.g. chess, bridge, poker), but also many situations in economics. politics, business and management. In fact, game theory was created by John von Neumann to model problems in economics. Consider the game with two players in which one player chooses a row and simultaneously the other player chooses a column of the matrix /0

fi
\\2

1 0 100

2 100 0

If a is the entry in the chosen row and column. the row player gets La from the column player (if a is negative, the column player receive s L(a) from the row player). Greed might encourage the row player to choose row 2 but then he could lose Li: similarly, the column player might be tempte d to choose column 2 but then he could lose Li. Avoidance by both players of a worse outcome will ensure that the first row and column are chosen. We shall see that a wide class of games can be reduced to a matrix game of this kind (possibly with a different matrix, of course. In general. the analysis will not be so simple, but eventually we shall obtain an algorit hm which allows us to solve all matrix games.

Contents

1 Games and their Trees 1.1 Introduction and Examples 1.2 Game Trees Problems 2 Pure Strategy Solutions 2.1 Pure Strategies 2.2 Examples 2.3 Matrix Games 2.4 Optimal Pure Strategies 2.5 Minimax Theorem (first version) Problems 3 Mixed Strategy Solutions 3.1 Mixed Strategies 3.2 Finding Mixed Strategy Solutions 3.3 Inferior Strategies (again) 3.4 The 2 x n Case 3.5 Convexity 3.6 Proof of Von Neumanns Minimax Theorem Problems 4 Solving Matrix Games in general 4.1 Simple Solutions 4.2 Convexity (again) 4.3 Shapley-Snow Algorithm 4.4 Proof of the Algorithm Problems

1 1 5 7 9 9 9 12 12 15 18 20 20 21 24 28 32 37 41 44 44 48 49 54 .56

Chapter 1: Games and their Trees

We begin by describing the kind of games we shall discuss in this course, how to formulate such games in a precise mathematical way, and how to represent them pictorially.

1.1 Introduction and Examples Before we attempt to formulate a mathematical definit ion of a game, we give a general discussion illustrated by some examples .Any game (or at least any game that we shall consider in this course) is characterized by a set of rules governing the behaviour of certain individuals or groups called players. An n-person game has n players (n 2,3... .). Of course, chess is a 2-person game. but so is bridge since two players in the same team have identical interes ts and so count as a single player in game theory. When the game is played, a sequence of moves are made, the nature of which is specified by the rules. Moves are of two kinds: person al moves, which involve a choice by one of the players of one of a finite set of alterna tives, and chance moves, in which a choice among a finite set of alternatives is made by a chance mechanism, the probabilities of each alternative being specified by the rules. The rules specify which player or chance mechanism is to make each move. For example, all moves in chess are personal moves, while the first move in a game of bridge is the deal, which is a chance move with 52!/(13!) alternatives. Finally, the rules specify which sequences of moves constitute a complete play of the game, and they assign a numerical score, the payoff, to each player for each play of the game. In a 2-person win, lose or draw game such as chess one would normally assign a score of + 1 to the winner and 1 to the loser (and 0 to each player in case of a draw). Example 1.1 In the 2-person game of NIM, one start s with a finite number of piles of matches, with at least one match in each pile. The first player selects one pile and removes at least one match from that pile. He can remove any number of matches from one up to the total number of matche s in the pile. The next player does the same with the remaining piles of matches. This continues until there are no matches left. The last player to remove a match loses. To simplify the discussion, we shall actually consider only the game of (2,2)NIM, which starts with two piles each with two matche s. Let us call the player who makes the first move player 1 and the other player player 2. We represent this game by a game tree, in which the initial state of the matches is represented by a vertex labelled 1 since player 1 makes the first move. Player 1 can either remove one match from one of the piles or both matches from one of the piles. Hence, at the initial vertex there are two possible (personal) moves , leading to states which we represent by

and

where the horizontal line represents one empty pile of matche s. These two new states are represented by two vertices labelled 2, becaus e it is now the turn of player 2. Continuing in this way, we get the game tree shown below.

Ci
......

I)

I 1-li
(- i ,i)

(Ii)

[1 H -I III L I
The vertices corresponding to two empty piles of matches do not have a label as neither player will then be making a move. They are called terminal vertices and represent the end of the game.

(I))

H I

It is very important to note that in general there are several differe nt vertices all of which correspond to the same match situation, Each vertex corresponds not just to the current match situation but to the entire sequen ce of moves of the play of the game that leads to that vertex. The game tree contains everything there is to know about the game. For example, we can see from the game tree that, if player 2 plays correct ly. he will always win no matter what player 1 does. Indeed, if player 1 removes 2 matches at the first move, then player 2 will win by removing 1 match at his first move; if player 1 removes 1 match at his first move, player 2 should remove both matches from the other pile. Unfortunately, riot all games can be analyzed so simply . Example 1.2 The game HI-LO is played with three cards, labelle d 1, 2. 3. Player 1 picks up the three cards, looks at them, chooses one and lays it face down on the table. Player 2 guesses high or low. The card on the table is turned face up; if the card was 3 and player 2 guessed high, or if it is 1 and player 2 guessed low, player 2 gets 3 from player 1; if the card was 3 and player 2 guessed low, or if it was 1 and player 2 guessed high, player 1 gets 2 from player 2. If the card was 2 and player 2 guessed low, player 2 gets 2 from player 1: hut if player 2 guessed high, player 1 must lay one of the two remaining cards face down on the table,

and then player 2 guesses high or low and player 2 gets 1 from player 1 if he is right but loses 3 to player 1 if he is wrong. The game tree is shown below, At the initial vertex, there are three (personal) moves according to which card player 1 selects; at the next three vertice s there are two (personal) moves according to whether player 2 guesses high (H) or low (L), etc. We label each terminal vertex with a pair of numbers representing the payoffs to each player.

(2,-2)

(-1,1) (1,1)

(2,2)

This game differs from NIM in that, after the first move, player 2 does not know which card is on the table. So although player 2 is at one of the vertices labelle d 2 just above the initial vertex, he does not know which. To indicate this we draw a box surrounding these three vertices. Similarly, the next two moves available to player 2 are also enclosed in a box. Since player 1 has complete information about the game at all times, each vertex labelled 1 should be in a separate box. The set of

vertices of the game tree in a given box is called an information set. So this game has four information sets. Our next example
is

of a game that involves chance moves.

Example 1.3 In the game RED-BLACK, player 1 is dealt face down a red or black card, each card being equally likely. Player 1 looks at the card but does not tell player 2 what it is. If the card is red, player 1 can either call for the toss of a fair coin or pass the move to player 2. If the card is black, player 1 must pass the move to player 2. If player 2 gets the move, he has to guess whether the card is red or black. If the coin is tossed, the garrie ends. The game tree is as follows. We label the vertices at which a chance move is made O and label the chance moves themselves with the probability of each alternative. Chance moves do nor belong to any inforthation set.

toss
1

Player 2 pays player 1 Lp, Lq, Lr, Ls, Lt or Lu, depending on the final state of the game as indicated on the game tree. The games in these examples all terminate after a finite number of moves, and in this course we shall only study games which have this property, although there are certainly games which do not have it. For example, it is well known that in chess there are some sequences of moves in the endgame which can be repeated cyclically, arid so could go on for ever. In such cases, however, some rule is normally agreed that causes the game to terminate; for example, it might be agreed in advance that if a cyclical sequence of moves has been repeated a certain number of times, the game is a draw. In this way, games which are potentially infinite are converted into the finite games studied in this course.

1.2 Game Trees

With these examples in mind, we now introduce an abstract mathematical model of a game tree. A game tree (sometimes called a game in extensive form) consists of the following
ten ingredients.

(i) A finite set V, whose elements are called vertices. The vertices of the game tree are identified with the states of the game. (ii) A set M of pairs (vi, v where u v E V and v ), 2 , 1 2 1 ; 2 v the elements of M are called edges. The interpretation is that in = (vi,v9) E Al if there is a move from the state 1 v of the game to state v Then v is called the initial point of m and v its final . 2 1 2 point. A vertex v E V that is not the initial point of any move is called a terminal vertex. These correspond to the states at the end of the game. (iii) A distinguished initial
vertex v 0

E V which is not the final point of any edge.

Of course. v corresponds to the state of the game before play has commenced. 0 (iv) A path in the game tree is a sequence of vertices (v 2 ,v 1 vk) such that, if k> 1, (vi, ui) Al for i = 1 ki. Then u is the initial point of the path, Vk 1 is its final point and k 1 is its length. The rank of the game tree is the maximum of the lengths of all paths in the game tree.

A path whose final point is a terminal vertex corresponds to a (complete) play of the game. The game tree is required to have the following property (it is this property which makes the game tree a tree). (v) For any
v. v

V, there exists a nnique path with initial point

vo

and final point

The players of the game are introduced into our mathematical game tree as follows: (vi) The set V of vertices has a partition into n + 1 disjoint subsets such that V = V U V U 1 U V and is non-empty if i > 0.
..

. 1 V

..

The interpretation is that the game tree is that of an n-player game. 1 being the set of states at which a chance move is made (if any), while V, (for 1 < j < n) is the set of personal moves for player i. (vii) For any v E let rn mk be the edges with initial point ,.. 1 are positive numbers ir 1 1 irk such that in + ir = 1.
. ,

v.

Then, there

...

Of course. ir is the probability with which the chance move rn. is made starting 3 at the state v of the game. It is also useful to partition V in another way. (viii) For j = 1, 2, 3,. precisely j edges.
..,

let A be the set of vertices which are the initial points of

Thus. if v e fl A. with j > 1 and 1 < i < ii. then at the state v of the game player i has to make one of j possible moves. If u E . the j moves with initial j. 4 point v are labelled 1,2 j.

The set V U 1

U V,

0 V\V has yet another partition: non-empty subsets called information then I C fl A for some i, j with are two different vertices in the same point u and final point iv.

(ix) There is a partition of V\V into disjoint 0 sets such that, if I is any information set, i 0. Moreover, we assume that, if v and w information set, there is no path with initial

The final ingredient is the introduction of the payoffs of each player. (x) There are a real-valued functions pr,. vertices. If v is a terminal vertex, state v.
pj(V)
. . ,

p,., defined on the set of terminal

is the payoff to player i if the game ends in the

Remark It is important that the indexing of the moves in (viii) is done properly, respecting the structure of the game, and in particular the information sets. For example. in the game HI-LO, there are two information sets for player 2. In each of these, the moves in which player 2 guesses high must all have the same label.

It might appear that games which involve simultaneous choices by two or more players cannot be described by using a game tree, but this is not the case. For example, consider the game of MATCHING PENNIES. This is a 2-person game in which players 1 and 2 simultaneously show a coin. If both coins are heads or if both are tails, player 1 receives 1 from player 2; otherwise, player 2 receives 1 from player 1. We can interpret this as a game in which player 1 makes the first move and then player 2 makes his move without knowing the result of player ls move. This gives the game tree

(l)

(1.1) T

(1.1)

(1,l)

This shows the importance of information sets, since the game tree

would represent a very different game. in which player 1 shows his coin and then player 2, having seen player 1 s coin, shows his. Obviously, player 2 will always win this latter game. whereas it is clear that there is no strategy that either player can adopt in the first game which will always result in a win.

Problems
1.1. In the game of n < n NOUGHTS AND CROSSES, plaYer 1 places an 0 in an n < n array, then player 2 places an X in one of the n 2 1 unoccupied positions, then player 1 places an 0 in one of the n 2 2 unoccupied positions. etc. The winner of the game is the first to form a complete row or column or diagonal of Os or Xs (and the game is a draw if this has not been achieved by the time the matrix is filled). Draw the game tree of the (completely uninteresting ) game of 2 x 2 NOUGHTS AND CROSSES. (If you have a very large piece of paper you could try the same for the more usual 3 x 3 case.)

1.2. Modify the game of MATCHING PENNIES by keeping the same rules except that the players now toss their coins instead of choosing whether to show heads or tails. Draw the game tree. 1.3. In the game STONE-SCISSORS-PAPER, two players simultaneously call out stone, scissors or paper, the rule being that stone beats scissors beats paper beats stone (and the game is a draw if the players make the same call). Draw the game tree. a game tree for the following game. The first i from {O, 1}. The second move is a chance move (0, 1}, both choices being equally likely. At the third move an integer k from (0, 1} and the game then terminates. If 1; otherwise player 2 pays player I Li.
Draw integer

1-1.

move is the choice by player 1 of an in which an integer j is chosen from player 2, knowing j but not i, selects i +j k = 1, player 1 pays player 2

1.5. Draw game trees for the following four variants of the game considered in Problem 1.4, where the moves and payoffs are the same hut player 2 has the following information when he makes his
move

8 (i) he knows neither i nor j; (ii) he knows both i and j; (iii) he knows z + j but not the separate values of i and j; (iv) he knows the value of the product of i and j but not the separate values of i and j. 1.6. Draw a game tree for the following 4-person game. The first move is a chance move at which an integer .r is chosen from (1, 2}, the probabilities that 1 and 2 are chosen being 1/3 and 2/3, respectively. For the second move, if .r = I player 1 chooses a number p from {1. 2, 3}, while if x = 2 player 2 chooses a number p from {1. 2, 3}. For the third move, if p 1 player 3, not knowing chooses an integer z from {1,2}. while if p 1 player 4, knowing .r but not knowing whether p = 2 or 3, chooses an integer z from {f, 2}. The game then terminates. 1.7. Draw game tree for the following 2-person game. Player 2 consists of two agents A and B. Player I and the agents A and B are isolated and play is conducted by an umpire At the start of the game, the umpire requires player 1 to choose an integer z from {i, 2}. If player 1 chooses 1. the umpire asks A to choose an integer y from {1.2}, while if player 1 chooses 2 the umpire asks B to choose an integer p from {1,2}. Finally , the umpire asks the agent for player 2 who has not yet participated to choose an integer z from {1,2}. Player 1 receives h(xy,z) from player 2, where h is a real-valued function. (In this game, it is important to note that when an agent of player 2 is asked to make a choice he does not know whether he is making the second or third move of the game.)
,
.

1.8. Draw a game tree for the 2-person game which starts with each player being dealt a card from a pack of cards labelled 1,2 and 3. Player 1 then calls high or low to which player 2 responds by calling high or low. The game then termin ates unless player 1 has called low and player 2 has called high, in which case player I mtist make one further move, calling either high or low. Denote the play of the game in which player 1 calls high and player 2 calls low by HL, and denote the other plays similarly. The payoffs are as follows. HH: the player with the higher card receives 2 from the other player; LL: the player with the higher card receives 1 from the other player: HL: player 1 receives 1 from player 2; LHL: player 2 receives 1 from player 1; LHH: the player with the higher card receive s 2 from the other player. 1.9. Show that, in any game tree: (1) there is no edge with final point vu; (ii) if v is a vertex different from v, there is exactly one edge which has v as its final point.

Chapter 2: Pure Strategy Solutions

In this chapter we discuss what it means to solve a game. i.e. to find optimal strategies for each player.

2.1 Pure Strategies Roughly speaking, a pure strategy for a player of a game is a plan made in advance of what he will do in every situation involving him that may arise in the game. (The word pure is added for reasons that will appear later.) The formal definition is as follows. We use the notation introduced in Section 1.2. Definition 2.1 Let Z be the set of information sets for player i in a game tree (i.e. I is the set of information sets I such that I C I. A pure strategy for player i is amapu :I *Nsuch that j(I) E {1,2,... .j} jf I A. The interpretation of this is as follows. Suppose that the game has reached a state v at which player i is to make a move. Let I be the information set containing v, and suppose that I C A, so that there are j possible moves that player i may make. Then, the pure strategy oj dictates that player i chooses the move labelled u(I). The fact that the pure strategy picks a move for each information set, rather than for each vertex of the game, reflects the fact that a player knows only which information set he is in at any stage of the game, and not necessarily which vertex he is at.

2.2 Examples Example 2.2 In (2,2)-NIM, there are five information sets for player 1, each consisting of one vertex, at two of which there are two possible moves and at the other three only one possible move. So there are 2 x 2 x 1 x 1 x 1 = 4 pure strategies for player 1. Similarly, there are 2 x 3 x 1 x 1 = 6 pure strategies for player 2. Example 2.3 In HI-LO, there are two information sets for player 1, one with three possible moves and one with two, giving 3 x 2 = 6 pure strategies for player 1. There are also two information sets for player 2, each with two possible moves, giving 2 x 2 = 4 pure strategies for player 2. Suppose that each player in a game has selected a pure strategy. If there are no chance moves in the game, then clearly the play is completely determined, and hence so is the terminal vertex of the play and hence the payoff to each player. However, if there are chance moves in the game, there will in general he several possible terminal vertices for any given choice of pure strategies for each player. Suppose that in an fl-person game player i selects pure strategy o, i = 1,... n. The probability of arriving at a particular terminal vertex v is the product of the probabilities assigned to each chance move in the unique path in the game tree with initial point v and final point v; denote this probability by 0 ,o).
,

10

Of course, it may he that. for the given choice of pure strateg y for each player. o is not a possible terminal vertex; in that case we put ir(ui a,,) = 0. For example, if there are no chance moves in the game. we shall have x, (a 1 a,,) = 1 if v is the unique terminal vertex determined by a 1 a,,, and ir (aj a,,) = 0 otherwise. Definition 2.4 Let given by

game. The pure pay[f funclion for player i is the map P, :

be the set of pure strategies for player i in an 71-per son x x x R

pi(v)xu(ai

where the sum is over the set of terminal vertices v of the game. Remark This definition follows the usual rules for calcula ting expected values of random variables in probability theory. Recall that, if X is a random variable with a finite number of possible values r N, and if pk is the probability that X = xe, the expected value of X is

E(X)

PkXk.

In this course, we shall be particularly interested in 2-person games. In that case. let us label the pure strategies for each plover, say = = {a1) where ni, n 1. Then, the pure payoff function for player i can conveniently he described by the mYn matrix whose (r. s)-entrv is P, (a, This is called the payoff niatri.v of the game for player i.
, . .

be described as follows:

Example 2.5 In RED-BLACK, each player has two pure strategies, which can
player 1: (1) if red then toss; (2) if red then pass move to player 2. player 2: (1) guess red; (2) guess black.

Suppose. for example, that are three possible terminal of these terminal vertices (1. 1 P 1) (p + q + 2t)/4. player I is

both players adopt their pure strategy (1). Then there vertices, with payoffs p, q or t. The probabilities being reached are 1/4, 1/4 and 1/2. respectively, so You should check that the complete payoff matrix for
(pq2u)/4

((p+q+2f)/4

(r+t)/2

(s+u)/2

Of course. the payoff matrix for player 2 is just the negativ e of this matrix, since for an play of the game the payoff to player 2 is minus that to player 1. Example 2.6 A 2-person game starts with the choice by player 1 of an integer x from {l, 2}. Next, an umpire tosses a coin and inform s player 2 of x if the outcome is heads, but not otherwise. Player 2 then chooses an integer p from {3. 4}. Finally, the umpire selects, by a chance device, one of the integers 1.2 or 3 with probabilities 2/5. 1/5 and 2/5, respectively. If z is the choice at the last move, player 1 receives (i + p + z) from player 2 if ,r + p + z is even, and player 2 receives (x + p + ) from player 1 if x + p + z is odd.

11

__-7

--T

-.7

c.

-7

-?

1/2

0
2
L

The game tree is shown above. We show only the payoff s to player 1 at the terminal vertices, since the payoffs to player 2 are simply the negatives of those to player 1. There is just one information set for player 1 at which there are two possible moves, so player 1 has two pure strategies which we denote by 1 or 2 according to which integer is chosen. For player 2, there are three inform ation sets, at each of which there are fwo possible moves, giving S pure strategies for player 2. We denote them by triples (i.j,k), this denoting the pure strateg which y consists in choosing i if the coin falls heads and player 1 has chosen 1, choosing j if the coin falls heads and and player 2 has chosen 2. and choosing k if the coin fails tails. We list the pure strategies for each player as follows: player 1: 1. 2 player 2: (3,33). (3,3,4), (3.4.3). (4,3.3), (3.4,4), (43,4). (4.4.3) , (4,4.4). You should check (see Problem 2.8) that the payoff matrix for player 1 is (18/s

\ 21/5

3/10 3/10

18/5 3/10

3/10 21/5

3/10 24/5

21/5 3/10

3/10 3/10

21/5 24/5

12

2.3 Matrix Games For the remainder of this course zero-sum, in the following sense.
we

shall only consider 2-person

games

that are

1 Definition 2.7 Let P (i = 1,2) be the pure payoff function for player i in a 0. 2 2-person game. Then, the game is said to be zero-sum if Pi + P
(1) (n) (in) (1) o ) 2 = {a {o Let ) be the sets of pure strategies 2 for players 1 and 2, respectively, in a 2-person game. Then the zero-sum condition is

for all i = 1,... , m, j = 1,... , n. This means that the payoffs to both players are determined by the payoff matrix for player 1; this matrix is called the payoff matrix of the game. Conversely, any rn x ri matrix (a) is the payoff matrix of some 2-person zero sum game, namely the game in which two players 1 and 2 independently select n}, respectively, and player 2 pays {1, 2 integers i E {1. 2,. , m} and j . 2 player 1 a The analysis of 2-person zero-sum games depends only on a knowledge of the payoff matrix. Accordingly, such games are often called matrix games.
. .

2.4 Optimal Pure Strategies We now consider the problem of finding the optimal strategies for each player in a given matrix game with rn x n payoff matrix (a). To simplify the notation, we and 2 simply by 1, 2,. , m shall denote the elements of the pure strategy sets and 1.2,... , n, respectively. If player 1 adopts pure strategy i, his payoff will be at least
. .

mm
1jn

b,

say.

0 Let b he the maximum of the numbers b for 1 < i <m. Then, by adopting pure i.e. not less than , 0 strategy i player 1 can ensure that his payoff is not less than max mm . 3 a
1<i<m 1<j<n

Similarly, if player 2 employs pure strategy min(_aj)

j, his payoff will be at least

so player 2 has a pure strategy which ensures that his payoff is at least max

maxa)

minmaxa

To summarize, player 1 has a pure strategy which ensures that his payoff is at least max mm
I

13

and player 2 has a pure strategy which ensures that player ls payoff is at most mm max
.1

Suppose that these two numbers are equal, i.e. that (1) max mm a 3
i

0 1ninrnax
j
i

say.

Then, player 1 has a pure strategy which gives him a payoff of at least v and player 2 has a pure strategy which prevents player 1 from obtaining a payoff of more than v. These pure strategies would then have to be considered optimal for each player, and the game would be solved. Unfortunately, condition (1) does not always hold: Example 2.8 In the game of MATCHING-PENNIES described at the end of Chapter 1, each player has two pure strategies. namely: 1: choose heads; 2: choose tails. The payoff matrix (a) is /1 and clearly max mma
z
j

1 1 mm max
3

i,

+ 1.

To understand when condition (1) holds, we shall consider a more general situ ation (which will be required later in any case), Definition 2.9 Let X and Y be sets of real numbers and let f: XxU any function, An element (x, y*) e X x Y is called a saddle point of if f

R be

(2)

f(x,y) <f(x,y) <f(x*,y)

for allx

X. yE Y.

Of course, we can regard an rn x n payoff matrix (a) to be a function of this kind by taking X = {1,2.. ,m}. U = {1,2.... ,n} and f(i,j) = Proposition 2.10 Let X and Y be sets of real numbers, let f: Xx Y * R be any function. and assume that maXmiflf(i,y) and miflmaxf() both exist. Then: xX
yE zEX

(i) max mm
zEX

f@ ) YEY f(,, )

, 1 <mmnmaxf( ).
yeY sEX

(ii) max mm

xE.V yE

mm max f(z, y) if and only if

has a saddle point.

Moreover, if (x*, y) is a saddle point of (3)

f,

then

maxminf(x.y) zmiflmaxf(x,y)f(x*y)
yEY sEX

Proof (i) For x E X, let rn(x)

minf(x,y),
YE Y

14

and for y

Y, let M(y)
-

rnaxf(x,).
EA

Then, rn(x) < f(x,y) < M(y) for all x Hence, for all x E X, X, y E Y.

m(x) <minM(y),
EY 5

andso maxm(x) <miflM(y),

as required.
(ii) Suppose that (x*,y*) is a saddle point off. By condition (2),

maxf(x,y*) <f(x*,y*) <minf(x*,y).

yEY

But obviously, minmaxf(x,y) <rnaxf(x,y),

rnaxrninf(z,y) > minf(,),

so minrnaxf(x,y) <f(x*,y*) <maxminf(x,y).

Part (i) now shows that both of these inequalities must be equalities. Conversely, suppose that maxminf(x,y)
zEX yCY

minmaxf(x.y).
yCY xcX

Let x E X, y

Y be such that, in the notation of part (i) of the proof,

m(x*)

maxm(x),
xEX

M(y)

minjw(y).

Then, the assumption says that m(x) f(x, ) <f(*) and so

M(y). But clearly, for all x E X, y E Y,

f( )

>

f(x,y) <f(x*,g). y* gives f(x,y*) < f(x*,y*), and taking x = x* gives f(x,y*) < f(x, y), so (x, y) is a saddle point. Warning If (x* y*) E X x Y is such that-condition (3) holds, it does not follow that (x, y) is a saddle point (see Problem 2.11).
Taking y
=

We now specialize to the case of the payoff matrix (a) of a matrix game. Recall that the indices i and j label the pure strategies of players 1 and 2, respectively.

15

Definition 2.11 Let (a) be the payoff matrix of a matrix game. A pair (j*,j) is called a saddle point of the payoff matrix, or a pure strategy solution of the game. if (4)
< <

for all i,j. A pure strategy i for player 1 is said to be optimal if there is a pure strategy j ) 1 for player 2 such that (j*j is a pure strategy solution of the game. Optimal pure strategies fpr player 2 are defined similarly. Thus, the condition (4) for (j,j*) to be a saddle point is that the entry of the payoff matrix is a largest entry in its column and a smallest entry in its row. Proposition 3.4 tells us that, if (i*,j*) is a saddle point, then
=

maxmina
i j

minmaxa. j z

but as we have remarked above the converse of this statement is false. Example 2.12 The payoff matrix

j4
\2

/2 6

3 4 3 3

1 1 3 2

4 5 2 4

has a saddle point in the (3. 4) position. since the (3. 4)-entry is clearly the smallest in its row and the largest in its column. Inspection shows that there are no other saddle points. Hence, 3 is the unique optimal pure strategy for player 1, and 4 is the unique optimal pure strategy for player 2.

2.5 Minimax Theorem (first version) We have already seen that not all games have pure strategy solutions. There is, however, a simple class of games which do have pure strategy solutions. Definition 2.13 A game is said to be of perfect information if every information set for each player contains only a single vertex. Thus, in a game of perfect information each player knows, at each stage of the game, the precise state of play and all that has gone before. Minimax Theorem 2.14 Every matrix game of perfect information has a pure strategy solution. Proof Let F be a matrix game of perfect information. The proof is by induction on the rank of F (see Chapter 1). Although we could start the induction when the rank off is zero (in which case there is nothing to prove), it is instructive to treat the rank 1 case in detail first. Suppose then that the rank of F is 1. If the initial vertex v off is a chance move, 0 then each player has a unique pure strategy (do nothing!), so there is nothing to prove. Suppose now that v is the initial point of a move for player 1: then player 2 0

16

has a unique pure strategy (do nothing), which is necess arily optimal. Let v 1 v be the final points of the moves with initial point j; thus. vi r are also the terminal vertices of F in this case. Let pj(v) be the payoff to player i (i = 1, 2) if the game ends at the terminal vertex v, and let s be such that
Pi(Vs)

max
1<t<r

m (v.

Then. an optimal pure strategy for player 1 is to choose the move with final point v. The argument when player 2 is to make a move at 0 v is similar. For the inductive step, we observe that any game of perfect information can be decomposed into the union of certain subgames of smaller rank. in the following sense. (This decomposition is possible for games of perfect information with any number of players, although we shall only need it for 2-person games.) Roughly speaking, if v 1 Cr are the final points of the moves rn 1 mr with initial point the initial vertex v of F, we define games 0 (for s = 1 ) with initial vertex u. consisting of all the branches of F eman ating from u . 5

0 V
The precise definition is as follows. Let 1 be the set of vertices of F, 141 the set of vertices at which a chance move is made. and 14 (i = 1, 2) the set of vertices at which player i must make a personal move. For s = 1 r, let p(s) be the game tree defined as follows: (i) the set V of vertices of F(s) is the set of final points of all paths in F with initial point v ; 5 (ii) the sets of vertices of at which chance moves (resp. personal moves for player i) are to be made are given by

14

fl

for i

0,1,2.

17

(iii) the set Al of edges of F(s) is Al n ((5) x 1 note that, except for the edges with initial point u every edge of Al is in one of the M, since no edge of , 0 Al can have initial point in some v) and final point in some ( with s (iv) the information sets of F> are the sets containing a single element of (v) the initial vertex of F(s) is v : 5 (vi) the pyffp5) to the ith player (i = 1,2) in F is the restriction to the set of terminal vertices of of the payoff p to player i in the original game F (these terminal vertices are simply the terminal vertices of F that are vertices of F(s)). It is easy to verify that these definitions make n(s) into a game tree in the sense of Chapter 1. But it is important to note that the fact that the information sets of F ard single element sets plays a crucial role here: if this condition were not satisfied, a given information set of F would not, in general. be contained in the set of vertices of one of the and there would then be no sensible way to define the information sets of Since every information set of F (s = 1 r) is an information set of F, the pure strategies of F for player i (i = 1, 2) are simply the restriction of the pure > 1 strategies of F to the information sets of We denote the restriction to F of 1 a pure strategy u for a player in F by a(s), With this construction understood, we can proceed with the inductive step in the proof of the minimax theorem. The induction hypothesis tells us that each game (s = 1,... , r) has a pure strategy solution (u,r). Thus. (5) ,r) 5 P <P()(u;,;) <P ) 5 (u,r 1
p(s),

for all pure strategies a and T for players 1 and 2, respectively, in 5 We must now consider three cases:

Case I: the moves with initial point v are chance moves. 0 Since Vo does not lie in any information set in this case, there are unique pure strategies u and r for players 1 and 2, respectively, in F such that *(s) = o; and = r. We shall prove that (o, r*) is a pure strategy solution of F. Let r be the probability that the move m is chosen, and let u and r be any pure 5 5 strategies for players 1 and 2, respectively, in I. If P is the pure payoff function of F and p(s) that of it is clear that (6) Hence, P(,r) By (5),
so by (6),
=

P(u,r)

(7)

P(u. r*)

p(s)

((5

r) < p(s) (u

;)

P(a*. r).

[8

Similarly,
(8) r) 5 P(u p(s)(g;T(s)) > p(S);)

Inequalities (7) and (8) show that (*, r*) is a pure strateg y solution of F. Case II: the moves with initial point t are personal 0 moves for player 1. In this case. a pure strategy for player 1 in F is determined by its restriction to each F(s) together with the move for player lit chooses at v Let o be the unique pure . 0 strategy for player 1 in F such that (s) = o for s = 1 r and which chooses the move mt at t, where t is such that p(t)(u*r) = max
i<s<r

Let r be the unique pure strategy for player 2 in F such that r s = 1,... ,r. We shall prove that (o. r is a pure strateg ) 5 y solution of F. Note first that (9) P(u*,
T*)

for all

Now let u and T be any pure strategies for players 1 and 2. respectively, in F. and suppose that u chooses the move m at v where 1 < < , 0 u r. Then,
(10)

P(,r) = P(u. *)

p(u)((u)r(u))

so
=

p(u)(j(n)r)

But p and p(

L) it)

) < p) (o, r) since (, r) is a pure strategy solution (, ) < p() (, T) by the choice of t. Hence, (9) gives
) 5 P(u,r <P(a,r).

((u),

of p( u)

On the other hand,

p(q(;, r(t)) > p(t)( r) = P(,

). 5 r

Hence, (o. r*) is a pure strategy solution of F. Case III: the moves with initial point u are personal moves 0 for player 2. The argument in this case is similar to that in Case II, arid we leave it as Problem 2.15. Remark It is possible for games to have pure strategy solutio ns even if they are not of perfect information (see the game in Problem 2.4). Problems
2.1. Write down the payoff matrix for player 1 in the game of STONE-SCISSORS-PAPER (assume that the winner gets 1 from the loser).

19 2.2 Find the payoff matrix for the game HI-LU in Example 1.2. 2.3. List the pure strategies of each player and find the payoff matrices for player I in each of the five games described in Problems 1.4 and 1.5. 2.1. List the pure strategies for each player and find the payoff matrix for player 1 in the game described in Problem 1.7. 2.5. In a 2-person game the first move is a chance move in which an integer x is selected from 1.2.3, 4}, all choices being equally likely, Player 1 is informed whether i (I, 2} or (3. 1}; if x E (1, 2}, player chooses an integer p from {3. 4}, while if .r 5 {3, 4} player I chooses an integer p from {f, 2}. Player 2 is informed only whether p is odd or even and player 2 then chooses an integer z from {1, 2}. The game then terminates and player 1 receives (x + p + z) from player 2 if x + p + z is even, while player 2 receives (i p z) from player 1 otherwise. For this game, draw a game tree, indicate the information sets for each player, list the pure strategies, and find the payoff matrix for player 1. 2.6. Repeat Problem 2.5 for the following 2-person game. First, an integer x is selected at random from (1,2. 3,4,5}. Then the two players. not knowing x, simultaneously select integers from {1. 2,3,4. 5}. Each players objective is to select, an upper bound for a. If only one player is successful, he wins. If both are successful, and their choices are not the same, the winner is the player who has chosen the smaller upper hound. In all other cases, the game is a draw. The winner gets 1 from the loser. 2.1. Repeat Problem 2.5 for the game of MORRA, in which each of two players simultaneously displays either one or two fingers and at the same time guesses how many fingers the opposing player will show. If both players guess correctly or if both guess incorrectly, the game is a draw. If only one player guesses correctly, he wins from his opponent an amount (in ) equal to the total number of fingers shown by both players. 2.8. Check that the payoff matrix given in Example 2.6 is correct.
2.9. Show that, if (C ,j) and (i,j) are pure strategy solutions of a matrix game with payoff matrix (aj), then (C ,j) and (,j*) are also pure strategy solutions.

2.10. Show that, if C is an optimal pure strategy for player 1. and for player 2, then (C, j) is a pure strategy solution. 2.11. A matrix game has payoff matrix
(aj)

is an optimal pure strategy

given by

(0

0 1 such that mm max

Show that there exists a pair of indices (C

.j)

max mm 3 but that the game has no pure strategy solution.

2.12. Find the pure strategy solutions of the games considered in Problems 2.1 to 2.7, if they exist. 2.13. A matrix game has mm x mm payoff matrix (aq), where a, = i j. Show that 3 the game has a pure strategy solution, and find the optimal pure strategies for each player.

2.14. Let two functions f.g

R x f(x,p)

f be defined by 2 c g() a p 2 2 .

Show that f has a saddle point hut p does not. (Thus. the notion of saddle point used in this chapter does not coincide with the notion of saddle point used in calculus, since both f and p have a saddle point in the latter sense.) 2.15. Give the proof of the third case of the Minimax Theorem 2.14.

20

Chapter 3: Mixed Strategy Solutions

In this chapter we consider a new kind of solution of a game that is more general than the pure strategy solutions we have considered so far. 3.1 Mixed Strategies We have seen that some matrix games have no pure strateg y solution. For example, the payoff matrix of the game of MATCHING PEN NIES
(1

1 1

has no saddle point. So what should the two players do in this case to maximize their payoffs? Nothing can really be done if the game is played only once, but if the game is played repeatedly each player might try to hedge his bets by sometimes calling heads and sometimes tails. If player 1 calls heads in a proportion x of plays of the game and tails in a proportion 1 x, and if player 2 calls heads in a proportiony of plays and tails in a proportion lg of plays (so that 0< x,y < 1), the average payoff to player 1 is

xyx(1y)(lx)y+(lx)(ly)= (2x1)(2y 1).

The function f : [0,1] x [0, 1] -4 given by f(x, y) = (2x point at (1/2, 1/2) (in the sense of Definition 29), since f(x, 1/2) for all xy. Hence, max mm f(x,y) mm
=

l)(2y

1) has a saddle

f(1/2. 1/2)

f(1/2. y)

E[0,1] y[0,1

yeo,1] xE0,1

maxf(x,y)0,

So the optimal strategy for each player is to call heads half the time, in which case the expected payoff for each player is zero. Such combinations of pure strategies are called mixed strategies: Definition 3.1 Let a player of a game have rn pure strateg ies, where in > 1. Then, a mixed strategy for the player is an m-tupie x = (x ....x 1 ,) of real numbers such that
In

O<x<l fori=1,,.,,rn and The subset of R consisting of the points (xi,,... is denoted by i,,-,.
rn)

satisfying these conditions

Thus, x is the proportion of games in which the player will adopt pure strategy i. Pure strategies then become special cases of mixed strateg ies: pure strategy i is

21

identified with the mixed strategy 6 = (0,... 0, 1,0, 0), where the 1 is in the ith place. If (a,) is the m x n payoff matrix of a matrix game, and if x = (Il,.. x) and Y (vi,.. y,) arc mixed strategies for players 1 and 2, respectively, the average payoff to player 1 is clearly
,
.

P(x.y)

cLjxiyj.
i=1 .j=1

since the ptoportion of plays of the game in which players 1 and 2 simultaneously adopt tle pure strategies i and j. respectively, is Xjj and the payoff in this case is The function P : Am x A, R is called the mixed payoff function of the game. As in the case of pure strategies. player 1 tries to choose x so as to maximize P(x,y), while player 2 tries to choose y so as to minimize P(x,y). Thus, we have try to find mixed strategy solutions of the game. in the following sense: Definition 3.2 Let P : Am x A be the mixed payoff function of a matrix game. A pair (x, y) E Am x A, is called a mixed strategy solution of the game if

(1)

P(x.y) <P(x*,y*) <P(x,y) for all XE -sm, yE

Am.

In that case, P(x, y) is called the value of the game. A mixed strategy X* E Am is said to be an optimal mixed strategy for player 1 if there is a mixed strategy y E A, for player 2 such that (x*. y) is a mixed strategy solution of the game: optimal mixed strategies for player 2 are defined similarly. Thus, the mixed strategy solutions are just the saddle points of the mixed payoff function P. It follows from Proposition 2.10 (ii) that, if (x*, y) is a mixed strategy solution of the game, then max mm P(x.y)
=

mm

yE1 xE.,,,

max P(x.y)

P(x*.y*).

This implies that the value of the game is well defined. 3.2 Finding Mixed Strategy Solutions The following theorem, perhaps the most important in game theory, was proved by John von Neumann in 1928. Von Neumanns Minimax Theorem 3.3 Every matrix game has at least one mixed strategy solution. Note that this theorem does not assume the game to be of perfect inform ation. We have already seen that a game that is not of perfect information need not have have any pure strategy solutions. We shall prove von Neumanns theorem in Section 3.6. However, althoug h this theorem is very important, its proof is non-constructive, i.e. it does not tell us how to actually find a mixed strategy solution of a given matrix game. Accordingly, we devote the remainder of this section. and the next two sections, to some simple methods which can often be used to find mixed strategy solutio ns.

22

The next two propositions can he used to test whether a candidate for an optimal mixed strategy is actually optimal. To prove them we shall need: Lemma 3.4 Let P : game. Then,
m

Z *

H be the expected payoff function of a matrix max P(Sj,y) for ally E mm P(x,6) for all x

max P(x,y)
Xrn

1<,<rn =

yE

Ifllfl P(x,y)

lJr

rn

Proof For y E

.,

let

Q()
Then, P(5,y) (y,. .y,,) E
<

max ,... 1 (x
,Xrn)

Q(y) for i

1,... ,rn, so for all x

rn,

P(x. y)

OjjXjj =

aiii) x

y)xj <Q(y)

Q(y).

Hence,

P(x,y) < max P(&,y). for all x E 1< <in This shows that
max P(x.y)
XE

exists for all y

and that

XEm

max P(x, y) < max P(&, y) for all E Y 1<m

On the other hand, since every pure strategy is a mixed strategy.

max P(x,y) > max P(6,y) for ally

Together with (2), this proves the first equation in the lemma. The second equation is proved similarly.

Proposition 3.5 Let P: x H be the mixed payoff function of a matrix game with value v. If a mixed strategy x E is optimal for player 1, then
.

(3)

mm
Jn 1

P(x.

u for all

1....

n.

Conversely, if a mixed strategy x for player 1 satisfies (4) then x is optimal. Analogous results hold for player 2. P(x*,o)
>

u for all

1,

23

Remark Of course, this shows that if a mixed strategy x for player 1 satisfies (4), then it satisfies the apparently stronger condition (3). Proof Suppose first that x* is an optimal mixed strategy for player 1. Then, there exists a mixed strategy 6 z for player 2 such that (x, y*) is a mixed strategy solution of the game. By Definition 32. P(x,y) Hence, rrun P(x,y)
y =

P(x,y)

u for ally 6 .

v, so by Lemma 4.4, mm
1jn

P(x*,j)=v.

Conversely, suppose that x 6 satisfies (1). Let (x.y) E x Zi,. be a mixed strategy solution of the game (this exists by Theorem 3.3). We shall prove that (x*. y) is also a mixed strategy solution of the game. Of course, this will prove that x* is an optimal mixed strategy for player 1. By Lemma 3.4, mm p(x*,y) = mm
ye 1jn

and this is > v by (4). Hence. (5) Now, P(x,y) (6)

P(x*,y) >

for all yE .

v and since (x,y) is a mixed strategy solution,

P(x.y) <P(x.y) <P(x.y) for all x 6

yE

It follows from (5) and (6) that P(x, y) < P(x, y) and hence that P(x*,y)
= =

u < P(x*,

I)

P(x.y). Thus.
=

P(x,y) < P(x,y)

v < P(x*,y) for all x E

Lm,

yE

This shows that (x*, y) is a mixed strategy solution of the game. Proposition 3.6 Let P : m x . + be the mixed payoff function of a matrix game. Let x 6 -sm, Y 6 V ER be such that P(5.y) <u < P(x5) for all i 1.... ,m,

1.. ,n.

Then, (x*, y) is a mixed strategy solution of the game and v is its value. Proof Let y = (y y) E ... and let x* = (I x). Then, if (a) is the payoff matrix of the game, P(x*,y) aijxyj

(aux:) y

P(x)y >

Vj

V.

21

Similarly, P(x,y*) < v for all (7)

XE

Hence,
,,

P(x,y) <v <P(x,y) for all x E

yE

Taking x = x and y = y shows that v (x, y) is a mixed strategy solution. E Example 3.7 A game has payoff matrix /3 (i 3 \2 Let 1 1 x*=(.O,O,), Then, ) 1 P(x,6 Similar calculations show that (8) P(*,S)

P(x, y*), and then (7) shows that

3 3 1 2

2 2 2 3 J 1 1

3 x

+2 x

P(,y*)

for all i

1.2.3,4.

1,2,3.

By Proposition 3.6, (x, y*) is a mixed strategy solution of the game and 1/2 is its value. (Note that it is just an accident that equality holds in (8) to conclude that we have a mixed strategy solution, it is only necessary to know that each term is > the next.)

3.3 Inferior Strategies In this section we describe a method which can sometimes be used to simpli fy the analysis of a matrix game, by reducing the size of the payoff matrix. Definition 3.8 Let (a) be the payoff matrix of a matrix game. A pure strategy 1 i for player 1 is said to be inferior to a pure strategy io for player I if (9) 11 a < a 0 for all

j;

1 i is comparable to io if equality holds for all j; and i is strictly inferior to 1 0 i if the inequality in (9) is < for all j. Similarly, a pure strategy j for player 2 1 is said to be inferior to a pure strategy Jo for player 2 if

ajq
and so on.

>

0 aq for all i,

Thus, the pure strategy i for player 1 is comparable (resp. inferior. resp. strictly 1 inferior) to the pure strategy i if each entry in the i th row of the payoff matrix 0 1 is equal to (resp. less than or equal to, resp. strictly less than) the corresponding entry in the ioth row, while the pure strategy Ji for player 2 is comparable (resp. inferio r,

25

resp. strictly inferior) to the pure strategy jo if each entry in the th column of the Ji payoff matrix is equal to (resp. greater than or equal to. resp. strictly greater than) the corresponding entry in the joth column. In this case, we shall sometimes say that the i row of the payoff matrix is comparable (resp. inferior, resp. strictly th 1 inferior) to the i row, etc. th 0 It is intuitively clear that, if the pure strategy i for player 1 is inferior to i 1 , 0 player 1 should adopt i in preference to i since his payoff when using i is at 0 1 0 least as great as his payoff when using i whatever pure strategy player 2 adopts. , 1 We expect, therefore, that if we delete the i row of the payoff matrix, any pure th 1 strategy solution of the new matrix game should be a pure strategy solution of the original matrix game. This is indeed the case, as the next proposition shows. Proposition 3.9 Let A be the payoff matrix of a matrix game F, and assume that some row (resp. column) of A is inferior to another row (resp. column). Let B be the submatrix of .4 obtained by deleting the inferior row (resp. column). Then, any pure strategy solution of the game FB with payoff matrix B is also a pure strategy solution of F. Conversely, if the deleted row (resp. column) is strictly inferior to another row (resp. column) of A. then every pure strategy solution of F arises from a pure strategy solution of FB. It follows that similar results hold when B is obtained from A by deleting a succession of inferior rows and columns. Warning If the deleted rows or columns are only inferior, and not strictly inferior, to other rows or columns, it is possible that not all pure strategy solutions of F arise from pure strategy solutions of FB. For example, for the payoff matrix /0 A=( 0 1 1 2 2 1 0

(1, 1) and (2, 1) are pure strategy solutions. However, pure strategy 2 for player 1 is inferior to pure strategy 1, so if we had reduced the payoff matrix by deleting the second row we would have lost the pure strategy solution (2, 1). The proof of Proposition 3.9 is the subject of Problem 3.16. We shall now prove a more general result. To state it we need one more piece of terminology. Let A be the m x n payoff matrix of a matrix game F, let 1 < r < m, 1 < s < ii, <...<j<n,andletBbethesubmatrix <j <...<z<m,1<j 2 <m 1 1<i of A consisting of the elements in rows i ,i,. and columns ,.. 1 1, ,j, i.e. B is the r x s matrix whose (k, 1)-entry is akJ. If E is a mixed (i = r) strategy for player 1 in the game FB with payoff matrix B, one can extend to a mixed strategy x E m for player 1 in F in the obvious way, i.e. the ith component of x is if i = it for some k = 1,... r and it is 0 otherwise. Proposition 3.10 Let .4 he the payoff matrix of a matrix game F, and let B be the submatrix of A obtained by deleting a row of A corresponding to a pure strategy for player 1 that is inferior to another pure strategy for player 1 (resp. a column of 4 corresponding to a pure strategy for player 2 that is inferior to another pure strategy for player 2). Let be a mixed strategy solution of the game FB with payoff matrix B, and let (x*, y) be its extension to a pair of mixed strategies for F. Then, (x, y) is a mixed strategy solution of F. If the row (resp. column)
...
,.

26

deleted corresponds to a pure strategy for player 1 (resp. 2) that is strictly inferior to another pure strategy for player 1 (resp. 2). then every mixed strateg y solution of F arises in this way from a mixed strategy solution of
FB.

It follows that similar results hold when B is obtained from A by deleting a succession of inferior rows and columns. Proof Let A = (a,) be m x n, and suppose that pure strategy 1 i for player 1 in the matrix game F is inferior to pure strategy i Let . 0 = (ci, emi), ?7 = (rh,.. ,r). so that y = i and x (It,... ,Xm) where x = j if 1 < i < i , 1 0 and z = if i <i < rn. Let P be the mixed payoff function of F, and 1 PB that of t Clearly, B.
,

(10) and (11) On the other hand, (12)

PB(,J)=P(X,)

forj=1

PB(5,)=P(,.y) foriii.

1 P( , y)
=

, 0 P( y*),

since r > 0 and a < a for all j = 1,... ri. Let v be the value 0 1 of PB; since (. i) is a mixed strategy solution of FB.
,

PB(Si,71*) <u <PB(*,oJ) for all i

11 and all

j.
. 1 i Finally,

By 10), P(x,) by (12), Hence.

v for all

j, and by (11) P(&, y)

< v for all i

PQ5j,,y*) < P( ,y) 0

P(&,y) <u <P(x,S) for all i

1,... rn,

1,... a,

so by Proposition 3.6, (x*,y) is a mixed strategy solution of F. and c is its value. Conversely, suppose that pure strategy i is strictly inferior to pure 1 strategy i , 0 let (x*,y*) be a mixed strategy solution of F, and let x (x ,...,Xm). y = (Yi.... ,y). Since y > 0 for all i and y >0 for at least one value of j. it follows that j 0 yj(ai
j1

auj) > 0

and hence that

Zyjai1J <Zyjai j 0

Since P(j,y*) < v for all i, we have < v, so by Problem 3.11 x = 0. Define , = y and = ( m.-i). where , 1 = x, if 1 < i < i j = x, if
,

27

1 i < i < m 1. The argument in the first part of the proof shows that mixed strategy solution of the game F . 8
Example 3.11 Suppose that the payoff matrix is

(*,j*)

is a

f 1 (1

1 1 2

2 3 4

Pure strategy 3 for player 2 is strictly inferior to pure strategy 1, so we delete the third column: /1 -1 1 1 2 In the game with this payoff matrix, pure strategy 3 for player 1 is strictly inferior to pure strategy 1, so we delete the third row: (1 1 1

We have analyzed this payoff matrix before. and we know that the value of the game is 0 and the unique optimal mixed strategY for each player is (4, 4). It follows that. for the original game, the value is 0 and the unique optimal mixed strategy for each player is ( 4,0). The following result shows that games of perfect information can be solved triv ially by successive deletion of inferior strategies.
Theorem 3.12 The payoff matrix of a game of perfect information can be reduced to a 1 x 1 submatrix (i.e. to a single entry) by successive deletion of inferior rows and columns.

Proof The proof is similar to that of the Minimax Theorem 2.14, so we shall be brief. We proceed by induction on the rank of the game F, the case in which the rank is zero being trivial. For the inductive step, we use the games s = 1,... r. defined in the proof of Theorem 2.14, and consider three cases.
,

Case I: the moves with initial point v are chance moves 0 Let = (a), with 1 <i < 1 <j be the payoff matrix of the game p(s) The payoff matrix .4 of F has rows indexed by r-tuples I = (j(), < m), columns indexed by r-tuples J = (J(l) 1 < j)), 1 < j(S) < a(s) and its IJ-entry is
ajj
=

where it is the probability that the sth move is chosen at the initial vertex n It 8 . 0 is easy to check that: (i) if i is an inferior row of for some 1 < ? < r, 1 < i < of .4 for which j(S) = i are inferior rows in .4: then all rows

28

(ii) if row i is deleted from A(s), the effect on A is to delete from .4 all rows for which (s) = Similar facts hold for the columns. Hence, each permissible deletion (i.e. the deletion of an inferio r row or column) in any of the matrices A, 1 < s < r. is associated with a corresponding set of permissible deletions in A. By the induction hypothesis, each matrix A( can be reduced to a 1 x 1 matrix by a sequence of permissible deletions. and this has the effect of reducing .4 to a 1 x 1 matrix. This completes the induct ive step in this case. Case II: th moves with initial point u are personal moves for player 0 1 In this case, the rows of the matrix A of F are indexed by the set of pairs I = (.s, i), where 1 < s < r and 1 < i < m(. its columns are indexed by the set of r-tuples J = (j,... ,j>) as in Case I, and the IJ-entry of .4 is given by ajj
=

a(S).

This time, one checks that: (i) if i is an inferior row of 4(5), for some 1 < s < r, 1 < i < then I = (s, i) is an inferior row of A, and deleting row i from A( corresponds to deleting row I from A: (ii) If j is an inferior column of A>, then all columns J with j(5> j are inferior columns of A, and deleting column j from A() corresponds to deletin g from .4 all jt columns with j(S) Hence, to each permissible row deletion in some ,4( corresponds permis a sible row deletion in A, and to each permissible column deletion in some A corresponds a set of permissible column deletions in .4. By the induction hypoth esis, there is a sequence of permissible deletions which reduces each A> to a 1 x 1 matrix. The corresponding sequence of permissible deletions reduces A to a matrix which has only one column. Clearly, this latter matrix can be reduced to a x 1 1 matrix by further deletions of inferior rows until only the maximum entry in the column remains. This completes the inductive step in Case II. Case III: the moves with initial point v are personal moves for player 0 2 The proof in this case is similar to that in Case II. Theorem 3.12 clearly implies the minimax theorem, since any 1 x 1 payoff matrix obviously has a pure strategy solution (!) which, by Proposition 3.9. will be a pure strategy solution of the original game.

3.4 The 2 x n Case We specialize now to the case m 2, i.e. player 1 has only two pure strateg ies. Let the payoff matrix be (am a 2 k\bl b 2 ) 11 b and let the mixed payoff function be P : x * R. A mixed strategy for player 1 is of the form x = (cm. 1 cm)
..

29

for some 0

<

<

1. Then, for ) 2 P(x,

1, (1

2 aa

2 (a

)a 2 b + b.

From Problem 3.7, the value v of the game is given by

max mm
2 xE 1jn

P(x,)

max mm o1 1jri

and by Proposition 3.5, x* is an optimal mixed strategy for player 1 if and only if (13) ruin P(x*,)
l<j<a
=

This means that the value of the game and the optimal mixed strategies for player 1 can be found by the following graphical method. We plot the straigh t line graph of 8 = (a 2 )u 2 b + b in the cr3-plane for j = 1,... n; the graph of 2 mm [(a b + b is the polygonal curve which )cr ] 2 2 forms the boundary of the

1Jn

region which lies below each of the lines. The value L is then given by the maximum 3-coordinate of the points on this polygonal curve corresponding to 0 <cm < 1. By (13), the value (or values) of cm which give this maximum value of 3 give the optimal mixed strategy (or strategies) for player 1. The following examples illustrate this method, arid also show how the optimal mixed strategies for player 2 may sometimes be determined. Example 3.13 We consider the matrix game with payoff matrix (1 3 1 5 3 3

30

We must plot the straight lines + 3(1

2n + 3, 6a + 5, 6n 3.

+ 5(1

n)

= =

3a

3(1

The polygonal boundary is clearly formed by segments of the last two lines (see the diagram above). These lines intersect when 6n + 5 i.e. c Hence, the value of the game is v=6
-

6n
=

3.

+5=1,

and the (unique) optimal mixed strategy for player 1 is x*

()

21

) 1 Since P(x*,6 = 1 + 3 > u, Problem 3.11 (applied to player 2) = shows that any optimal mixed strategy for player 2 must be of the form

()

()
y

(0. n, 1

a).

31

for some 0 (14)

a < 1. By the analogue of (13) for player 2. 1

mill max O<<1 1<1<2

Now, P( y) = a + 3(1 a) 1 Ia + 3. P( y) . 2 5a 3(1 a) = 8a 3. The graph of the righthand side of (14) against a is the polygo nal curve which represents the region above the graphs of the two lines
,

4a + 3.

8a

3
,

(see the.diagram above). These lines intersect when a and this is clearly the value of a which satisfies (14). Hence, the (unique) optimal mixed strategy for player 2 is y
=

(0,

).

Hence, this game has a unique mixed strategy solution. Example 3.14 We consider the matrix game with payoff matrix /2 4 (13 4 11 1 2

Of course, this is not a 2 x n matrix. However, we observe that pure strategy 2 for player 1 is strictly inferior to each of the other pure strategies for player 1. so Proposition 3.8 shows that the optimal mixed strategies for the origina l game can be obtained from those of the game with payoff matrix (2 4 4 11 2

to which the methods of this section can be applied.

219

3/5

32

The appropriate diagram for player 1 is shown above, from which it is clear that any a with < a gives an optimal mixed strategy = (a, 1 a) for player I in the reduced game. Hence, the optimal mixed strategies for player 1 in the original game are 9 3 x=(a,0,1a),

The value of the game is v game, we observe that

4. If P is the mixed payoff function for the original

P(x,i)=75a>4

ifa<,

)=9a+2>4 3 P(x*, ifa>. It follows from Problem 3.11 that the unique optimal mixed strategy for player 2 is (0,1,0), i.e. the pure strategy 2. Remark The methods of this chapter apply, with obvious modifications to , the case of in x 2 payoff matrices. This time, one analyzes player 2s optimal mixed strateg ies first. If the payoff matrix is 1 a 1 b 2 b a2 am the value of the game is
=

bm

mm

O<a<1 1<i<m

max [(at

b)a + b].

The optimal mixed strategies for player 2 are found by examining the polygonal boundary of the region above each of the straight lines 3 = (ab)ab, 1 <i < m. 3.5 Convexity To prove Von Neumanns Minimax Theorem 3.3 we introduce a property of certain subsets of Rtm (m > 1). But first we review some concepts from real analys is. Recall that a subset X of Rtm is said to be bounded if there is a constant D such that, if X (xi,... .Xm) is any point of X. we have xj < D for i = 1 in. (You may have been given a different definition in another course; if so, it is an exercise for you to prove that the definitions are equivalent.) We also recall that a sequence of points x E H. k = 1.2,3,.. where (x c), is said to converge to a point x (xi, m) e H
.,

urn x

x for all i

1,..

rn.

A subset X of Rtm is said to be closed if, whenever (x(k)) is a sequen of ce points of X converging to a point x E H it follows that x tm e X. It is obvious that the intersection of any family of closed sets is closed. (Again, these definit ions are

33

equivalent to aiiy other definitions of convergence and closed sets you might have been given.) The following definition introduces the concept that will be needed to prove Theorem 3.3.
Definition 3.15 The line segment joining two distinct points x and y of R m (m > 1) is the set of points of the form Ax + (1 A)y for some < < 0 A 1:

-:7--

x The points x and y are the endpoints of the line segment and (x + y) is its midpoint. A subset C of m (m > 1) is said to be convex if, whenever x and y are points of C, the whole of the line segment joining x and y is contained in C. For example, the interior of a triangle is convex

but the interior of the following polygon in the plane is not:

The importance of the idea of convexity for us derives from the follow ing result: Proposition 3.16 If a player in a matrix game has rn pure strateg ies, the set of his optimal mixed strategies is a bounded closed convex subset of R. Proof Let X be the set of optimal mixed strategies for player 1 in a matrix game

31

with mixed payoff function F: m x * R, and let v be the value of the game. It is clear that Lm is bounded: if x 1 (x Xm) E Lm then Xj < 1 for = 1,... rn. Since X C X is also bounded. Next, we show that X is closed. Let (x(k)) he a sequence of points of X coti (k) k) (k verging to a point x irt \e have to show that x E A. If x (x 1 ,....im and x = (xi x,). then x(k> E m implies that. for all k.
,
.

0< Letting k
+

< 1 for i

1,... rn, and Zx

1.

we get
rn

0<x<1fori1....,m.andx=1, which tells us that x E i.e. that x is a mixed strategy for player 1. Proposition 3.5, x E V if and only if (15)
Since E X for all k.

By

P(x,) > v for all

1,... ,n.

(16)

P(x,) > v for all

1.

..

,0.

Now, if (a) is the payoff matrix of the game,

F(x,6) Hence. condition (16) is that

Zxjajj.

>

for all j, k. Letting k in (17) gives (15), so x e X. Hence, X is closed. Finally, we prove that X is convex. Let x = (x x,,) and y = (yj,... ,y) ,... 1 be elements of X, and let 0 <) < 1. Then 0< xj,yj < 1 for all i and x = = 1, so that

\.0 (1

\).0

Ax + (1 A)y <).1 (1

).1

for all i, while

(Ax + (1

)yj)

x + (1

A)

A.1 + (1

A).1

1.

It follows that Ax + (1 A)y E optimal is equivalent to P(x,6)

By Proposition 35, the fact that x and y are

v and P(y,6)

> i;

1,

rn.

35

This implies that P(/\x + (1

(Ax + (1
111

)yj)ajj
Tn

(1 )) yiaj

) 5 )P(x, + (1 >Xv+(1A)v=v.

Hence, .Xx + (1

,\)y E X. This proves that X is convex.

Remark In particular, the set Am is bounded, closed and convex for all m 1. The proof of this is actually contained in the proof just given. Alternatively, consider a game in which player I makes one of rn possible moves at the initial state and then the games ends, the payoff to player 1 being the same whichever move he made. In such a game, the set of optimal mixed strategies for player 1 is the whole of Am, so our assertion follows from the proposition. We shall now consider the smallest convex set containing a given subset of R. The following (nearly obvious) result shows that this makes sense: Proposition 3.17 The intersection of any family of convex subsets of R is a convex subset of R. Proof Indeed, if x and y are two points in the intersection, they are in each of the sets; then so is the line segment joining x and y since the sets are convex; and this means that this line segment is contained in the intersection. It follows from this proposition that, if X is any subset of R the intersection , m of all the convex subsets of R containing X is itself convex. It is called the convex hull of X, is denoted by (X), and is obviously the smallest convex subset of R m containing X (i.e. it is contained in every convex subset of iR that contains X). It can be described explicitly as follows: Proposition 3.18 The convex hull of a subset X of R is the set of points of the m form (18) wherexEX,a>Ofors=1,.. ,randa=1.

A linear combination of points x ...,x as in the statement of the proposition , 1 is called a convex combination of the points. Proof Let X be the set of points of R of the form (18). It is obvious that X C X m (take r = 1). We prove first that X is convex; this implies that (X) C X. Suppose then that x 1 x and 3i,... Yq are points of X, and that pr,... and va,... 1 are non-negative real numbers such that q 3 IL = Vt = 1. The points
, , ,

, 8 XZJLSX

YVtYt

36

are in X, and we have to show that, if 0 < A < 1, the point (1 A)x + y is also inX. Definez,. xr forr= 1,...,p. and Zrp=Yr forr= 1, ...,q; define Qrjirforr=1,,par1dcr=Oforrp+1,..,p+q:aflddefiTle3r0for r1....,pand3rp=rforr1....,q.Then.zrEXforr1 ,p+qand

x=Urzr: Now,

YZ3rZr

p+q

(1_A)X+AyZyrZr

where 7r (1 A)ar + Ar for r = 1,... ,p + q. Since ar and r are > 0 for all r, we have 7r > 0 for all r, and
p+q

p+q

p+q

7r

(1

A)n + AZr = 1

A+A

1.

This shows that (1 A)x + y E X. We must now show that, for any r > 1, any convex combination of r points of X belongs to (X). We proceed by induction on r, the case r = 1 being obvious. Assume next that r > 1 and that the result is true for convex combinations of fewer than r points of X. Let x be a convex combination of r points of X, as in the statement of the proposition. If a,. = 1, then x = x,. and there is nothing to prove. Assume from now on that 0 < a,. < 1. Note that a/(1 a,.) > 0 for a/(1 a,.) 1. By the induction hypothesis, s= 1,... ,r 1 and that

(X).

Since (X) is a convex set,

(1

ar)yarxr

E (X).

To prove Von Neumanns minimax theorem we shall need one more result about convex sets. For this, we shall need the concept of a hyperplane in am.: this is a subset of R of the form H
=

{x E

I b.x

= d},

where b is a non-zero vector in R and d E R. If rn = 2, a hyperplane is a straight line, and if in = 3 a hyperplane is a plane in the usual sense. Two subsets X and Y of are said to be on opposite sides of the hyperplane H if either b.x > d for allxfiXandb.y<dforallyfiY,orb.x<dforalixfiXandb.y>dforall y E Y. The result we want is the Separating Hyperplane Theorem 3.19 If X and Y are disjoint convex subsets m of F. there is a hyperplane H in 11 such that X and Y are on opposite sides of H.

37

Of course, this result is false if the sets X and Y are not assumed to be convex:

3.6 Proof of Von Neumanns Minimax Theorem To prove Von Neumanns minimax theorem we shall need, in addition to the results on convex sets stated in the preceding section, the following simple lemma from real analysis. Lemma 3.20 Let X be a subset of R, and let fi,... fm be continuous real-valued functions on X (m, n 1). Define functions F and f on X by
,

F(x) Then, F and

max
1<1< rn

f(x)

mm
1 <i < rn

f(x),

for x

X.

are continuous.

Proof We prove that F is continuous. The proof for f is similar (alternatively, the result for f can be deduced from that for F by replacing each f by ft).
0 Fix x E X and let e > 0. Since that x x 11< 0

, 0 is continuous at x there exists O > 0 such (x) 1 jf

f,(xo)I

<E.

Let 5 (19)

mm
1< i < rn

; then, > 0 and <eforalli=1,

xxo<rrlfj(x)fj(xo)I
-

rn.

i =

0 then f(x) < f(xo) + < F(xo) + for all Now (19) implies that, if x x < 7fl and hence that F(x) < F(xo) + e. On the other hand, let io be such 1 that (xo). 1 F(xo) = f = max f (xo) 0
,

1<

i< in

Then (19) also implies that, if and hence that F(x) > F(xo)

0 x x 1< e. Hence.

then

fj (x)

>

0 f (xo)

F(xo)

XO

F(x)

F(xo) <.

38

This shows that F is continuous at x Since x was an arbitrary point of X, this . 0 0 completes the proof. We are finally ready for the proof of the minimax theorem. Consider a matrix gamer with mx a payoff matrix (a) and mixed payoff function P : ii m X For each i = 1, ..., in, the function f IR given by f(y) P(.y)
gjajj.

for y

(yr,.., y) E R given by

is continuous. By Lemma 3.20. the function F F(y)

max f()

is continuous. Since L is a closed hounded subset of , the function F is hounded and attains its bounds. In particular, mm exists. By Lemma 3.4, so does mm
y,. yE-X, 1<i<rn

max p(,

max P(x,y).

Similarly, max mm
XEm yE

P(x.y)

exists. We have to show that these two quantities are equal. By Lemma 3.4, mm max P(x,y)
=

mm

yc I<<m

max P(&,y) =v (say).

Let y E Z be such that max

For j = 1 a, let ,cJ = (a a2J... . . a) e m (thus, 1 , 1 is the jth column of the payoff matrix. but written as a row). Let T be the convex hull of the n T points By Proposition 3.18, T is the set of points of ff of the form yJK(3>, where Yj > 0 for j = 1, a and Z= = 1. We can therefore define a map * T by
=

if y

= (y.... ,n)

Clearly, (z = T. Note that, if y E ) 11 yjajj = P(, y), so (y) is (y)

Ln

and i

m, the ith component of

.y),... 1 (P(

39

By the definition of v, max P(&.y) Hence, if t

v for ally E

tm)

E T, then max
i<i<m

>

andso (20) if t ,.... 1 (t

E T. then t > v for some i max PQ5, y)

1,

. .

m.

On the other hand,

1< i < rn

so if th(y) (21) Now let

(t.

t), then t <u foralli=1....,m.

Z{z=(zi,...,znz)eRmzi<vforalli1,...,m}, and let Iz<vforal1i=1,...,m}. m )ER 1 Z={z=(zi,.,.,z It is easy to see that Z is a convex subset of and by (20) it does not intersect the convex set T. By the separating hyperplane theorem, there exists a non-zero vector b = (b , b,) E m and a real number d such that .... 1 (22) b.z<difzEZ. b.t>diftET.

On the other hand, if z E Z, it is clear that there is a sequence of points z(k) e Z, for k = 1, 2, 3,. . ., which converges to z as k + (for example. one can take z(k) = z (1/k,... , 1/k)). Since b.z(k) <v for all k, it follows on letting k * -c that b.z < d. Hence, (22) actually implies that (23) By (21), t b.z<difzE, b.t>diftET.

Tn Z, and hence by (23) and the preceding remarks,

b.t For i
=

d.

1.

m. let u

(u,
U)

. . .

u, where ifji, ifj=i.

It;
t1

By (21), u E Z for all i

1,... , m, and so b.u <d. But

b.

40

It follows that, for all i = 1,..., rn, we have d b < d and so b 0. Since b = (b bm) is a non-zero vector, some b is > 0 and so ,.. 1 b > 0. Then

/
=

1 b

b,

is well-defined, and clearly x* E Lm. Also, if = ci

thenby(23)x.t wift e Tandx*.z <wifz E Now,ifv=(v v) ER tm then v E Z so x*.v < w. But x.v = u, so v < w. If tj E then o(y) E T and so x(y) > w > u. Since x(y) = P(x, y),
.

it follows that P(x*, y) > v for all y E

Hence,

mm and so

P(x*,y) >v

< max mm

P(x.y).

yEs,,

By Proposition 2.10 (i), the reverse inequality holds, and the minirnax theorem is proved.

41

Problems
3.1. Verify that, for any A x [O,1}, 3 A 8 5 1.6 5A 16 5 5AA 111

=(-+8

- -), 16 16 2

=(- -) 4 4 2

gives a mixed strategy solution (x* ,y) of the game with payoff matrix 3 1 3 2 3.2. The payoff matrix of a matrix game is 0 2 1 1 Verify that x (0,0,
,

3 3 1 2

2 2 2 3

01 1 1 1 2 1 1 3 010

),

(,

,0,0) gives a mixed strategy solution (x*,y) of the game.

3.3. A matrix game has payoff matrix (a), where a,, 1 if i mixed strategy solution of the game and write down its value. 3.4. A matrix game has payoff matrix /0 0 1 1 0 1 1

j and

1 for all i. Find a

(o

Show that there is a pure strategy solution of this game, and also that each player has an optimal mixed strategy that is not a pure strategy. 3.5. The n x n payoff matrix of a matrix game has the property that the sum of the entries in each row, and the sum of the entries in each column, are each equal to the same number a. Find a mixed strategy solution of this game and show that its value is s/n. 3.6. Let (a,,) be the payoff matrix of a matrix game and suppose that (i,j) is a pure strategy solution of the game. Show that ( ,6-) is a mixed strategy solution of the game. Deduce that is the value of the game. 3.7. Let P
m

1. be the mixed payoff function of a matrix game. Prove that

max

mm

xE,,, 1<j<,,

) 3 P(x,6

mm

yEz,, 1<,<m

max P(5,.y.

and that both sides are equal to the value of the game. 3.8. A matrix game with square payoff matrix is said to be symmetric if the payoff to plaverl when he uses pure strategy i and player 2 uses pure strategy j is the same as the payoff to player 2 when he uses pure strategy i and player I uses pure strategy j. For a symmetric game, show that (i) the payoff matrix (a,,) of the game is skew-symmetric, i.e. that a, = a for all , 3 i,j: (ii) (x,y) is a mixed strategy solution of the game if and only if (y ,x) is a mixed strategy solution; (iii) the value of the game is zero. 3.9. A matrix game F has m x ri payoff matrix (a,,), and another matrix game F has payoff matrix (a1). where a, = a, + A and .1 is a constant. Show that 3

42 (i) if the mixed payoff functions of F and F are P and F, respectively, P(xy) P(x,y)+A for au XE

s,,,,

ys

(ii) if ii and u are the values of F and F, respectively, then v = u + A; (iii) each player has the same set of optimal mixed strategies in the two games. 3.10. Let (ajj) be the payoff matrix of a matrix game with value v. Show that, if A > 0 is a constant, the game with payoff matrix (Aa) has value Au and that a player has the same set of optimal mixed strategies in the two games. Is the same result true if the condition that A > 0 is omitted I 3.11. Let P x ,, * IR be the mixed payoff function of a matrix game with value u, and let x* x) S s,-,, and y S (x be optimal mixed strategies for players 1 and 2, respectively. Show that, if P(5, y) < u for some i = 1 m, then x 0. Write down the corresponding result when the roles of players 1 and 2 are interchanged . 3.12. A game is played with five ordinary playing cards, the 2.3, and 4 of clubs anti the 2 and 4 of diamonds. Player 2 holds 2 and 3 and player 1 holds the remaining cards. Player 1 places two of his cards face up on the table. Then player 2 plays one of his cards. For each of the two pairs consisting of a card played from each players hand, player 1 receives from player 2 the difference in values of the two cards if they are of the same suit and pays player 2 the difference of the values if they are of different suits. Draw a game tree, list the pure strategies, and write down the payoff matrix for this game. Find the pure strategy solutions, and verify that each one can be obtained by reducing the payoff matrix to a single entry by successive deletion of inferior rows and columns. 3.13. In a certain game, a coin is tossed and if the outcome is heads, player I chooses one of the integers 1, 6, while if the outcome is tails player 1 chooses one of the integers 2, 7. Player 2, knowing the choice made by player 1, then chooses one of the integers 3, 9. The choices of the two players are then added to yield the sum .s. The final move is the selection of a winner by a chance process which selects player 1 with probability 4/5 and player with 2 probability 1/5. The winner then receives s from the loser. Draw a game tree for this game, list the pure strategies and find the payoff matrix. Find a pure strategy solution of the game and show that the payoff matrix can be reduced to a single entry by successive deletion of inferior strategies. 3.14. Let ij be an optimal pure strategy for a player of a matrix game, and suppose that i is 1 inferior to a pure strategy i Show that io is an optimal pure strategy for . 0 the player. 3.15. Show that the matrix /2 3 .1 (140 \i 0 6 is not the payoff matrix of any game of perfect information, although it does have a saddle point. 3.16. Prove Proposition 3.9 starting from the definitions (not by deducing it from Proposition 3.10!). 3.17. A matrix game has payoff matrix (2 3 5 11 2

(7

Find the value of the game and the optimal mixed strategies for each player. 3.18. Repeat Problem (i) 3.17 for the following payoff matrices: (5 (1 2 (ii) 4 ,2 1 2 3

3 3

1 0

43 (iii) (19 \0 (iv) 15 20 17 15 16 5

f-i
1 \O (v) 4 2 1

1 2 l

o
3
(Vi)

3 1 3 i 0

(i 2

3 4

5 6

7 8

9 10

3.19. Repeat Problem 3.17 for the following payoff matrices, deleting inferior strategies where possible: (i)

fi
(6 (ii) 2 4 2 4 (iii) 16 14 0 \. 22 (iv) 4 8 0 2

7
2 1

2
7 6

0 2 4 2

2 6 2 0

14 4 2 12

6 10 12 5

11 8 6 10

0 365 15 10 8 9 10 15 11 7 5942 3.20. In a certain game, player I holds two Kings and one Ace. He discards one card, lays the other two face down on the table, and announces either two Kings or Ace King. Ace King is a better hand than two Kings. Player 2 must either accept or question player ls call. The hand is then shown and the payoffs are as follows: if if if if player player player player 1 1 1 2 calls the hand correctly and player 2 accepts, player 1 wins 1 from player 2; calls the hand better than it is and player 2 accepts, player 1 wins 2 from player 2: calls the hand worse than it is and player 2 accepts, player 2 wins 2 from player 1; questions the call, the above payoffs are doubled and reversed.

Draw a game tree for this game, list the pure strategies. write down the payoff matrix, and find an optimal mixed strategy solution. What is the value of the game?

4.1

Chapter 4: Solving Matrix Games in General

In this final chapter we describe an algorithm which allows one to find all the mixed strategy solutions of any matrix game.

4.1 Simple Solutions In this section we discuss certain special mixed strategy solutions of matrix games, called simple solutions. Although not every game has a simple solution, it turns out that all the mixed strategy solutions of any matrix game can be constructed starting from the special solutions of other games contained in the original game. Definition 4.1 A simple solution of a matrix game with mixed payoff function P -m X L * R is a pair of mixed strategies (x, X Zn such that, for y) E some v E III. (1) P(x*,6j) = u = P(6,y) for all i = 1,... m.,
,

j = 1,... ,n.

It follows from Proposition 3.6 that a simple solution of a matrix game is a mixed strategy solution of the game and that the value of the game is v. Proposition 4.2 Assume that the payoff matrix of a game is invertible (and hence square). Then, if a simple solution of the game exists, it is unique. Proof Let the payoff matrix B of the game be r x r, and suppose that (x*, y*) and (x*. y) are two simple solutions of the game. If we think of x*,x (resp. y*y*) as row (resp. column) matrices, the equations in (1) can be written in matrix form as 1 1 1 Hence, (x*

x*B = v (1

...

1) = x**B,

By* = v

By*.

x**)B

0.

B(y*

y) = 0.

Since B is invertible, these equations imply that x = x and y = y. We shall now establish the conditions under which simple solutions exist. We shall need to introduce some notation. Namely, if B = (b) is an r x r matrix, let p(B) (resp. K,(B)) be the sum of the entries in the ith row (resp. column) of B (i = 1, r), and let (B) be the sum of all the entries of B. Note that
.

/1 (B) 1 (
...

kr(B)) = (1

...

1)B, pr(B))

= B

15

Now recall that the adjoint adj(B) of B is the transpose of the matrix of cofactors of B, and that Badj(B) = adj(B)B = det(B)I, where clet(B) is the determinant of B and I denotes the r x r identity matrix. We have
1 1 ...

1
ba,.

(adj(B)) 1 = det

21 b :

29 b

bri br2 since on expanding the determinant along its first row we get the sum of the co factors of the entries in the first row of B, i.e. the sum of the entries on the first column of adj(B). Similar formulas hold for the other column sums and the row sums of adj(B). Lemma 4.3 Let r > 1, let B = (b) be an r x r matrix, and let C = (b + A), where A E R. Then. i(adj(C)) = k(adj(B)), p(adj(C)) = p(adj(B)). and det(C) = det(B) + A(adj(B)). Proof From the above remarks, 1 A 21 b : bri+A 1 +A 9 b
...

for i

1,... ,r.

(adj(C)) 1 = det

...

1 rA 2 b

brr+A Subtract A times the first row of this matrix from each of the other rows; this does not change the determinant of the matrix, so 1 21 b bri 1 22 b br 1
. .

ti(adj(C)) = det

...

kj(adj(B)).

Similar arguments apply for the other column sums and for the row sums. Next, 11 b +A 91 b : b.riA +A 22 b bir r+A 2 b

det(C)

...

det

A 2 br brrA 1 1 1 +A +A 21 22 b b r+A 2 b
. .

=det

11 b +A 91 b : bri + A

A br +A 2 2 1 b +A 9 b)
. . ...

brrA bir (adj(B)). 1 +At

2 br A

...

brr A

46

Iterating this calculation, we get det(C) det(B) + \(tci(adj(B)

r(idj(B)))

det(B) (adj(B)).

Note that Lemma 4.3 is not true when r = 1! (Where does the proof break down when r = 1?) This trivial remark will have a non-trivial consequence in Section 4.3. We can now state the main result about simple solutions: r
Proposition 4.4 Suppose that a matrix game F has r x r payoff matrix B, where 1.

(2)

(adj(B))

o,

(a

> 0,

and

>0, for i

the game has a unique simple solution (x*,y*), which is given by

(i(adj(B))
E(adj(B))

r(adj(B))

(pi(adj(B)) pr(adj(B)) (adj(B))

and the value of the game is

det(B) E(adj(B))

Conversely, suppose that B is invertible and that the game with payoff matrix B has a simple solution (which is necessarily unique by Proposition 4.2). Then, B satisfies the conditions in (2). We shall say that an r x r matrix B satisfying the conditions in (2) is a simple matrix. Thus, the first part of Proposition 4.4 says that a matrix game with a simple payoff matrix has a simple solution. Proof Assume first that B is simple. Let x*
X= =

,... 1 (x ,xr), where r.

k(adj(B)) fori=1 E(adj(B))

By(2),x>0foralliand

(adj(B))

i(adj(B))

1,

50 X*

E
=

Let B

(b) and let P :

R be the mixed payoff function ofF. Let

for

1,... ,r.

47

Then,
((1 (r)=X

(adj(B))
=

(in (adj(B)) (1
.

inr(adj(B))) B

F(adj(B))

1)adj(B)B.

Since adj(B)B

det(B)I, we get
j)

det(B) F(adj(B))

for

r.

Similarly, y E

and P(&,y
=

det(B) fori E(adj(B))

1,... ,r.

We have now proved that the game has the stated simple solution and value, and it remains to prove that the solution is unique. We have already noted in Proposition 4.2 that this is true if B is invertible, so assume now that det(B) = 0. Then we must have r > 1, for if r = 1 we would have B = 0 and then B would not be simple. From (2) the value of the game is 0. Let C = (ej) be the r x r matrix such that = + 1 for all i, j = 1. r. By Lemma 4.3. det(C) = E(adj(B)), so by (2), C is invertible. It follows from Problem 3.9 that, if (x, y) is a simple solution of F, it is also a simple solution of the game F with payoff matrix C, the 0 value of the latter game being 1. By the first part of the proof, Pc has a unique simple solution, hence so does F. The converse is proved by essentially reversing the above argument. Let B be invertible and let (x*, y) be a simple solution of F; let v be the value of F. Let x* = (x xv), and let P be the mixed payoff function of F. For = ,... 1 j 1,... r, we have P(x,6) = v, i.e.
. . .

xB=v(1 so

...

1), inr(adj(B))).

x*Badj(B) =v(inj(adj(B)) Since Badj(B) det(B)I, we get det(B)x Since

...

vin(adj(B)) for i

1,

r.

1, this gives det(B)

vF(adj(B)), 0. Also,
=

so since B is invertible it follows that F (adj(B)) in(adj(B)) S(adj(B)) A similar argument shows that p(adj(B)) F(adj(B)) Hence, B is simple. 0 for z
=

0 for z

1,... r.

1,...

,.

48

4.2 Convexity (again)

In fact, we can define the analogue of vertices for any convex set: Definition 4.5 Let C be a convex subset of R (m > 1). A point x E C is said m to be an extreme point of C if x is not the midpoint of any line segment whose endpoints are in C. Example 4.6 The extreme points of Am are precisely the pure strategies. To see this, we show first that = (1.0.... .0) is an extreme point of Am: a similar argument works for each of the m. pure strategies. Suppose for a contradiction that is the midpoint of a line segment with endpoints x ,... 1 (x x), y (yr,... ,Ym) EArn. Then, Xi+i = 0 for i >1, so since xj,yj >0 for all i, we have = 0 for all i > 1. Hence, x. = = = 1, contradicting the fact that x y. Conversely, let a , 1 (a am) be a proper mixed strategy, i.e. a mixed strategy that is not a pure strategy. Then, at least two of the components of a are non-zero, say a and a (i <j). Let
,

miii

1 a, 1 a}

then, e> 0. Let

a,+e....ae.

am).

It is easy to see that x, y E Am and that a point of A . 1

4(x + y). Hence, a is not an extreme

We shall accept the following result without proof.

49

Theorem 4.7 Let C be a bounded closed convex subset of R (m > 1), and assume that the set E of extreme points of C is finite. Then. C (B) (arid hence C is a convex polytope). For example, since the set rn of mixed strategies is bounded, closed and convex (see the remark following the proof of Proposition 3.16), the set of mixed strategies is the convex hull of the set of pure strategies (it is easy to see this directly).

4.3 Shapley-Snow Algorithm Proposition 3.16 shows that the sets of optimal mixed strategies for the two players in a matrix game are bounded closed convex sets. By Theorem 4.7, such a set is the convex hull of the set of its extreme points. Let us say that an optimal mixed strategy for a player is an extreme optimal strategy for the player if it is an extreme point of the set of his optimal mixed strategies. If x and y are extreme optimal strategies for players 1 and 2: respectively, then (x,y) is called an extreme solution of the game. Solving the game therefore reduces to determining the extreme solutions. Proposition 4.8 Let A be the payoff matrix of a matrix game LA, let B be a simple submatrix of A, and let LB be a game with payoff matrix B. Let (,i) be a simple solution of LB, and let (x*, y*) be its extension to a pair of mixed strategies for LA. Then, if (x, y*) is a mixed strategy solution of F. it is an , 4 extreme solution of LA. Moreover, F and LB have the same value. 4 Conversely, any extreme solution of LA is either a pure strategy solution of LA or is the extension to FA of the simple solution of a game with payoff matrix a simple submatrix B of A. Remark Note that any pure strategy solution is obviously an extreme solution, since any pure strategy is an extreme point of the set of all mixed strategies by Example 4.6, and a fortiori any optimal pure strategy is an extreme point of the set of optimal mixed strategies. Before we spell out the algorithm which this proposition gives us, we point out some corollaries. Let us say that a mixed strategy x = (x Xm) for a player ,... 1 uses the pure strategy i if x > 0.
,

Corollary 4.9 Each of the players of a matrix game with m x mm payoff matrix has an optimal mixed strategy which uses at most min(m, mm) pure strategies. Proof Let .4 be the payoff matrix of the game and let x* be an extreme optimal strategy for player 1 (a similar argument works for player 2). If x is an optimal pure strategy, the conclusion obviously holds. Otherwise, it is obtained by extending a simple solution of a game with payoff matrix a simple submatrix B of .4, say. Then B is r x i for some r < min(rn, mm), and x uses at most r pure strategies. Corollary 4.10 The set of optimal mixed strategies for a player of a matrix game is a convex polytope. Proof The payoff matrix of the game has only a finite number of simple submnatrices, and by Proposition 4.4 each such submatrix has a unique simple solution. Since the original game obviously has only a finite number of pure strategy solutions,

50

by Proposition 4.8 it has only finitely many extreme solutions. Hence, the set of optimal strategies for each player is the convex hull of a finite set of points. fl Proposition 4.8 translates into a method which can be used to solve any matrix game, called the Shapiey-Snow algorithm. Consider a matrix game F with m x n payoff matrix A. Let be the set of all simple submatrices of A. (Note that, if A itself is a square matrix (i.e. if m = n) it may be a simple submatrix of itself.) Step I Determine the pure strategy solutions of F, if any. Step II If B E let (,11*) be the simple solution, and v the value, of the game with payoff matrix B given by Proposition 4.4. Let (x*,y*) be the extension of ,11*) to a pair of mixed strategies for players 1 and 2, respectively, of the game F. Step III Determine which of the pairs of mixed strategies (x*, y*) found in Step H
,

are mixed strategy solutions of F. Step IV For i = 1,2, let E be the set consisting of the optimal pure strategies for player i found in Step I together with the set of optimal mixed strategies found in Step III. Then, the convex hull (E) of E is the set of all optimal mixed strategies for player i. All of this is an immediate consequence of Proposition 4.8. However, let us spell out in more detail what must be done in Step III. Let B B, and assume that B consists of the elements of rows i ,. 1 i and columns ,Jr of A. Then, the mixed strategies X = (x ,Xm) and ,... 1 ,y) for players land 2, ( respectively, associated to B as in Step II are given by
..

(adj(B)) 3 K (adj(B))

p(adj(B)) for s (adj(B))

,...

and Xi = 0 if j ,.. 1 {i ,ir}, Yj 0 if j {ji,... (x*, y*) is a mixed strategy solution of F if and only if (4) forj=l,... ,nand det(B) (adj(B)) P(x*.)

,jr}. By Proposition 3.5,

for i = 1,. m. However, we already know that (4) holds for j = Ji, j since for such values of j we have Pn(*,) = P(x*,) as is a mixed strategy solution of the game with payoff matrix B; similarly, and we know that (5) holds for i = ii Hence, we need only verify () for j {ji,... ,jr} and (5) for
. .

{ii,...

,ir}.

It is important to note that one often does not need to consider all the simple subrnatrices of A. For, once the value v of F has been determined, for example by finding a pure strategy solution as in Step I or an extreme solution as in Step III, one need only consider the simple submatrices B E B such that det(B) E(adj(B))

51

Note finally that we do not have to consider the 1 x 1 simple submatrices of A, since the corresponding extreme solutions (if any) are pure strategy solutions. (See Problem 1.8.) We now give some examples of the algorithm in action. Example 4.11 Consider the game F with mixed payoff function P and payoff matrix 1 f 1 1 2 1 1 1 0 2 1 0

.=1

For the submatrix B=2 2)

consisting of the entries in rows 1 or 3 and columns 1 or 2 of A, we find that adj(B)=(

),
(a.,

so E(adj(B)) = 6 is positive and so are the row sunis (3 and 3) and column sums (4 and 2) of adj(B). Hence, B is a simple subrnatrix of A. By Proposition 4.4, the simple solution of the game PB with payoff matrix B is where ). = (. ), and since det(B) = 0. the value of the game isO. Extending to solutions of the original game gives the mixed strategies x = (,0, ), y = (, ,O,0) for players 1,2. respectively, in F. Then (x, y) will be a mixed strategy solution of F if and only if ) 3 P(x*, > 0, P(x ) 4

0,

,y) 2 P(S <0.

We find that P(x, = 4/3 4/3 = 0, P(x, ) 3 ) 4 (x, y) is not a mixed strategy solution of P. Next, consider the submatrix B=(1

2/3, but P(,y*)

1/2, so

)
1)
i.),

consisting of the entries in rows 1 or 3 and columns 2 or 3 of A. We find that adi(B)=(j

so E(adj(B)) = 9 is negative and so are the row sums (6 and 3) and column sums (6 and 3) of adj(B). Hence, B is a simple submatrix of A. The simple solution of the game PB with payoff matrix B is where = (, ), , = (. ). and since det(B) = 0, the value of the game is 0. Extending to solutions of the original game gives the mixed strategies x = (, 0. y = (0, ., , 0) for players 1.2. respectively, in F. Then (x*, y) will be a mixed strategy solution of F if and only if (x*,5i) 4 P >0. P >0. ) (x,6 4 Pt(.y) <0.

52

We find that PA(x*,oi) = 0, PA(x,Sl) = 2/3, P = 1/3, so (x,y) is ,y) 2 (6 1 a mixed strategy solution off. Hence, ((i, O ), (0, , , 0)) is an extreme solution of P. If you have enough energy, you can consider the remaining sixteen 2 x 2 subma trices and the four 3 x 3 submatrices of A There are no pure strategy solutions. Example 4.12 A more manageable example is the game P with payoff matrix /1 .4=(3 \\3 There are no pure strategy solutions. We find that /3 adj(A)= (3 \9 Thus, S (adj(A))
=

22 0 1 3 0

6 6 3

2 7 6

15 and
= =

(adj(A)) 1 ic (adj(=4)) 1 p

(adj(A)) 2 3. i 5, p. (adj(.4)) 2

= =

(aclj(A)) 3 9, i 10, p (ad(A)) 3

= =

3. 0.

Thus, A is a simple submatrix of itself. Since det(A) = 15, the value of the game is 15/1.5 = 1 and extreme optimal strategies for players 1 and 2 are (i-, , ) and (i., , 0), respectively. We must now consider the nine 2 x 2 submatrices:
B B
=

(
(3

2):

adj(B)
=

B B
=

( 1): ) ( ):
_2): (3

( ),

E(adj(B))

6.
-3.
=

det(B)/dj(B))

-1; -1;
=

adj(B)
=

(1
= =

(adj(B))

det(B)/(adj(B)) -9,
det(B)/(adj(B))
=

adj(B) adj(B) adj(B)=

(j ),

(adj(B))

-1;

B=( B B
=
=

adj(B)
=

B
=

() ; ):
( )

), ), ),
2

E(adj(B))

5,

det(B)/(adj(B))

-7/5;

E(adj(B))=O; (adj(B))
=

-5. 1. 3,

det(B)/(adj(B)) det(B)/(adj(B))

3/5;

adj(B)
=

(
(3

(adj(B)) (adj(B))

-2:

adj(B) adj(B)

2)

det(B)/(adj(B)) det(B)/E(adj(B))

2;

B
=

( ),

E(adj(B))

4,

-3/4.

The second, fifth, sixth and seventh matrices are not simple, and the fourth, eighth and ninth, although simple, give the wrong value for the game. So only the first and third niatrices could give rise to extreme solutions.

53

The submatrix (1 3 gives rise t.o pair of mixed strategies x and 2 in F, respectively, but since
= =

2 0

(4, 4,0)

and y

(k

0) for players 1

(2) x

+ (1) x

<1,

this is not a mixed strategy solution of F. The siibrnatrix 0 (3 3 3 gives the possible extreme solution (x y), where 1 21 x=(0,), y=(,0). y* is an optimal The fact that A is a simple submatrix of itself already showed that mixed strategy for player 2, and since ) 3 (x,S 4
=

(1) x

>

1,

x* is an optimal mixed strategy for player 1. Hence, we have found the second extreme solution. The most general optimal mixed strategy for player 1 is the convex hull of the 3 subset of 1R consisting of the two points 21 131 and namely the straight line segment
=

for 0<A< 1.

(0,

Z/s 13)

Player 2 has a unique optimal mixed strategy (,

0)

54

4.4 Proof of the Algorithm

In this section we give the probf of Proposition 4.8. We use the notation in the statement of the proposition. Let A be rn x n and let B he the r x r submatrix of 4 consisting of the entries in rows i ,i,. and columns j, ,.. 1 Let P arid 4 PB be the mixed payoff functions of L,4 and TB, respectively. Since P y*) = (x. 4 PB(,rJ*). it follows that F.. and T have the same value. savu. 1 8 We now show that x* is an extreme optimal 1 in F. (a similar 4 argument will, of course, show that y* is extreme). Let x = 4(x x), where x (xi,: ,x) and x = (xv,. ,x,) are optimal mixed strategies for player 1 = 0, so since x and x7 are > 0, we have TA. Fo i ,ir}, (x + {Ii,. x=xOfori{ii where = x arid x for s = 1,... ,r. Then, = 4( +), so that
strategy for piaer
.

..

in

..

x)

= PB(,) + s) 6 s = 1,... ,r. But PB(*, = v, and since x and x are optimal mixed strategies for player 1 in TA,
for

Jj 5 PA(x, >

V,

s) 5 PB(C, = E4(x,j) =
V

v.

Hence,
s = 1, This shows that (,i) and (.i) are simple solutions of T B. But, since B is simple, Proposition 4.4 shows that TB has a unique simple solution. x*. This proves that x is extreme. Hence, and so x = x = = For the converse, let (x*. y) be an extreme solution of TA that is not a pure strategy solution. say X = (x 1 Xm), y = (yi,... ,y,), and let u be the value of F We consider two cases: . 4
for
.

Case I: v 0. y*) < v for alli, and by Problem By Definition 3.2, 3.11, P(3, y = v if z > ) 8 0. If the rows of A are permuted to give a matrix A and the same permutation is applied to the components of x to give an element x it is clear that (x, y) is an extreme solution of the game with payoff matrix A. Hence, we might as well assume that there exist positive integers p and s such that p < s < m, x > 0 if and j,y*) 6 only if i <p, PA( = v if i <s. and PA(j,y*) <v if s <i < m. Similarly, we may assume that there exist positive integers q and t such that q < t < n, y > 0 if and only if j <q, F.i(x = v if j <t, and F(x,) > v if t <j <a. ,i) 8 Consider the submatrix
11 a a

C =

1 a 81 a

am

t 8 a

of A. We shall prove that the first q columns of C are linearly independent. Simi larly, the first p rows of C are linearly independent. Assuming this for the moment, the rank r C satisfies
of

mnax{p.q} < r < min{s,t},

30

so we can choose an invertible r x r submatrix B of C which contains

1 a

apq

as a submatrix. Let B and JB be the sets of integers which index the rows and columns of B, respectively. Since {1, for , p} C B, there is a mixed strategy player 1 in the game LB with payoff matrix B which extends to the mixed strategy x for player 1 in F. Similarly, there is a mixed strategy , for player 2 in LB which extends to y* If j E JB, then 1 < j < t so Pt(x*,6j) = v. Similarly, if i E B then (6,y) 4 F v. Thus, is a simple solution of LB. Since B is an invertible matrix such that LB has a simple solution, it follows from Proposition 4.4 that B is a simple submatrix of A. To complete the proof in the case u 0, we must now show that the first q columns of C are linearly independent. For j = 1, , q, let i be the jth column of C, so that 13 a
. . .

i=

apj

3 a and suppose for a contradiction that there exist real numbers cii, zero, such that
.
. .

, c, not all

i.e.,
na=O fori=1,...,s.

Then,

(aui)

Since u point

0, we must have
y

cij

0. Put a

0 if q <

< n and consider the

2 ,y 1 = (y +An

2 +A

+Aa,) 0 y E

for A E R. Note that

+An)= yj +Aa

Yj

1.

Since y > 0 if (6)

< q, there exists

> 0 such that

>O Ac 3 y forl<j<q

56

if .! < e; arid (6) also holds for q < j < n since in that case y 0 and cj = 0. Thus, y E . 1 ifJ <e We now show that there exists 0 < e e such that y, is an optimal mixed strategy for player 2 if Aj < e. For this we compute
aj(yj +

t)

+ A
=

PA(6j,y*) +

If 1 < < s, then PA(oi,y*) = v and 1 = 0, so P(&,y>) = v. On cia the other hand, PA(Sj, y*) < v if s < i < m, so there exists 0 < e < e. such that PA(,yA) <v for all s <i < m provided that I) <e. We have thus shown that y,, is an optimal mixed strategy for player 2 if A < e. In particular, y and y are distinct optimal mixed strategies for player 2. But this is impossible, since
y
=

(y, y),

and y is extreme. This contradiction shows that the first q columns of the matrix C are indeed linearly independent. Case II: v 0 Let A be the matrix obtained by adding 1 to each entry of A. The game FA with 1 1 payoff matrix -4 has value 1 and (x, y) is an extreme solution of it. It follows from Case I that there exists a simple submatrix D of .4 (which is r x r, say) and a simple solution (,i) of the game 1 with payoff matrix D such that x* and D y* are the extensions of and i to mixed strategies for players 1 and 2 of FA , 1 respectively. Let B be the submatrix of A obtained by subtracting 1 from each entry of D. By Lemma 4.3, (adj(B)) = (adj(D)) and 9 (adj(B)) (adj(B)) 8 p = p for s = 1,... ,r. It follows that B is a simple subrnatrix (adj(D)) 5 of .4. Moreover, (,j*) is a simple solution of the game with payoff matrix B.

Problems
dl. Show that the game with payoff matrix

ft
1

1 1 2

1 3 1

has a unique simple solution, and find it. 4.2. Show that the game with payoff matrix /i

4 2 0

5 0 1

has a unique mixed strategy solution. 4.3. Repeat Problem 4.2 for the paoff matrix fi 5 4 6 0 2 3 2 2

4.4. Find the value of the game with payoff matrix /1 1 \,2 1 0 2 2 1 4

Show that player 1 has a unique optimal mixed strateg y and find all optimal mixed strategies for player 2. Sketch the set of optimal mixed strategies for player 2 as a subset of. 1.5. Find the value of the game with payoff matrix

o
1 1

0 0 2 3

1 0 0 2

given that each extreme strategy for player 2 uses three pure strategies. Show that player 2 has a unique optimal mixed strategy and find all optimal mixed strategies for player 1. 4.6. Use the methods of Section 3.4 to find the value, and the optimal mixed strategies for player 1, of the game with payoff matrix (i 2

k,.o

4 3

0 2

Find all the optimal mixed strategies for player 2. 4,7. A matrix game is said to be completely mixed if every optima l mixed strategy for each of the players uses every one of his pure strategies. Show that (i) a completely mixed game has a square payoff matrix; (ii) a completely mixed game has a unique mixed strategy solutio n; (iii) the value of a completely mixed game is non-zero if and only if its payoff matrix is invertible. 4.8. Show that every pure strategy solution of a game with payoff matrix .4 arises from a 1 x 1 simple submatrix of .4, provided the value of the game is non-zero. Why does the proof of Proposition 4.8 not show that this result is true whatev er the value of the game ? Give an example to show that the result is not, in fact, true if the value is zero.

only one such subset

4.9. Show that the open interval (0, 1) is a convex subset of that has no extreme points. Give an example of a non-empty closed convex subset of with no extreme points. (In fact, there is

can you prove that?)