COMP9020 Lecture 3 Session 2, 2014 Sets, Functions, and Sequences. Revision: 1.3

Transcription

1 1 COMP9020 Lecture 3 Session 2, 2014 Sets, Functions, and Sequences Revision: 1.3

2 2 Notation for Numbers Definition Integers Z = {... 2, 1, 0, 1, 2,...} Reals R. : R Z floor of x, the greatest integer x. : R Z ceiling of x, the least integer x Example π = 3 = e

3 3 Simple properties x = x, hence x = x x + t = x + t and x + t = x + t, for all t Z Fact Let k, m, n Z such that k > 0 and m n. The number of multiples of k in the interval [n.. m] is m k n 1 k.

4 4 Examples 1.1.4(b) = 1; = 0 (d) 3 3 = 1; the same for every noninteger (a) Give x, y s.t. x + y < x + y. 3π + e = = 11 < 12 = = 3π + e

5 5 Divisibility Let m, n Z. m n means m is a divisor of n, defined by n = km for some k Z Also stated: n is divisible by m, m is a divisor of n m n - negation of m n Notion of divisibility applies to all integers positive, negative and zero. 1 m, 1 m, m m, m m, for every m 0 n 0 for every n; 0 n except n = 0

6 6 Notions derived from divisibility Numbers > 1 divisible only by 1 and themselves are primes. The greatest common divisor of n, m Z is gcd(m, n) = max { d N : d m d n }. Numbers m and n s.t. gcd(m, n) = 1 are said to be relatively prime or co-prime. The least common multiple of n, m Z is lcm(m, n) = min { d N : m d n d }. NB gcd(m, n) and lcm(m, n) are always taken as positive gcd( 4, 6) = gcd(4, 6) = gcd( 4, 6) = gcd(4, 6) = 2 lcm( 5, 5) =... = 5

7 7 Absolute Value x = { x, if x 0 x, if x < 0 NB gcd(m, n) lcm(m, n) = m n

8 Examples True or False. Explain briefly. (a) n 1 only if n = 1 (for n Z also n = 1) (b) n n always (c) n n 2 always 1.2.7(b) gcd(0, n) = n Can two even integers be relatively prime? No; why? (b) Find m, n, s st 1 m + 1 n = 1 s with s < lcm(n, m). In fact, if such s exists it must be s < m, s < n. NB Representing fractions (i.e. rational numbers) in the form r = 1 n n n k, n 1 > n 2 >... > n k 8 is called Egyptian fractions. Note that all denominators must be different; it is always possible, but not always easy.

9 Examples True or False. Explain briefly. (a) n 1 only if n = 1 (for n Z also n = 1) (b) n n always (c) n n 2 always 1.2.7(b) gcd(0, n) = n Can two even integers be relatively prime? No; why? (b) Find m, n, s st 1 m + 1 n = 1 s with s < lcm(n, m). In fact, if such s exists it must be s < m, s < n. NB Representing fractions (i.e. rational numbers) in the form r = 1 n n n k, n 1 > n 2 >... > n k 9 is called Egyptian fractions. Note that all denominators must be different; it is always possible, but not always easy.

10 Examples True or False. Explain briefly. (a) n 1 only if n = 1 (for n Z also n = 1) (b) n n always (c) n n 2 always 1.2.7(b) gcd(0, n) = n Can two even integers be relatively prime? No; why? (b) Find m, n, s st 1 m + 1 n = 1 s with s < lcm(n, m). In fact, if such s exists it must be s < m, s < n. NB Representing fractions (i.e. rational numbers) in the form r = 1 n n n k, n 1 > n 2 >... > n k 10 is called Egyptian fractions. Note that all denominators must be different; it is always possible, but not always easy.

11 11 Sets A set is defined by the collection of its elements. Sets are typically described by: (a) Explicit enumeration of their elements S 1 = {a, b, c} = {a, a, b, b, b, c} = {b, c, a} =... three elements S 2 = {a, {a}} S 3 = {a, b, {a, b}} S 4 = {} S 5 = {{{}}} S 6 = { {}, {{}} } two elements zero elements three elements one element two elements

12 12 (b) Specifying the properties their elements must satisfy; the elements are taken from some universal domain. A typical description involves a logical property P(x) S = { x : x X P(x) } = { x X : P(x) } We distinguish between an element and the set comprising of this single element. Thus always a {a}. Set {} is empty (no elements); set {{}} is nonempty it has one element. There is only one empty set; only one set consisting of a single a; only one set of all natural numbers.

13 13 (c) Constructions from other sets (already defined) Union, intersection, set difference, symmetric difference, complementation Power set P(X ) = { A : A X } Cartesian product (below) Empty set, also written {} X for all sets X. S T S is a subset of T ; includes the case of T T S T a proper subset: S T and S T NB An element of a set and a subset of that set are two different concepts a {a, b}, a {a, b}; {a} {a, b}, {a} / {a, b}

14 14 Cardinality Number of elements in a set X ; (various notations) X = #(X ) = card(x ) Always P(X ) = 2 X [m.. n] interval of integers; it is empty if n < m [m.. n] = n m + 1, for n m = 0 P( ) = { } P( ) = 1 P(P( )) = {, { }}... {a} = 1 P({a}) = {, {a}} P({a}) = 2...

15 15 Examples Find the cardinalities of sets 1 { 1 n : n [1.. 4] } = 4 four indices, no repetitions of values 2 { n 2 n : n [0.. 4] } = 4 one repetition of value 3 { 1 n 2 : n P 2 n n < 11 } = 5 4 { 2 + ( 1) n : n N } = 2 which are the two elements?

16 16 Sets of Numbers Natural numbers N = {0, 1, 2,...} Positive integers P = {1, 2,...} common notation N >0 = Z >0 = N \ 0 Integers Z = {..., n, (n 1),..., 1, 0, 1, 2,...} Rational numbers (fractions) Q = { } m n : m, n Z, n 0 Real numbers R In P N Z different symbols denote different numbers. In Q and R the standard representation is not necessarily unique.

17 17 NB Proper ways to introduce reals include Dedekind cuts and Cauchy sequences, neither of which will be discussed here. Natural numbers etc. are either axiomatized or constructed a from sets (0 = def {}, n + 1 = def n {n}) a If we need to emphasise that an object (expression, formula) is defined through an equality we use the symbol def =. It denotes that the object on the left is defined by the formula/expression given on the right.

18 18 Number sets and their containments Derived sets of positive numbers P N Z Q R P = N >0 = Z >0 = {n : n 1} Q >0 = {r : r = k l > 0} R >0 Derived sets of integers 2Z = { 2x : x Z } 3Z + 1 = { 3x + 1 : x Z } the even numbers

19 19 Intervals of numbers (applies to any type) [a, b] = { x : a x b } (a, b) = { x : a < x < b } [a, b] [a, b) (a, b] (a, b) NB (a, a) = (a, a] = [a, a) = ; however [a, a] = {a}. Intervals of P, N, Z are finite: if m n [m.. n] = {m, m + 1,..., n} [m.. n] = n m + 1

20 20 Set Operations Union A B; Intersection A B Note that there is a correspondence between set operations and logical operators (to be discussed in Topic-2) One can match set A with that subset of the universal domain, where the property a holds, then match B with the subset where b holds. Then A B a or b; A B a and b We say that A, B are disjoint if A B =. NB A B = B A B A B = B A B

21 21 Other set operations A \ B difference, set difference, relative complement. It corresponds (logically) to a but not b A B symmetric difference A B = def (A \ B) (B \ A) It corresponds to a and not b or b and not a; it is termed xor (exclusive or) A c = A set complement w.r.t. the universe. It corresponds to not a.

22 22 Cartesian Product S T = def { (s, t) : s S, t T }, where (s, t) is an ordered pair n i=1s i def = { (s 1,..., s n ) : s k S k, for 1 k n } S 2 = S S, S 3 = S S S,..., S n = n 1 S,... S =, for every S S T = S T, n i=1 S i = n i=1 S i

23 23 Venn diagrams p23 26: are a simple graphical tool to reason about the algebraic properties of set operations. De Morgan laws p 25, Table 1: are the rules satisfied by set operations, esp. those involving complementation. They are often termed set algebra rules, due to their partial correspondence to the algebra of positive integers.

24 24 Examples Σ = {a, b} (a) All subsets of Σ :, {a}, {b}, {a, b} (b) P(Σ) = A A? =, A? = A for all A Relate the cardinalities A B = A + B A B hence A B + A B = A + B A \ B = A A B A B = A + B 2 A B

25 25 Formal Languages Σ alphabet, a finite, nonempty set Examples (of various alphabets and their intended uses) Σ = {a, b,..., z} for single words (in lower case) Σ = {,, a, b,..., z} for composite terms Σ = {0, 1} for binary integers Σ = {0, 1,..., 9} for decimal integers The above cases all have a natural ordering; it is not required in general, thus the set of all Chinese characters forms a (formal) alphabet.

26 26 word - any finite string of symbols from Σ ω = aba, ω = ,... also: empty word λ length(ω) # of symbols in ω length(aaa) = 3, length(λ) = 0 The only operation on words (discussed here) is concatenation, written as juxtaposition νω, ωνω, abω, ωbν,... NB λω = ω = ωλ length(νω) = length(ν) + length(ω)

27 27 Notation: Σ k set of all words of length k We often identify Σ 0 = {λ}, Σ 1 = Σ. Σ set of all words (of all lengths) Σ + set of all nonempty words (of any positive length) Σ = Σ 0 Σ 1 Σ 2... ; Σ n = Σ + = Σ 1 Σ 2... = Σ \ {λ} n i=0 Σ i A language is a subset of Σ. Typically, only the subsets that can be formed (or described) according to certain rules are of interest. Such a collection of descriptive/formative rules is termed a grammar.

28 28 Examples Number of elements in the sets (cont d) (e) Σ where Σ = {a, b, c} Σ = (f) { ω Σ : length(ω) 4 } where Σ = {a, b, c} Σ 4 = = = = 121

29 29 Examples Number of elements in the sets (cont d) (e) Σ where Σ = {a, b, c} Σ = (f) { ω Σ : length(ω) 4 } where Σ = {a, b, c} Σ 4 = = = = 121

30 30 Functions and their Properties We deal with functions as a set-theoretic concept, it being a special kind of correspondence (between two sets). f : S T describes pairing of the sets: it means that f assigns to every element s S a unique element t T. To emphasise where a specific element is sent, we can write f : x y, which is equivalent to f (x) = y. S domain of f, symbol: Dom(f ) T codomain of f, symbol: Codom(f ) { f (x) : x Dom(f ) } image of f, symbol: Im(f ) Im(f ) Codom(f ) We observe that every function maps its domain onto its image, but only into its codomain.

31 31 Function is called 1 1 (one-to-one) or injective if different x implies different f (x), i.e. f (x) = f (y) x = y Examples (of functions that are not 1 1) absolute value floor, ceiling length of a word most functions of two or more arguments, when viewed as a function of a single pair of arguments

32 32 Composition of Functions Auxilliary notation f : x y, f : A B The former means that x is mapped to y; the latter means that B is the image of A under f. NB Observe the difference between and Composition of functions is described as g f : x g(f (x)), requiring Im(f ) Dom(g)

33 33 If a function maps a set into itself, i.e. when Dom(f ) = Codom(f ) (and thus Im(f ) Dom(f )), the function can be composed with itself iterated f f, f f f,..., written also f 2, f 3,... Composition is associative h (g f ) = (h g) f, can write h g f Identity function on S i S (x) = x, x S; Dom(i) = Codom(i) = Im(i) = S For g : S T g i S = g, i T g = g

34 34 Examples Regarding length : {a, b} N (c) length(λ) = 0 (d) Im(length) = N Σ as above and g(n) = def { ω Σ : length(ω) n }, n N. Here g(n) is a function that has a complex object as its value for any given argument it maps N into P(Σ ). (a c) g(0)? = {λ} g(1)? = {λ, a, b}, g(2)? = {λ, a, b, aa, ab, ba, bb} In general g(n) = n i=0 Σi = Σ n.

35 35 Examples (cont d) (d) Are all g(n) finite? Yes. (e) Give an example of a set in P(Σ ) that is not in Im(g) any infinite subset of Σ (infinite language) any finite language that excludes some intermediate length words, e.g. {λ, a}, {a, b}, {λ, a, aa},... Find g(n) It is n = 2 n Regarding gcd : P P P (c) Im(gcd) = P as gcd(n, n) = n x 3 x f (x) = x 0 x < 1 x 3 x < 0 (c) Im(f )? = R 0

36 36 Examples (cont d) (d) Are all g(n) finite? Yes. (e) Give an example of a set in P(Σ ) that is not in Im(g) any infinite subset of Σ (infinite language) any finite language that excludes some intermediate length words, e.g. {λ, a}, {a, b}, {λ, a, aa},... Find g(n) It is n = 2 n Regarding gcd : P P P (c) Im(gcd) = P as gcd(n, n) = n x 3 x f (x) = x 0 x < 1 x 3 x < 0 (c) Im(f )? = R 0

37 37 Examples (cont d) (d) Are all g(n) finite? Yes. (e) Give an example of a set in P(Σ ) that is not in Im(g) any infinite subset of Σ (infinite language) any finite language that excludes some intermediate length words, e.g. {λ, a}, {a, b}, {λ, a, aa},... Find g(n) It is n = 2 n Regarding gcd : P P P (c) Im(gcd) = P as gcd(n, n) = n x 3 x f (x) = x 0 x < 1 x 3 x < 0 (c) Im(f )? = R 0

38 38 Examples (cont d) (d) Are all g(n) finite? Yes. (e) Give an example of a set in P(Σ ) that is not in Im(g) any infinite subset of Σ (infinite language) any finite language that excludes some intermediate length words, e.g. {λ, a}, {a, b}, {λ, a, aa},... Find g(n) It is n = 2 n Regarding gcd : P P P (c) Im(gcd) = P as gcd(n, n) = n x 3 x f (x) = x 0 x < 1 x 3 x < 0 (c) Im(f )? = R 0

39 39 gcd Example Reconsider gcd as a higher-order function, defined by m if m = n gcd(f )(m, n) = f (m n, n) if m > n f (m, n m) if m < n Its type is now gcd : (P 2 P) (P 2 P) that is, it maps each partial function (from pairs of positive integers to a positive integer) to a (partial) function of the same type. The worst such function is the nowhere defined function f (m, n) =.

40 40 Consider the sequence gcd Example cont d f, gcd(f ), gcd(gcd(f )),..., gcd i (f ),... and observe that the i th element of this sequence is an approximation of the gcd function that works as long as the depth of the recursion is less than i 1. Since we proved that the original gcd function terminates, we can deduce that the limit of this sequence exists, and is the original gcd. It also is the least fixpoint of gcd i.e. the simplest solution f to the equation f = gcd(f ). This, in a nutshell, explains how the semantics of recursive procedures is defined in CS. How all this works is somewhat beyond the scope of COMP9020 but still serves the purpose of motivating why we discuss functions, their composition, iteration, and fixpoints.

41 41 Properties of Functions We ve met: one-to-one (1 1), onto Inverse function f 1 : T S; for a given f : S T exists exactly when f is both 1 1 and onto. Image of a subdomain A under a function f (A) = { f (s) : s A } = { t T : t = f (s) for some s A } Inverse image - f (B) = { s S : f (s) B } S; it is defined for every f If f 1 exists then f (B) = f 1 (B) f ( ) =, f ( ) =

42 42 Examples f and g are shift functions N N defined by f (n) = n + 1, and g(n) = max(0, n 1) (c) f is 1 1, not onto: f (N) = N \ {0} = P (d) g is onto, not 1 1: g(0) = g(1) (e) f and g do not commute: g f : n (n + 1) 1 = n, thus g f = i N f g : 0 1, hence f g i N NB f g is the identity when restricted to P

43 43 NB For a finite set S and f : S S the properties 1 onto, and are equivalent. (Proof suggestion?) Examples 1.7.6(c) Is length : Σ N onto? Yes: length ({n}) = Σ n (d) length (2)? = {aa, ab, ac, bb,..., cc}

44 44 NB For a finite set S and f : S S the properties 1 onto, and are equivalent. (Proof suggestion?) Examples 1.7.6(c) Is length : Σ N onto? Yes: length ({n}) = Σ n (d) length (2)? = {aa, ab, ac, bb,..., cc}

45 45 Examples Verify that f : R R R R defined by f (x, y) = (x + y, x y) is invertible. The inverse is f 1 (a, b) = ( a+b 2, a b 2 ); substituting shows that f f 1 = i R R

46 46 Sec. 1.8 Supplementary exercises 1.8.2(b) When (A \ B) \ C = A \ (B \ C)? From Venn diagram (A \ B) \ C = A B C; A \ (B \ C) = (A B) (A C). Equality would require that A C A B C; however, these two sets are disjoint, thus equal only when both are empty. One verifies that A C = is also a satisfactory condition and that, in this case, both set expressions simplify to A \ B How many third powers are 1, 000, 000 and end in 9? (solve without calculator). n 3 = 9 (mod 10) only when n = 9 (mod 10), and n 3 1, 000, 000 when n 10. Hence all such n are 9, 19,..., 99. Try the same question for n 4.

47 47 Examples Σ = {a, b}; relate it to Σ (a) Is there an onto Σ Σ? No: Σ = 2, Σ =. (b) Is there an onto Σ Σ? Yes, eg f (ω) = a when length(ω) - odd, f (ω) = b when length(ω) - even. The following is not completely correct f : ω first letter of ω Reason: f (λ) is not defined.