Scribd
Upload
74 views50 pages

Chapter 3-Textbook

Uploaded by

Aseel mujahed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
74 views50 pages

Chapter 3-Textbook

Uploaded by

Aseel mujahed
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
You are on page 1/ 50
Chapter 3 The Integers 3.1 The Division Algorithm What are some of the important properties of the real numbers? Most of us learned early in life that they have a definite order. Later, we met the symbol < whose properties are best summarized by saying that it is a Partial order; that is, the binary relation Sis reflexive: @ Sa forallacR, antisymmetric: if q Sbandb 0 implies ac < be. 9. @< band c <0 implies ac > be. A third well-known operation, subtraction, is defined in terms of addition by the rule a-b=a+(-b). Tt is not unusual for a set of real numbers to have no smallest element; for example, there is no smallest element in the set da 44,..-}. Similarly, there is no smallest positive number. On the other hand, this sort of thing does not occur with sets of natural numbers, according to the Well Ordering Principle 3.2 WELL ORDERING PRINCIPLE. Any nonempty set of natural numbers has a PAUSE 2. smallest element. In the rest of this section, we concentrate on the integers. All the properties of paragraph 3.1 hold for the integers, in most instances because they hold for all real numbers. One exception is the closure property, which simply says that the sum and product of integers are integers. We also say that the integers are closed under addition and under multiplication, Not alll sets of real numbers are closed under these operations; for instance, the set of odd integers is not closed under addition since, for example, 1 +3 = 4, which is not odd. 1s the set of negative integers closed under multiplication? Is the set of natural numbers closed under multiplication? Is the set of natural numbers closed under subtraction? When we divide 58 by 17, we obtain a quotient of 3 and a remainder of 7. As children, many of us were taught to write 38 Z = =34+5. 17 as 17 Generally, when we divide one natural number, a, by another natural number, 2, we write a4? he aa This can be done in several ways—for example, § = 2+ 24s, in order to make the quotient and remainder unique, we require that the'remainder be lece than the divisor. 98 Chapter 3. The Integers by the Well Ordering Principle, the set of Inultiples of b which exceed a has smallest element, say (g-+1)b. So we have % Sa < (q+ 1)b, (See Fig. 3.1.) Set Hon per Since 9 S a, we have r > 0. Since (9+ 1)b > a, we have r 0 anda <0. Since ~a > 0, we can apply Theorem 3.3 to ~a and b obtaining g and r, with Or <6, such that a= bq+r. Therefore, a= (—q)b—r. Ifr <0, a = (-a)b, While ifr > 0, a = (—¢—1)b-4-(b—r) with 0< b—7 < b= |b), In elthes case, we have expressed a in the desired form. Case 2: 6 <0 and cither a> 0 or a <0. Here —b > 0, so Theorem 3.3 and Case 1 tell us that there exist integers g and 7, 0 0, we can write a = qb+b—b—r = (q+ 1)b-+ (b=) with O<—b—r<—b=[b), Again we can always express a in the desired form. The proof of uniqueness follows as in Theorem 3.3, a definite procedure for determining q and , given a and b. For a,b > 0, q is the integer part of $ and r = a — qb, as illustrated in Problem 1. For possibly negative a and/or b, the procedure is given in Proposition 3.6 below 2 ¥ {Division by 0 always causes problems! BG Sur) 100 Chapter 3. The Integers Example 1. The reader should verify that a= qb-+r in each of the following cases and, in the process, notice that q is not always what one might expect a oT alr 38] i7{ 3] 7 58 | -17|-3] 7 -58/ 17] -4] 10 —58 | -17 4| 10 ——t When a is written in the form a = qb +r, with 0 0 a {ere ifb <0. Proof. We consider the case b > 0 and leave the other possibility to the exercises. By definition, c—1 < [2] < « for any 2. Letting x = a/b and k= [a/b]; therefore, we have iz = Multiplying by the positive number b (aud using Property 8 of paragraph 3.1), we obtain a— < kb 1, the base b representation of a natural number N is the expression (4—1@n—9...ao)s Where @9,@1,.-. ,@y—1 are those integers 0 < a; < b which satisfy N = an—15"-! +ay_26"~24---tarb-+a9. Thus, (an—14n—2 ++ A0)b = An—1b"-! + ay_gb"-? 4+ +++ a1b + a9. The base 5 representation of 117 is 432, written 117 = (432)s, because 117 = 4(5*) +3(5) +2. In base 3, 117 = (11,100) because 117 1(34) + 1(33) + 1(3?). While 10 is the most familiar base, nowadays bases 2, 8, and 16 are also common. Base 2 is also known as binary, base 8 as octal, and base 16 as heradecimal notation. The basic hexadecimal digits are 0-9, A, B, C, D, E, and F. For example, 10 = Ayg, 11 = Big and (4C)1g = 4-16 + 12 = 76. To convert from base b to base 10 is easy: Use the definition of (...), in Definition 3.7). What about converting from base 10 to base b? Suppose that NV = (@q—1an—2--.ao)s is a number given in base . Then N= ina 18" * + and"? +++ + a1b+ a9 = (ay—10"-? + and" + +--+ a1)b + a0 = god + a9 where go = an—1"-? + an—2d"-> ++---+ay. Since 0 < ao 1, then ged(a,b) = b = O(a) + 1(b) is also a linear combination of a and b, Using our remark, it is clear that if ged({a|, |b) is a linear combination of |a| and |b], then so is ged(a,b). For example, since ged(1800, 756) = 36 = 8(1800) — 19(756), ged(1800,—756) = 36 = 8(1800) + 19(-756). The following theorem, the proof of which follows by noting that each remainder in the Euclidean algorithm is a linear combination of the previous two, describes the most important property of the greatest common divisor. (See also Exercise 23.) 3.16 THEOREM. The greatest common divisor of integers a and 6 is an integral linear combination of them; that is, if g = ged(a,b), then there are integers m and n such that g = ma +nb. The corollaries which follow illustrate ways in which this theorem is used. 3.17 COROLLARY. Suppose «,a,b are integers such that x | ab, Ifa and a are relatively prime, then a | b. Proof. We know that there are integers m and n so that ma-+na = 1 (because ged(x,a) = 1). Multiplying by 6, we obtain bmx + bna = b. But x | bma and x | bna because bra = n(ab) and we are given that « | ab. Thus, « divides the sum bma + bna = b. Qo 3.2. Divisibility and the Euclidean Algorithm pent 3.18 COROLLARY. The greatest common divisor of nonzero integers a and 6 is divisible by every common divisor of a and b. Proof. Let g = ged(a,b). By Theorem 3.16, g = ma+ nb for integers m and n. Thus, by Proposition 3.9, if c is a common divisor of a and 6, thene|g. O In paragraph 1.28, the greatest lower bound of two elements @ and bina Partially ordered set (4, ~) was defined to be an element 9 = ab € A with the properties 1. gXa,gXb, and 2. ife Xa and ¢ xb for some c€ A, then eg. The preceding corollary therefore shows that every pair of natural numbers has a glb in the poset (N, |), namely, their greatest common divisor: For a,b € N, aAb= ged(a,6). It is also true that every pair of natural numbers has a least upper bound. Problem 7. Suppose a, b and ¢ are three nonzero integers with a and ¢ relatively prime. Show that ged(a, 6c) = ged(a, b). Solution. Let 91 = ged(a,bc) and go = ged(a,6). Since gp | ®, we know that 92 | be. Since also gy | a, we have go < gi. On the other hand (and just as in Corollary 3.17), since ged(a,c) = 1, there are integers m and n such that ma + ne = 1. Multiplying by 6, we obtain mab + nbc = b. Since g, | a and s1 | be, it must also be the case that 9, |b. Then, since g | a, g1 < gp. Therefore, 9; = gp. a The Least Common Multiple 3.19 DEFINITION. Ifa and 6 are nonzero integers, we say that ¢ is the least common multiple (lem) of a and § and waite ¢ = lem(a,b) if and only if € is a positive integer satisfying 1. a| 6,6] ¢and, 2. if m is any positive integer such that a | m and b | m, then €< m. 2 Chapter 3. The Integers Examples 5. © The least common multiple of 4 and 14 is 28. Iem(—6,21) = 42. # lem(—5, -25) = 25. ee Since |ab| is a common multiple of a and 6, the least common multiple always exists and does not exceed |ab]. Remember also that a least common multiple is always positive (by definition). In the exercises (see also Exercise 29 of Section 3.3), we ask the reader to derive the formula (2) gcd(a, b) Iem(a, b) | which holds for any nonzero integers a and 6, and gives a quick way to compute least common multiples Examples 6. © Since ged(6,21) = 3, it follows that lem(6, 21) = (6 - 21)/3 = 126/3 = 42. * Since ged(630, -196) = 14 (Problem 5), it follows that. lem(630,—196) = (630 - 196) /14 = 8820. __—_'f Just as the greatest common divisor of a and b is divisible by all common divisors of a and b, one can show that the least common multiple of a and b is a divisor of all common multiples of a and b (see Exercise 26). Thus, the least common multiple of natural numbers a and 6 is their least upper bound in the poset (N, |) (see again paragraph 1.28). For a,b € N, aVb=lem(a,b). Remembering that a lattice is a partially ordered set in which every two elements have a greatest lower bound and a least upper bound, the following proposition is immediate. 3.20 PROPOSITION. The poset (N,|) is a lattice. 3.2. Divisibility and the Euclidean Algorithm 113 The Lattice of Divisors of a Natural Number] Implicit in the definition of lattice is the fact that the greatest lower bound and least upper bound of every pair of elements should lie in the set. The poset ({2,3,4}, |), for instance, is not a lattice because while, for example, 2 and 3 have a glb in N, this element is not in {2,3, 4} We have seen that (N,|) is a lattice. Also, for many subsets A of N, (A, |) is a lattice. For example, if n is any natural number and A = {d € N | d | n} is the set of positive divisors of n, then (A, |) is a lattice. (See Exercise 32.) Example 7. With A = {1,2,3,5,6,10,15,30}, the set of positive divisors of n = 30, (A,|) is a lattice whose Hasse diagram is shown in Fig. 3.2. Notice that every pair of elements of A has a glb and a lub. For every pair of elements a,b € A, a Ab is the unique element below (and connected with lines to) both a and b and aV b is the unique element above (and connected with lines to) both and b, aay FIGURE 3.2: The Hasse diagram for (A,|), where A is the set of divisors of 30. Compare with Fig. 1.5, p. 58. Answers to Pauses 1. For any integer n, we have 0 = qn for the integer g = 2, Since 1 | a for all a € A, no a # 1 can be 8. & minimal. The only minimal element is 1, and it is a minimum. The elements 4, 5, and 6 are maximal. For instance, 4 is maximal because there is no a € A such that 4 | a, except a = 4. There are no maximum elements; for example, 4 is not maximum because a | 4 for alla € A 1 is not true. 114 Chapter 3. The Integers 3. The largest divisor of a is jal. Since jal is a common divisor of a and b, this must be their ged. 4. Since a is a common divisor of a and 0, the greatest common divisor of these numbers is lal. 5. 17,369 1 0 5472 0 2 953 1 -3 707 -5 16 246 6 -19 215 ~LT 5 31 23° -73 29° -155 492 2 178-565 1 ~—2647 8402 The last nonzero remainder— in this case—is the ged. So 17,369 and 5472 are relatively prime. 6. 1 = —2647(17,369) + 8402(5472). EXERCISES 1. [BB] We have seen in this section that (N, |) is a partially ordered set. (a) Is it totally ordered? (b) Does it have a maximum? A minimum? Explain your answers. 2. Consider the partially ordered set ({2,4,6,8}, |). (a) Explain why every pair of elements in this poset has a greatest lower bound. (b) Does every pair of elements have a least upper bound? (c) Is the poset a lattice? Explain your answers. 3. Draw the Hasse diagrams for each of the following partial orders. (a) [BB] ({2,3,4 (b) ({1,2.3,4,5,6,7,8,9, 10}, |) 4. List all minimal, minimum, maximal, and maximum elements for each of the partial orders in Exercise 3. 3.2. Divisibility and the Buclidean Algorithm 115 5. [BB] Let n be a natural number. Given n consecutive integers, a, a + ty @+2,...,a+n—1, show that one of them is divisible by n. = Prove that n? — 2 (n an integer) is never divisible by 4. x Given that a and x are integers, a > 1, a| (11e+3), and a | (65x + 52), find a. % [BB] Suppose a and 6 are integers with the same remainder upon division by some natural number n. Prove that n | (a). © True or false? In each case, justify your answer with a proof or a coun- terexample (all variables represent integers). (a) [BB] ifa| band 6 | ~c, then a | (b) ifa| band | 6, then ac| b. (©) ifa|b anda | c, then a| be. (4) ifa | band e | d, then ac | bd. (©) ifa|bande| 2, then e| and a| 2. . a é 10. Suppose a and 6 are relatively prime integers and c is an integer such that a| cand | c. Prove that ab | c. 11. In each of the following cases, find the greatest common divisor of a and b and express it in the form ma +nb for suitable integers m and n. (a) [BB] a= 93, b= 119 (b) [BB] a = ~93, b (c) [BB] a = -93, 5 (d) a= 1575, b= 231 (e) a= 1575, b = ~231 (f) a= ~1575, b= —231 (g) a = ~3719, 6 = 8416 (h) a = 100,996, b= 20,048 (i) a = ~28,844, b = ~15,712 () a = 12,345, b = 54,301 12. Which of the pairs of integers in Exercise 11 are relatively prime? 13. [BB] Ifa and bare relatively prime integers, prove that ged(a+b,a—b) =1 or 2. 116 14. 15. 16. a: 18. 19. 20. 21. 22. 23. 24. 25. 26. j27. Chapter 3. The Integers Prove that integers a and b can have at most one greatest common divisor, Prove that for integers a and b, ged(a,a+b) = ged(a,). (Hint: Mimic the proof of Lemma 3.12.) (@) [BB] Find a pair of integers w and y such that 17,3692 + 5472y = 4, (See Pause 6.) (b) Find integers x and y such that 15dx +260y = 4. (c) [BB] Show that no pair of integers 2, y satisfy 154s + 260y = (4) Show that no pair of integers 2, y satisfy 196x + 260y = 14 (a) [BB] Given integers d, x and y, suppose there exist integers m and n such that d= mz +ny. Prove that ged(,y) | d. (b) Is the converse of (a) true? If ged(z,y) | d, need there exist integers mand n such that d= mx +ny? Ik EN, prove that ged(3k +2,5k +3) =1 [BB] Let a, b, c€ N. Prove that ged(ae, be) = e(ged(a, b)). 1 fais odd Ifa EN, prove that god(a,a +2) = { ois [BB] Prove that ged(n,m +1) = 1 for any n € N. Find integers x and y such that na + (n+ 1)y = 1. Prove that if a, b, and c are natural numbers, gcd(a,c) = 1 and 6 | c, then ged(a,b) = 1. Use the Well Ordering Principle to prove Theorem 3.16. (Hint: Consider the set of positive lincar combinations of a and b.) {BB] Find lem(63,273) and lcm(56,200). [Hint: An easy method takes advantage of equation (2).] [BB; (a)] Find the lcm of each of the pairs of integers given in Exercise 11. ' ,oered vf f Let a and 6 be nonzero integers with lem(a, 6) Let m be any common multiple of a and b. Prove that ¢ | m. he [BB] Prove that ged(a,b) | lem(a,) for any nonzero integers a and b. 3.2. Divisibility and the Euclidean Algorithm 117 28. Establish formula (2) by proving that the least common multiple of nonzero integers a and b is |ab|/ ged(a, 6). 29. (a) [BB] Let x be an integer expressed in base 10. Suppose the sum of the digits of « is divisible by 3. Prove that « is divisible by 3. (b) [BB] Prove that if an integer « is divisible by 3, then, when written in base 10, the sum of the digits of x is divisible by 3. (c) Repeat (a) and (b) for the integer 9. 3.21 DEFINITION. The greatest common divisor of n integers a1, a2,... , am, nob all zero, is a number g which is a common divisor of these integers (that is, g | a1,9 | @2,...,9 | an) and which is the largest of all such common divisors (that is, if ¢ | a,,¢ | a2,...,¢ | an, then c < g). It is denoted ged(ay,... ,an) 30. (a) Suppose a,b,c are three nonzero integers. Show that ged(a,b,¢) = ged(ged(a, ),c). (b) Show that the ged of nonzero integers a,b, is an integral linear com- bination of them. (©) Find gcd(105, 231, 165) and express this as an integral linear combi- nation of the given three integers. (d) Answer (c) for the integers 6279, 8580, and 2873. (e) Answer (c) for the integers 5577, 18,837, and 25,740. 31. Suppose that (Aj, <1) and (Ag, <2) are partial orders. (a) Show that the definition (21,22) X (1,92) => 21 1 ys and 22 <2 yp for (21, £2), (y1,y2) € Ai x Ap makes Aj x A2 a partially ordered set. (b) Let Ai = Ag = {2,3,4}. Assign to A; the partial order < and to Ay the partial order | . Partially order A, x Ag as in part (a). Show all relationships of the form (1,22) 3 (yi. ya). (c) Draw the Hasse diagram for the partial order in (b). (d) Find any maximal, minimal, maximum, and minimum elements which may exist in the partial order of (b). (e) With A; and Ag as in part (b), find those glbs and lubs which exist for each of the following pairs of elements. * (2,2), (3,3) i. (4,2), (3,4) us Chapter 3._ The Integers iii. (3,2), (2,4) iv. (3,2), (3,4) (f) Show, by example, that if (41,1) and (Ag, Xo) are total orders, then (Ai x Ag, 4) need not be a total order. 32. (a) Let n be a natural number, n > 1. Let A = {aE€N|a|n}. Prove that (A, |) is a lattice. (b) [BB] Draw the Hasse diagram associated with the poset given in (a) for n = 12. (c) Draw the Hasse diagram associated with the poset given in (a) for n=36. (d) Draw the Hasse diagram associated with the poset given in (a) for n=90. 3.3 Prime Numbers In the last section, we saw what it means to say that an integer a “divides? another integer 6. The basic building blocks of div ibility are the famous prime numbers. We examine some of the properties of prime numbers in this section. 3.22 DEFINITION. A natural number P 2 2 is called prime if and only if the only natural numbers which divide p are 1 and p. A natural number n> 1 which is not prime is called composite. Thus, n > 1 is composite if n = ab where a and b are natural numbers with 1 < a,b 1, there exists a prime p such that p | n. Proof. We give a proof by contradiction. Thus, we suppose the lemma is false. In this case, the set of natural numbers greater than 1 which are not divisible by any prime is not empty. By the Well Ordering Principle, it contains a smallest element m. Since m | m, but m is not divisible by any prime, m itself cannot be prime, so m is divisible by some natural number a, with 1 1 is not prime, then n is divisible by some prime Pause 2. number p < /n. Proof. Since n is not prime, n can be factored n = ab with 1 Y/n/n = n, a contradiction. As a natural number greater than 1, it follows that a is divisible by some prime p. Since p | a and a | n, p | n by transitivity. Also, since a < viv and p | a, we must have PS yn. Thus, p is the desired prime factor of n. a ‘As a consequence of Lemma 3.25, when testing a natural number n for pri- mality, we need only consider the possibility of a prime divisor less than or equal to the square root of n. For example, to verify that 97 is prime, we need only check the prime numbers less than or equal to V97. Since none of 2, 3, 5, 7 is a divisor of 97, we are assured that 97 is a prime number. It is apparent that Lemma 3.25 reduces dramatically the work involved in a primality check. One simple procedure which goes back over 2000 years for finding all the Primes less than or equal to some given integer is named after the Greck, Er- atosthenes (ca. 200 B.C.), chief librarian of the great library at Alexandria and a contemporary of Archimedes. Reportedly, Eratosthenes was the first person to estimate the circumference of the earth. Who was Archimedes? 3.26 THE SIEVE OF ERATOSTHENES. To find all the prime numbers less than or equal to a given natural number n, « list all the integers from 2 to n, © circle 2 and then cross out all multiples of 2 in the list, © circle 3, the first number not yet crossed out or circled, and then cross out all multiples of 3, © circle 5, the first number not yet crossed out or circled, and cross out all multiples of 5, 3.27 THEOREM. Every natural number n > 2 can be written n = pypo- 3.3. Prime Numbers 121 # at the general stage, circle the first number which is neither crossed out nor circled and cross out all its multiples, © continue until all numbers less than or equal to Vt have been circled or crossed out. When the procedure is finished, those integers which are not crossed out are the primes not exceeding n. Example 8. As an example of this procedure, we verify that the list of primes p < 100 given in Table 3.3 is correct. All the integers from 2 to 100 are listed in Figure 3.4. Initially, 2 was circled and then all even integers were crossed out with a single stroke. At the second stage, 3 was circled and all multiples of 3 not yet crossed out were crossed out with two strokes. Then 5 was circled and all multiples of 5 not yet crossed out were crossed out with three strokes. Finally 7 was circled and all multiples of 7 not previously crossed out were crossed out with four strokes. The primes less than 100 are exactly those listed earlier. =e @@r Or Oy g wu woB WM po Ww Ww Bw 19 wo oh 2 23 HK KC 3 8 Me 87 43 oh aT BO Bo 53 tt KD SCL ss gf of fi ow 67 6 gw WwW 7m bm A bp HB my Bm Bw sf 83 of AST 8D 2 98 9 MH 9 97 9S HH 100 FIGURE 3.4: The Sieve of Eratosthenes used to determine the primes pS 100. The prime numbers are considered “building blocks” becau important theorem which generalizes observations such as 8 = and 60 = 2-2-3-5. of the following 2-2,15=3-5 as the product of prime numbers p),p2,-.. ,Py of, by grouping equal primes, in the form n= qi" gg? -+-q2* as the product of powers of distinct primes q1,q2,--- .4s- Proof. If the result is false, then the set of integers n > 2 which cannot be written as the product of primes is not empty and so, by the Well Ordering 122 Chapter 3. The Integers Principle, contains a smallest element m. This number cannot be prime, so m = ab with 1 < a,b < m. By minimality of ‘m, each of a and 6 is the product of primes, hence so is m, a contradiction. Qo = 252 1176 = 2.2-2-3-7-7 21340 = 2-2-5-11-97 Example 9. ‘Theorem 3.27 can be strengthened. There is, in fact, just one way to express @ natural number as the product of primes, but in order to prove this, we need a Preliminary result which is actually a special case of something we proved earlier (see Corollary 3.17). Notice that 3 | 12 and regardless of how 12 is (properly) factored—12 = 2(6) or 12 = 3(4)—8 always divides one of the factors. The next Proposition describes the most, important property of a prime number, 3.28 PROPOSITION. Suppose a and b are integers and p is a prime such that p | ab. Then p| aor p|o. This indeed follows from Corollary 3.17 because, if p } a, then p and a are relatively prime. Now suppose prime p divides the product ayaz-+-ay of & integers. By Proposition 3.28, p | a1 or p | ag-++ay. In the latter case, applying the proposi. tion again, we see that p | az or p | a3+++a,. In this way, we obtain the following corollary. 5.29 COROLLARY. If a prime p divides the product aiay---ay of integers, then p divides one of the a; Now to prove that a natural number n > 1 can be factored in just one way as the product of primes, assume that: n can be factored in two different ways (3) n= PiP2 “Pk = 1192 °** 4 with all the pj and q; primes. After canceling equal factors on each side of this equation, we either have an equation which says that the product of certain primes is 1 (an absurdity) or an equation just like (3) where none of the primes: Pi is among the primes qj. Then, since py | pipe -++pe, it would follow that Pr | 9192 -++ ge. By Corollary 3.29, py | 4; for one of the primes g;. Since both p, and qj are primes, this forces p; = q;, a contradiction, The fact that every integer n > 1 can be written uniquely as the product of prime numbers is called the Fundamental Theorem of Arithmetic. Because of its importance, we state it again. 3.3. Prime Numbers 123 3.30 THEOREM. (The Fundamental Theorem of Arithmetic) Every integer n > 2 can be written in the form (4) N= PiP2-**Pr for unique primes py, p2, ... . prs equivalently, every integer n > 2 can be written (6) n = gStgi?+++q% (the prime decomposition of n) as the product of powers of distinct prime numbers q1,42,... ,qs- These primes and the exponents a1,02,... a are unique. As noted, the decomposition in (5) is called the prime decomposition of n. For example, the prime decomposition of 120 is 2%. 3-5. 3.31 DEFINITION. The pritiie factors or prime divisors of an integer n > 2 are the prime numbers which divide n. The multiplicity of a prime divisor p of n is the largest number @ such that p® | n. of ‘Thus, the prime factors of an integer n are the primes p; or q; given in (4) and (5) above and the multiplicities of q,,...,q, are the exponents a1,... ,a5, respectively. The prime factors of 14 are 2 and 7 and each has multiplicity 1 The prime factors of 120 are 2, 3, and 5; 2 has multiplicity 3, while 3 and 5 each have multiplicity 1. Problem 8. Find the prime decomposition of the greatest common divisor of nonzero integers a and b Solution. By the Fundamental Theorem of Arithmetic, a and b can be ex- pressed in the form a= tpipy per, b= bpp? pee for certain primes p1,p2,... ,pr and integers a1,02,... ,@r, 31, 82,... ,8,. (BY allowing the possibility that some of the a; or 3; are 0, we can assume that the same primes occur in the decompositions of both a and b.) We claim that the greatest common divisor of a and b is g=PL rin(ers Bt) pmin(oas2a) , pminarBr), where min(aj, 3;) denotes the smaller of the two nonnegative integers a; and /3;- Note that g | a since the exponent min(a;, 4) does not exceed a, the expo- nent of the prime p; in a. Similarly, g | 6. Next, assume that c | a and c | b. Since any prime which divides c also divides a and b, the only primes dividing cc must be among p1,p2,... ,pr. Thus, c = pj'p3? ---p2" for some integers ~% Since c | a, however, we must have +; < aj for each i and, similarly, + < 9) for each i because c | b. Hence, + < min(aj, ;) for each i, so c | g and the result follows. a 124 Chapter 3. The Integers Prime mumbers have held a special fascination for mankind ever since the Greeks realized there were infinitely many of them. ‘They are so familiar, yet there are many questions concerning them which are easy to state but hard to answer, | Many remain unsolved today. In the rest of this section, we briefly survey some of what is known and some of what is unknown about primes, but few details and no proofs will be given. For more information, the interested reader might consult the text by Calvin Long,? for example. Mersenne Primes While it is very difficult to determine whether or not a given large integer is prime, the problem may be easier for special classes of integers. For example, integers of the form 7n, n € N, are never prime if n > 1 since they are obviously divisible by 7. What about integers of the form n? — 1? Well, since n?— 1 = (n~ 1)(n +1), these cannot be prime either as long as n > 3. Father Marin Mersenne (1588-1648) was interested in integers of the form 2" — 1. He showed that these could only be prime if n itself was prime. (See Exercise 17.) Then he noted that it wasn’t sufficient just for n to be prime since 2!!—1 — 2047 = 23-89 is not prime. He conjectured that 2? — 1 is prime if p is any of the primes 2,3,5,7,18, 17, 19, 31, 67, 127,257 and composite for the other primes p< 257. Unfortunately, he was wrong on several counts. For example, 25! — 1 is prime while 2°°7—1 is not. (Resist the urge to factor!) As of 1994, just 32 primes of the form 2? — 1 had been found but interestingly, the list includes the largest primes known today; for example, 28°9,433 _ 1, as mentioned earlier in this section, and 271601 — 1 which, for several years, held the record as being the largest known Prime of the form 2? — 1. Today, new primes are being discovered at such a rate that it is difficult for the authors of textbooks to keep pace. In Mersenne’s honor, primes of the form'2”= are called Mersenne primes. While ‘Mersenne Primes continue to be discovered, the following problem is still unresolved, 3.32 OPEN PROBLEM. Are there infinitely many Mersenne primes? Fermat Primes Another interesting class of prime numbers is the set of so-called Fermat primes, these being prime numbers of the form 2241 For n =0,1,2,3, 4, indeed 22" +] 's prime and it was a guess of the seventeenth-century lawyer Pierre de Format (1901-1665), perhaps the most famous amateur mathematician of all time, that 2" +1 is prime for all n > 0. In 1732, however, the great Swiss mathematician SCalvin T. Long, Elementary Introduction to. Number Theory, Englewood Cliffs, NJ. Prentice-Hall, 1987. PAUSE 3. 3.3. Prime Numbers 125 Léonhard Euler (1707-1783) showed that 2?” + 1 is not prime—it is divisible by 641—and, to this day, no further Fermat primes have been discovered. When a Cray supercomputer tires of hunting for Mersenne primes, it turns to a search for Fermat primes! In a number of cases, complete factorizations of 22" 41 are known; in other cases, one or two prime factors are known. There are Fermat numbers known to be composite, though without knowledge of a single factor! As of 1995, the smallest Fermat number whose primality was unsettled was 22" +1, More details, together with an indication as to how testing huge numbers for primality is used to check the reliability of some of the world’s largest computers, can be found in an article by Jeff Young and Duncan A. Buell.* What are the first five Fermat primes? 3.38 OPEN PROBLEM. Are there more than five Fermat primes? How Many Primes Are Ther 2) We have discussed Euclid’s observation that there are infinitely many primes (Theorem 3.24), so what does the question just posed really mean? ‘There is great interest in the proportion of natural numbers which are prime. Students who have enjoyed advanced calculus will know that the series Pee Ss 1'as pipes A diverges (the partial sums increase without bound as more and more terms are added), while the series ieee eek pratptpt converges (to 72/6 in fact).> Notice that the second sum here is part of the first. So the fact that the second sum is finite while the first is infinite shows that the sequence of perfect squares 17,2?,3?,... forms only a tiny part of the sequence of all natural numbers 1,2,3,... Which of these sequences does the set of primes most resemble? Does the sum of reciprocals of the primes gra, al al converge or diverge? As a matter of fact, it diverges, so, in some sense, there really are a “lot” of prime numbers, many more than there are perfect squares, for example. {Mathematics of Computation, 50 (1988), 262-263. 5The reader who has not studied advanced calculus can relax! The ideas to which we make reference here are not critical to an overall appreciation of the mysteries of the primes. 126 Chapter 3. The Integers A quite sophisticated Tesult concerning the number of primes is the important Prime Number Theorem, which was Proven independently by J; jacques Hadamard and Charles-Jean de la Vallée-Pousin in 1896, Tt gives an approximation to the function (2), which is the number of Primes p < x. For example, z(5) = 3 since equivalently, (2) ~ is Students uncertain about limits might wish a translation! First of all, one reads “is asymptotic to” for ~. So one reads the second statement of the theorem, “n(z) is asymptotic to x/Ina” and infors that (x) is approximately equal to z/Ine for large x, the approximation getting better and better as x grows, Setting z = 100, Theorem 3.34 asserts that the number of primes p < 100 is roughly 100/In 100 ~ 21.715. Note that (100) 25 100/In100 ~ 21.745 * 1151. Setting © = 1,000,000, the theorem says that the mumber of primes under 1 million is roughly 1,000,000/ In 1,000,000 ~ 72,382. In fact, there are 78,498 such primes. Note that (1,000,000) 78,498 ae 1,000,000/In 1,000,000 ~ 72,389 © 1.084. As x gets larger, the fraction F7ivs Bets closer and closer to 1. On June 23, 1993, at a meeting at the Isaac Newton Institute in Cambridge, England, Andrew Wiles of Princeton University, announced a proof of Fermat’s Last Theorem, arguably the most famous open mathematics problem of all time. 3.85 FERMAT’S LAST THEOREM, For any integer n > 2, the equation a” +5" = on has no nonzero integer solutions a, , Notice that it is sufficient to prove this theorem just for the case that n is a prime. For example, if we knew that a9 +8 = had no integral solutions, then neither would A® + 8% = C%. If the latter had solution, so would the former, with a = A", b= B", on. 3.3. Prime Numbers 127 ? = c; for example, 3,4,5 and 5, 12,13, but are there triples of integers a,b,c satisfying a5 + 6° = c3 or a? +67 =! or a" +b" =" for any values of n except ? Pierre de Fermat was notorious for scribbling ideas in the margins of what- ever he was reading. In 1637, he wrote in the margin of Diophantus’s book Arithmetic that he had found a “truly wonderful” proof that a” + b" = c had no solutions in the positive integers for n > 2, but that there was insufficient space to write it down. Truly wonderful it must have been because for over 350 years, mathematicians were unable to find a proof, though countless many tried! Amateur and professional mathematicians alike devoted years and even life- times to working on Fermat's Last Theorem. The theorem owes its name, by the way, to the fact that it is the last of the many conjectures made by Fermat dur- ing his lifetime to have resisted resolution. By now, most of Fermat's unproven suggested theorems have been settled (and found to be true). In 1983, Gerd Faltings proved that, for each n > 2, the equation a” +0" =o? could have at most a finite number of solutions. While this was a remarkable achievement, it was a long way from showing that this finite number was zero. Then, in 1985, Kenneth Ribet of Berkeley, showed that Fermat’s Last Theorem was a consequence of a conjecture first proposed by Yutaka Taniyama in 1955 and clarified by Goro Shimura in the 1960s. It was a proof of the Shimura-Taniyama conjecture which Andrew Wiles announced on June 23, 1993, a truly historic day in the world of mathematics. The months following this announcement were extremely exciting as mathematicians all over the world attempted to understand Wiles's proof. Not unexpectedly in such a complex and lengthy argument, a few flaws were found. By the end of 1994, however, Wiles and one of his former graduate students, Richard Taylor, had resolved the remaining issues to the satisfaction of all. Among the many exciting accounts of the history of Fermat's Last Theorem and of Wiles’s work, we draw special attention to an article by Barry Cipra, “Fet- mat’s Theorem—at Last!”, which was the leading article in What's Happening in the Mathematical Sciences, Vol. 3 (1995-96), published by the American Math- ematical Society. Faltings himself wrote “The Proof of Fermat’s Last Theorem by R. Taylor and A. Wiles” for the Notices of the American Mathematical Soci- ety, Vol. 42 (1995), No. 7. In fact, many excellent accounts have been written. We cite just a few. There is one entitled “The Marvelous Proof” by Fernando Q. Gouvéa, which appeared in the American Mathematical Monthly, Vol. 101 (1994) and others by Ram Murty, Notes of the Canadian Mathematical Society, Vol. 25 (September 1993) and by Keith Devlin, Fernando Gouvéa, and Andrew Granville, Focus, Vol. 13, Mathematical Association of America (August 1993). Some Open Problems So far, we have only peeked into the Pandora's box of fascinating but unanswered problems concerning prime numbers. Here are a few more. There are intriguing questions concerning prime “gaps,” the distances be- 128 Chapter 3. The Integers tween consecutive primes. On the one hand, there are arbitrarily long gaps in the list of primes. One way to see this is to observe that if the first n +1 primes are py, p2, .-- , Pn41, then all the numbers between pipy+++py +2 and P1P2*-Pn+Pn41~Lare composite. (See Exercise 26.) On the other hand, there are also very small gaps in the primes, for example, gaps of length two such as between 3 and 5 or 11 and 13. Integers x and x +2 which are both prime are called twin primes. For example, 3 and 5, 5 and 7, 11 and 13, 41 and 43 are twin primes. Also, 1,000,000,009,649 and 1,000,000,009,651 are twin primes. Nobody, however, knows if there are infinitely many such pairs. One tantalizing result, proved by Viggo Brun in 1921 using a variation of the Sieve of Eratosthenes, is that the sum of the reciprocals of just the twin primes converges. Having noted that the sum of the reciprocals of all the primes diverges, Brun’s result is evidence that the number of twin primes is “small.” On the other hand, there are 224,376,048 twin prime pairs less than 100 billion and this seems like a pretty big number. Is the total number finite? No one knows. 3.36 THE TWIN PRIME CONJECTURE. Are there infinitely many integers x such that both x and x + 2 are prime? Observe that 4 = 2+2,6=3+3,8=543, 28=17+11, and 96 = 43453. Every even integer appears to be the sum of two primes. Is this really right? The conjecture that it is, first made by Christian Goldbach in 1742 in a letter to Euler, has proven very resistant to solution. In 1937, I. M. Vinogradov showed that every sufficiently large integer can be written as the sum of at most four primes. Sufficiently large means that there is a positive integer no such that every integer larger than no satisfies the condition. In 1966, J. Chen showed that every sufficiently large even integer can be written as x + y where x is prime and y is either prime or the product of two primes. So we seem to be close to a solution of Goldbach’s conjecture, though the last step is often the hardest. 3.37 THE GOLDBACH CONJECTURE. Can every even integer greater than 2 be written as the sum of two primes? Answers to Pauses 1. For instance, numbers of the form 7n are not prime and there are infinitely many of these. 2. Archimedes was a Greek scientist of the third century B.C. perhaps best known for the Principle of Archimedes. This states that the weight of the fluid displaced by a floating object is equal to the weight of the object itself. 3.3. Prime Numbers 129 3. The first five Fermat primes are 2241 = 24153, 2241 2) 41 = 241 = 17, 20) $1 = 28 +1 = 257, and 22) + 65537. 241 26 41 EXERCISES i o s Determine whether or not each of the following integers is a prime. (a) [BB] 157 (b) [BB] 9831 (c) 9833 (4) 55,551,111 (c) 2226.090 _ 1 . Can Lemma 3.25 be strengthened? In other words, does there exist a number r < \/n such that if n is not prime, then n has a prime factor psr? . (a) Suppose p is the smallest prime factor of an integer n and p > V/n/p. Prove that n/p is prime. (b) Express 16,773,121 as the product of primes given that 433 is this number's smallest prime factor. (a) To list the primes between 100 and 200, one could use the Sieve of Eratosthenes as described in the text to list all primes less than 200 and then ignore those less than 100, but this procedure is not very efficient. How might the procedure be modified in order to list the primes between two given integers A and B, A < B, without listing as well those primes less than A? (b) Use the method described in (a) to find all primes between 100 and 200. Find the prime decomposition of each of the following natural numbers. (a) [BB] 856 (b) 2323 (c) 6647 (a) 9970 (e) [BB] (28 - 1)” (f) 55,551,111 130 Chapter 3. The Integers 6. Use the Fundamental Theorem of Arithmetic to prove that for no natural number n does the integer 14” terminate in 0. 7. Let A ™EQ|3/n}. (a) List five different elements of A at most one of which is an integer. (b) [BB] Prove that A is closed under addition. (c) Prove that A is closed under multiplication, 8. Let A be any subset of Z {0} and, for a,b € A, define a ~ b if ab is a perfect square (that is, the square of an integer). Show that ~ defines an equivalence relation on A. 9. True or false? Explain your answers. (a) For all n EN, n > 1, there exists a prime p such that pln (b) [BB] There exists a prime p such that p | n for alln €N,n> 1. (The position of the universal quantifier makes a world of difference!) 10. Define f: N+ N.by setting f(n) equal to the largest prime divisor of n. (a) Find the range of f. (b) Is f one-to-one? (c) Is f onto? Explain your answers. 11. Determine whether or not each of the following functions N x N > N is one-to-one. Explain your answers. (a) [BB] f(n,m) = 26" (b) f(m,m) = 366" 12. (a) Find 7(10) and approximate values of 10/In10 and iets (three decimal place accuracy). (b) Find (50) and approximate values of 50/150 and ah. (c) Pind (95) and approximate values of 95/1n95 and aes: 18. (a) Use the Sieve of Eratosthenes to list all the primes less than 200. Find (200) and the values of 200/1n 200 and arietas (to three decimal places), (b) Use the Sieve of Eratosthenes to list all the primes less than 500. Find (500) and the values of 500/In500 and xi, (to three decimal places). 3.3. 14. 15. 16. 17, 18. 19. 20. 21. 22. 23. 24. 25. Prime Numbers 131 Estimate the number of primes less than 5000, less than 50,000, less than 500,000 and less than 5,000,000. [BB] Let p and q be distinct primes and n a natural number. If p |nand q| n, why must pg divide n? For any natural number n, Jet d(n) denote the number of positive divisors of n. For example, d(4) = 3 because 4 has three positive divisors; namely, 1,2, and 4, (a) Describe those natural numbers n for which d(n) = 2. (b) [BB] Describe those natural numbers n for which d(n) (c) Describe those natural numbers n for which d(n) = 5. (a) [BB] Is 2! — 1 prime? Explain your answer. (b) Is 2% —1 prime? Explain your answer. (c) Show that if 2" —1 is prime, then necessarily n is prime. (d) Is the converse of (c) true? (If n is prime, need 2” — 1 be prime?) (a) [BB] Show that 2° +1 is not prime. (b) Show that 22 +1 is not prime. (c) Show that if 2” +1 is prime, then necessarily n is a power of 2. (a) Show that the sum of two odd prime numbers is never prime. (b) Is (a) true if the word odd is deleted? [BB] Show that the sum of two consecutive primes is never twice a prime. If n is an odd integer, show that x? — y? = 2n has no integer solutions. True or false: {rn € N| n> 2 and a” +b" = c" for some a,b,c € N} = 0? [BB] Suppose that a, b, and c are integers each two of which are relatively prime. Prove that ged(ab, bc, ac) = 1. Ifa, b, and c are integers each relatively prime to another integer n, prove that the product abe is relatively prime to n. Given that p is prime, ged(a,p?) = p and ged(b, p) = p*, find (a) [BB] gea(ab,p*) (b) ged(a +b, p') 132 26. 27, 28. 29. 30. 31. 32. 33. 34. Chapter 3. The Integers - Let pi, pa, --. , Pn4i denote the first n+ 1 primes (in order). Prove that every number between pip +++ py +2 and P1P2"**Pn+Pn+1—1 (inclusive) is composite. How does this show that there are gaps of arbitrary length in the sequence of primes? If the greatest common divisor of integers a and 6 is the prime p, what are the possible values of (a) [BB] ged(a2, 5)? (b) ged(a¥, b)2 (c) ged(a?, 23)? [BB| Let a and b be natural mumbers with ged(a,b) = 1. It ab = 2? for seme natural number z, prove that a and b are also perfect squares, Let a and 6 be natural numbers, (@) [BB] Find the prime decomposition of lem(a,b) in terms of the prime decompositions of a and b and prove your answer. (See Problem 8) (b) Use (a) to prove formula (2) of Section 3.2: ged(a,8)lem(a, b) =ab. Prove that an integer which is both a square (a? for some a) and a cube (8° for some 4) is also a sixth power. Let a and b be integers. Let pbeaprime. Answer true or false and explain: (@) [BB] 1p | a, then p | a. (b) Tf p | a and p| (a? +6: , then p | b. (©) Ip | (@° +a"), then p | a. Show that there are infinitely many triples of integers a,b,c which satisfy 4h 2 (@) [BB] Prove that every odd positive integer of the form 3n +2, n € N, has prime factor of the same form. What happens if the word odd is omitted? (b) Repeat (a) for positive integers of the form 4n +3. (c) Repeat (a) for positive integers of the form 6n +5, (d) Prove that there are infinitely many primes of the form 6n +5, Ifp and p +2 are twin primes and P > 3, prove that 6 | (p +1). 3.3. Prime Numbers 133 35. Let Q* denote the set of positive rational numbers. Then, if 2 is in Q*, we may conclude that © m= orm = pi'py --- pj is the product of powers of distinct primes P1,P2,- Pri en=lorn 1:92 of qf? «+ qf" is the product of powers of distinct primes % and, in the case m # 1 and n # 1, we may assume that no p; equals any gy. Now define f: Q* + N by fQ = 1 phesphe? «Pee ifn=lmAl1 f(2) = GPirtgpfe-t .. ghhmt ifm=1n#1 pie pategihh gh? ifm él né1 (a) Find f(8) [BB], f(4), £(100) and (48). (b) Find « such that f(x) = 1,000,000 [BB], t such that f(t) = 10,000,000 and s such that such that f(s) = 365,040. (c) Show that f is a bijection. (This exercise, which gives a direct proof that the positive rationals are countable, is due to Yoram Sagher. See the American Mathematical Monthly, Vol. 96 (1989), p. 823.) 36. For positive integers a and b, define a ~ b if there exist integers n > 1 and m > 1 such that a” = 0" (a) Prove that ~ defines an equivalence relation on N. (b) Find 3, 4, and 144. (c) Find the equivalence class of a € N. (Hint: Write a = pf'p$? ---pg*.) 37. (a) Write a computer program which implements the Sieve of Eratos- thenes as described in the text. (b) Use your program to enumerate all primes less than 1000. (c) What is (1000)? Compare this with approximate values (three dec- imal place accuracy) of 1000/In 1000 and 3qz47 Og - 38. Write a computer program which implements a modified version of the Sieve of Eratosthenes in order to determine those primes which lie between two input integers A and B. (See Exercise 4.) 134 Chapter 3. The Integers 3.4 Congruence If it is 11 p.m. in Los Angeles, what time is it in Toronto? If he or she were aware that Los Angeles is three time zones west of Toronto, a person might well respond (correctly) “2 a.m.” The process by which the time in Toronto was obtained is sometimes called clock arithmetic; more properly, it is addition modulo 12: 1143 = 14 = 2 (modulo 12). It is based upon the idea of congruence, the subject of this section. 3.38 DEFINITION. Let n > 1 be a fixed natural number. Given integers a and , we say that @ is congruent to b modulo n (or a is congruent to b mod n for short) and we write a = b (mod n), if and only if n | (a— 4). The number n is called the modulus of the congruence. Examples 10. 3=17 (mod 7) because 3—17 14 is divisible by 7; (mod 3) because —2 ~ 13 = ~15 is divisible by 3; 60 = 10 (mod 25); (mod 9). 4 As a binary relation on Z, congruence is reflexive: a =a (mod n) for any integer a, symmet if a= (mod n), then b= a (mod n) and transitive: if a= b (mod n) and 6 =c (mod n), then a=c (mod n). Congruence is reflexive because a ~ a = 0 is divisible by n. It is symmetric because if n | (a — 5), then n | (b— a) because b— a = —(a— 6). It is transitive because if both a —b and b ~c are divisible by n, then so is their sum, which is a—c. Thus, for any n> 1, congruence mod n is an equivalence relation on Z. We urge the reader at this time to review the basic features of equivalence relations which were discussed in Section 1.4. Recall, for instance, that an equiva- lence relation partitions the underlying set into subsets called equivalence classes, the equivalence class of an element being the set of those elements to which it is equivalent. The equivalence classes of congruence mod n are called congruence classes. 3.89 DEFINITION. The congruence class mod n of an integer a is the set of all integers to which a is congruent mod n. It is denoted @. Thus, (Since congruence mod n is symmetric, one does not have to remember whether to write a = (mod n) or b = a (mod n) in this definition. You can’t have one without the other!) 3.4. Congruence 135 Example 11. Let n = 5. Since —8 — 17 = ~25 is divisible by 5, we have —8 = 17 (mod 5). Thus, —8 belongs to the congruence class of 17; in symbols, -8 € T7. Notice also that 17 € —8. In fact, the reader should check that =8 =T7. (See Proposition 3.40.) “1 Let us find all congruence classes of integers mod 5. To begin, T={beZ|b=0 (mod 5)} = {bEZ|5| (6-0)} = {b Z| b= 5k for some integer k} and 1 {bEZ|b=1 (mod 5)} = {b€Z|5|(b-0} = {b€Z|b—1=5k for some integer k} = {b€Z|b=5k +1 for some integer k}. In Section 1.4, we introduced the notation 5Z and 5Z + 1 for these sets 0 and T, respectively, Continuing, we find that 2={bEZ|b=5k+2 for some k € Z}=5Z+2 B= {beZ|b=5k+3 for some k € Z} = 5243 and T= {beEZ|b=5k +4 for some k EZ} =5Z +4. It is useful to observe that the five congruence classes determined so far partition the integers in the sense that they are pairwise disjoint and their union is Z. Why are they pairwise disjoint? If a € (5Z +r) (5Z +s) for0 <7,5 <4, then a = 5k-+r = 5¢-+8 for some integers k and ¢ and so r—s = 5(€—k) would be a multiple of 5. For r,s between 0 and 4 this can only happen if r= s. Why is their union Z? For any a € Z, by the Division Algorithm (Theo- rem 3.5), we can write a = 5q+7r with g,r integers and 0 1. Then the following statements are equivalent (1) n| (a-0). (2) a=b (mod n). (3) aed. (4) bea. (5) @=6. Proof. Each of the implications (1) > (2) = (3) is a direct consequence of definitions and (3) => (4) follows from the symmetry of congruence. We may therefore complete the proof by establishing (4) => (5) and (5) => (1). To soe that (4) => (5), let 6 € @ Then b = a (mod n) soa = 5 by Proposition 1.19. To see that (5) => (1), suppose that a= b. Then a € 5 Decause a € 7, so a =b (mod n) and n | (a —d). a Of all the equivalences established in Proposition 3.40, we emphasize perhaps the most important one. 3.41 COROLLARY. For integers a, b, and n with n > 1, a=b (modn) = 7=5. The next proposition generalizes the special case n = ready investigated. , Which we have al- 3.42 PROPOSITION. Any integer is congruent mod n to its remainder upon division by n. Thus, there are n congruence classes of integers mod n corresponding to each of the n possible remainders 0 a a n=1 = nZ+(n-1). These congruence classes partition Z; that is, they are disjoint sets whose union is the set of all integers. Proof. Suppose a is an integer. The remainder when a is divided by n is the number r, 0 1 is a natural number and a is any integer, a (mod n) is the remainder r, 0 < r = (1017)*(1017) = 2(2) 1 (mod 7) (mod 7) (mod 7) 3.4. Congruence 139 and so on. Here is an easy way to compute (1017)!?; (1017)'? = ((1017)?) 1 (mod 7). The following proposition describes the general principles which guarantee that the sorts of calculations we have been performing are valid. 3.45 PROPOSITION. Ifa=a (mod n) and b= (a) a+b=2-+y (mod n) and (mod n), then (b) ab = ey (mod n). Proof. Direct, proofs of each part are suggested, and these we provide. (a) We have to check that (a +) — («+ y) is divisible by n. This difference is (a— ©) + (b—y), which is the sum of two numbers each divisible by n, so the difference is itself divisible by n. (b) We have to check that ab ~ zry is divisible by n. Adding and subtracting ay, we notice that ab — xy = ab— ay + ay ~ zy = a(b — y) + (a—2)y, each term on the right again being divisible by n. Thus, ab — zy is divisible by n. o Problem 10. Suppose a and 6 are integers and 3 | (a? +b). Show that 3 Ja and 3 | b. Solution. We wish to prove that a = 0 (mod 3) and b= 0 (mod 3). If this is false, then 1 or 2 (mod 3) and so, by part (b) of Proposition 3.45, a? = or 4 (mod 3). Since 4 = 1 (mod 3), we must have a? = 1 (mod 3). Similar! & =1 (mod 3) and so, by (a) of the proposition, a? + 6? =14+1=2 (mod 3), which is a contradiction. o While addition and multiplication of congruences behave as one would hope, one must be exceedingly careful when dividing each side of a congruence. Divid- ing each side of the congruence 30 = 12 (mod 9) by 3, for example, produces the false statement 10 = 4 (mod 9). In general, one can only divide a congruence by an integer which is relatively prime to the modulus, as asserted by the next proposition. (We leave its proof to the exercises.) 3.46 PROPOSITION. If a be (mod n) and ged(c,n) = 1, then =b (mod n). For example, given 28 = 10 (mod 3), we can divide by 2 and obtain 1d (mod 3) because ged(2, 3) = 1. The linear congruences 2x = 1 (mod 7) and 6r 3 (mod 7) have the same solutions since we can divide each side of the second congruence by 3, this being relatively prime to 7. On the other hand, multiplying a congruence by a number not relatively prime to the modulus changes the congruence. For example, the solutions to 2x = 1 (mod 9) and 6x = 3 (mod 9) are different! Pause 2. 140 apter 3._ The Integers Solve 2a = 1 (mod 9) and 6x = 3 (mod 9) Problem 11. Solve each of the following pairs of congruences no solution exists, explain why not. vossible. If (a) 2r+3y=1 (mod 6) w+ 3y =4 (mod 6) (b) 2c +3y=1 (mod 6) + 3y =5 (mod 6) Solution. We solve simple systems of linear congruences by the ad hoc methods with which we fist solved systems of linear equations, by adding and subtracting and occasionally multiplying, though only by numbers relatively prime to the modulus (for a reason illustrated by PAUSE 2). (a) Adding the two congruences gives 3x + 6y = 5 (mod 6). Since 6y = 0 (mod 6), we have 3x = 5 (mod 6). This congruence has no solution because the values of 3x (inod 6) are 0 and 3. ‘Thus, no x,y satisfy the given pair of congruences. (b) This time, adding the two congruences gives 3x = 0 (mod 6) and hence v= 0.0=2orx=4 (mod 6). Ife = 0, then the second congruence says 3y = 5 (mod 6) to which there is no solution. If « = 2, then the second congruence says 2+ Sy = 5, hence 3y = 3 (mod 6), so y = 1 (mod 6), y =3 (mod 6) or y = 5 (mod 6). If © = 4, the second congruence reads 4+ 3y = 5, so 3y = 1 (mod 6) and there is no solution. The pair of congruences has three solutions: ¢ = 2 (mod 6) and y = 1 (mod 6), y = 3 (mod 6) or y =5 (mod 6) o We have seen that the congruence 3x = 6 (mod 6) may not have a solution (for example, when b= 1). On the other hand, the congruence 32 =b (mod 7) has a solution for any since each of the integers 0,1,2,3,4,5,6 is 3 (mod 7) for some 2: 30) =0, 80)=3, 3(2)=6, 3(8)=9=2, 3(4) = 12=5, 36)=15=1, 3(6)=18=4 (mod 7). A solution to 3 = 1 (mod 7), for instance, is z = 5. The difference between 3x = b (mod 6) and 3x = b (mod 7) is, of course, the modulus. In the second case, 3 is relatively prime to the modulus whereas in the first case, it is not 3.47 PROPOSITION. Let n> 1 be a natural number, let a be an integer and suppose ged(a,n) = 1. (a) There exists an integer s such that sa of a (mod n).) (mod n). (We call s an inverse (b) For any integer 6, the congruence ax =b (mod n) has a solution

You might also like