The mathematics of relations A set is a collection of objects, called its elements, or members. Examples: the set of all dogs the set of students in this class the set comprising the numbers 1, 2, and 3 Typically, we use uppercase Roman letter A, B, ..., S, ... for sets For elements we typically use lowercase Roman letters a, b, c, ..., x, y, z ‘aS’ means that a is a member of the set S A set is standardly defined in two ways: by providing a list of its members, or by providing a property which picks out its members in some unique way Examples: S1 = {1, 2, 3} S2 = {x : x is a natural number between 0 and 4} Beside these two ways, there is an informal way of defining sets by drawing a diagram: 1 2 S3 3 Identity of sets (the principle of extension): Two sets are identical iff they have the same members Notation: iff stands for ‘if and only if’ therefore, S1 = S2 = S3 S1 = {1, 2, 3} S2 = {x : x is a natural number between 0 and 4} 1 2 S3 3 From the principle of extension it follows: {x, y} = {y, x} Ordered pairs: (x, y) or <x, y> (x, y) (y, x) (x, y) = (x’, y’) iff both x = x’ and y = y’ Example: {(x, y) : x is a natural number between 0 and 4, y = x2} = = {(1, 1), (2, 4), (3, 9)} Special symbol for the empty set: Quantifiers: , universal quantifier: x(x 1) existential quantifier: y(y2 = 1) Subsets: S S’ (S is a subset of S’) iff x(if xS, then xS’) Real subsets: S S’ (S is a real subset of S’) iff S S’ but not S’ S All of known mathematics can be built out of sets and logic. In particular, everything in this course can. Relations In math, science, and everyday life we are continually concerned with relationships among objects. It is therefore useful to have a mathematical language that can talk about relations. Examples of relations: x is bigger than y x is better than y x loves y x causes y x is preferred to y Notation: We denote relations by uppercase Roman letters, e.g. R, R’. Some important relations have special symbols, e.g. >, =. We write xRy to mean that relation R holds between x and y Defining relations: We will use three ways to define particular relations: 1. List all the objects of interest between which the relation holds Examples: Let S = {Victoria, Alberta, Manitoba, Edmonton, Winnipeg, B.C.} be our set of objects of interest. Let R be the relation ‘is capital of’ Then R is defined as: Victoria R B.C. Edmonton R Alberta Winnipeg R Manitoba Let S’ = {1, 2, 3} Then the relation ‘>’ is defined on the set S’ as: 3>2 3>1 2>1 2. By description xRy holds iff x is the capital of y and both x and and y are in S x > y iff x is greater than y and both x and y are in S’ 3. By picture Victoria B.C. Edmonton Alberta Winnipeg Manitoba Formal definition of relations: Relations are sets A binary relation is a set of ordered pairs A relation is always defined on a set Thus R is defined on S, > is defined on S’ R = {(Victoria, B.C.), (Edmonton, Alberta), (Winnipeg, Manitoba)} > = {(3, 2), (3, 1), (2, 1)} Properties of relations: A relation R on a set S is: reflexive iff symmetric iff transitive iff total (connected) (xS) xRx (x,yS) if xRy then yRx (x,y,zS) if xRy and yRz, then xRz iff (x,yS) either xRy or yRx Functions Functions are relations Definition: A relation F on a set S is a function if F relates all elements of S to at most one element of S, i.e. (x,y,zS) if xFy and xFz, then y z Notation: Functions are denoted by the lowercase Roman letters: f, g, u, p If f is a function, we typically write f(x) y instead of xfy If f(x) y, then x is called the argument of function f, and y is called the value of f for x. Examples: ‘son-of-father’ ‘is-capital-of’ ‘is-parent-of’ is a function is a function is not a function Preference relations and score functions Decision theory studies relations of preference among a given set of options. Notation: Let O be a set of options among which an agent A is choosing (flavours of ice cream, stocks, lotteries, strategies in a game, etc.). 1. We write x y to mean that ‘A prefers option x to option y’ 2. We write x ~ y to mean that ‘A is indifferent between x and y’ 3. We write x y to mean that ‘A does not prefer option y to option x’ is called strict preference relation is weak preference relation ~ is indifference relation Some preference relations are rankings of the available options. Rankings of options can be represented by assigning a number to each option such that an option is preferred just in case its number is greater. Example: Suppose Jane ranks TV shows as follows. Seinfeld is the best, then Friends, then the Simpsons score(Seinfeld) 3 score(Friends) 2 score(Simpsons) 1 x y iff score(x) > score(y) It’s only order that matters: score2(Seinfeld) 1000 score2(Friends) 2 score2(Simpsons) –123 x y iff score2(x) > score2(y) We take the weak preference relation as a primitive notion Strict preference and indifference relations are then defined as: xy x~y iff iff x y and not y x x y and y x Definition of the score function (or utility function): A score function or utility function u assigns a number to each option in a given set O. A utility function represents a preference relation on a set of options O just in case (x,yO) x y holds iff u(x) u(y) Rational preferences Definition: A preference relation is rational iff it is: 1. total 2. reflexive 3. transitive Theorem 1: If a weak preference relation is rational, then 1. the corresponding strict preference relation is transitive 2. the corresponding indifference relation ~ is reflexive and transitive Theorem 2 Given a finite set of options, there is a score function that represents a weak preference relation just in case is rational. Money pumping coke 3 sprite 2 juice 1 1 2 csjc 3 1 2 3