2. a. 2; 10 b. 4; 40 c. 4: 120; d. 1; 1050 e. pq2 ; p2q3
3. 12, 2, 2, 10, 1, 0, 4, 5.
4. s = 3, t = 2; s = 8, t = 5
5. Let a be the least common multiple of every element of the set and b be any common multiple of every element of the set. Write b = aq + r where 0 r < a. Then, for any element c in the set, we have that c divides b aq = r. This means that r is a common multiple of every element of the set and therefore is greater than or equal to a, which is a contradiction.
6. If n = 0 mod 3, we are done. If n = 1 mod 3, the n + 2 = 0 mod 3. If n = 2 mod 3, then n + 1 = 0 mod 3.
7. By using 0 as an exponent if necessary, we may write a = pm 1 pm k and b = p 1 · · · p k , where the p’s are distinct primes and the m’s and n’s are nonnegative. Then lcm(a, b) = p 1 · · · p k , where si = max(mi, ni) and gcd(a, b) = p 1 · · · p k , where k ti = min(mi, ni) Then lcm(a, b) · gcd(a, b) = p1 1 1 · · · p = ab.
≤ n n 1 k 1 k s s t t 1 k m +n m +n 1 k k k
8. The first part follows from the Fundamental Theorem of Arithmetic; for the second part, take a = 4, b = 6, c = 12.
9. Write a = nq1 + r1 and b = nq2 + r2, where 0 ≤ r1, r2 < n We may assume that r1 ≥ r2. Then a b = n(q1 q2) + (r1 r2), where r1 r2 ≥ 0. If a mod n = b mod n, then r1 = r2 and n divides a b If n divides a b, then by the uniqueness of the remainder, we have r1 r2 = 0. Thus, r1 = r2 and therefore a mod n = b mod n.
10. Write as + bt = d. Then, a ′ s + b′t = (a/d)s + (b/d)t = 1.
11. By Exercise 9, to prove that (a + b) mod n = (a ′ + b′) mod n and (ab) mod n = (a ′b′) mod n it suffices to show that n divides (a + b) (a ′ + b′) and ab a ′b′ . Since n divides both a a ′ and n divides b b′ , it divides their difference. Because a = a ′ mod n and b = b′ mod n, there are integers s and t such that a = a ′ + ns and b = b′ + nt. Thus ab = (a ′ + ns)(b′ + nt) = a ′b′ + nsb′ + a ′nt + nsnt. Thus, ab a ′b′ is divisible by n.
12. Write d = au + bv. Since t divides both a and b, it divides d. Write s = mq + r, where 0 ≤ r < m. Then, r = s mq is a common multiple of both a and b so r = 0.
13. Suppose that there is an integer n such that ab mod n = 1. Then there is an integer q such that ab nq = 1. Since d divides both a and n, d also divides 1. So, d = 1. On the other hand, if d = 1, then by the corollary of Theorem 0.2, there are integers s and t such that as + nt = 1. Thus, modulo n, as = 1.
15. By the GCD Theorem there are integers s and t such that ms + nt = 1. Then m(sr) + n(tr) = r.
16. It suffices to show that (p2 + q2 + r2) mod 3 = 0. Notice that for any integer a not divisible by 3, a mod 3 is 1 or 2 and therefore a2 mod 3 = 1. So, (p2 + q2 + r2) mod 3 = p2 mod 3 + q2 mod 3 + r2 mod 3 = 3 mod 3= 0.
17. Let p be a prime greater than 3. By the Division Algorithm, we can write p in the form 6n + r, where r satisfies 0 r < 6. Now observe that 6n, 6n + 2, 6n + 3, and 6n + 4 are not prime.
18. By properties of modular arithmetic we have (71000 ) mod 6 = (7 mod 6)1000 = 11000 = 1. Similarly, (61001) mod 7 = (6 mod 7)1001 = 11001 mod 7 = 1 = 6 mod 7.
19. Since st divides a b, both s and t divide a b. The converse is true when gcd(s, t) = 1.
20. If n = ab, where 1 < a < n and 1 < b < n, then both a and b are terms in the product (n 1)!. But then, n = ab is a factor of (n 1)! and therefore, (n 1)! = 0 mod n
21. If gcd(a, bc) = 1, then there is no prime that divides both a and bc. By Euclid’s Lemma and unique factorization, this means that there is no prime that divides both a and b or both a and c Conversely, if no prime divides both a and b or both a and c, then by Euclid’s Lemma, no prime divides both a and bc.
22. If one of the primes did divide k = p1p2 · · · pn + 1, it would also divide 1.
23. Suppose that there are only a finite number of primes p1 , p2 , . . . , pn . Then, by Exercise 22, p1 p2 . . . pn + 1 is not divisible by any prime. This means that p1 p2 pn + 1, which is larger than any of p1 , p2 , , pn , is itself prime. This contradicts the assumption that p1 , p2 , . . . , pn is the list of all primes.
24. Let z1 √= a + bi and z2 √ = c + di. Then, z 1 √ z 2 = (
+ bc); |z
|
|
|
25. x NAND y is 1 if and only if both inputs are 0; x XNOR y is 1 if and only if both inputs are the same.
26. If x = 1, the output is y, else it is z.
≤ ·
27. Let S be a set with n + 1 elements and pick some a in S By induction, S has 2n subsets that do not contain a. But there is one-to-one correspondence between the subsets of S that do not contain a and those that do. So, there are 2 2n = 2n+1 subsets in all.
28. Use induction and note that 2n+132n+2 1 = 18(2n32n) 1 = 18(2n33n 1) + 17.
29. Consider n = 200! + 2. Then, 2 divides n, 3 divides n + 1, 4 divides n + 2, . . ., and 202 divides n + 200.
30. Use induction on n.
31. Say p1p2 · · · pr = q1q2 · · · qs, where the p’s and the q’s are primes. By the Generalized Euclid’s Lemma, p1 divides some qi, say q1 (we may relabel the q’s if necessary). Then p1 = q1 and p2 · · · pr = q2 · · · qs Repeating this argument at each step we obtain p2 = q2, · · · , pr = qr and r = s.
32. 47. Mimic Example 17.
33. Suppose that S is a set that contains a and whenever n a belongs to S, then n + 1 S. We must prove that S contains all integers greater than or equal to a. Let T be the set of all integers greater than a that are not in S and suppose that T is not empty. Let b be the smallest integer in T (if T has no negative integers, b exists because of the Well Ordering Principle; if T has negative integers, it can have only a finite number of them so that there is a smallest one). Then b 1 ∈ S, and therefore b = (b 1) + 1 ∈ S. This contradicts our assumption that b is not in S.
34. By the Second Principle of Mathematical Induction, fn = fn 1 + fn 2 < 2n 1 + 2n 2 = 2n 2 (2 + 1) < 2n
35. For n = 1, observe that 13 + 23 + 33 = 36. Assume that n3 + (n + 1)3 + (n + 2)3 = 9m for some integer m. We must prove that (n + 1)3 + (n + 2)3 + (n + 3)3 is a multiple of 9. Using the induction hypothesis we have that (n + 1)3 + (n + 2)3 + (n + 3)3 = 9m n3 + (n + 3)3 = 9m n3 + n3 + 3 · n2 · 3 + 3 · n · 9 + 33 = 9m + 9n2 + 27n + 27 = 9(m + n2 + 3n + 3).
36. You must verify the cases n = 1 and n = 2. This situation arises in cases where the arguments that the statement is true for n implies that it is true for n + 2 is different when n is even and when n is odd.
37. The statement is true for any divisor of 83 4 = 508.
38. One need only verify the equation for n = 0, 1, 2, 3, 4, 5. Alternatively, observe that n3 n = n(n 1)(n + 1).
39. Since 3736 mod 24 = 16, it would be 6 p.m.
40. 5
41. Observe that the number with the decimal representation a9 a8 a1 a0 is a9109 + a8108 + + a110 + a0 From Exercise 9 and the fact that ai10i mod 9 = ai mod 9, we deduce that the check digit is (a9 + a8 + · · · + a1 + a0) mod 9. So, substituting 0 for 9 or vice versa for any ai does not change the value of (a9 + a8 + + a1 + a0) mod 9.
42. No.
43. For the case in which the check digit is not involved, the argument given Exercise 41 applies. Denote the money order number by a9a8 a1a0c where c is the check digit. For a transposition involving the check digit c = (a9 + a8 + + a0) mod 9 to go undetected, we must have a0 = (a9 + a8 + · · · + a1 + c) mod 9. Substituting for c yields 2(a9 + a8 + · · · + a0) mod 9 = a0. Then cancelling the a0, multiplying by sides by 5, and reducing module 9, we have 10(a9 + a8 + + a1) = a9 + a8 + + a1 = 0. It follows that c = a9 + a8 + a1 + a0 = a0 In this case the transposition does not yield an error.
44. Observe that for any integer k between 0 and 8, k ÷ 9 = .kkk . . . .
46. 7
47. Say that the weight for a is i. Then an error is undetected if modulo 11, ai + b(i 1) + c(i 2) = bi + c(i 1) + a(i 2). This reduces to the cases where (2a b c) mod 11 = 0.
{ | ≤ }
48. a a = 0; if a b is an integer k, then b a is the integer k; if a b is the integer n and b c is the integer m, then a c = (a b) + (b c) is the integer n + m. The set of equivalence classes is [k] 0 k < 1, k is real The equivalence classes can be represented by the real numbers in the interval [0, 1). For any real number a, [a] = {a + k | where k ranges over all integers}.
49. No. (1, 0) ∈ R and (0, 1) ∈ R but (1, 1) /∈ R.
50. Obviously, a + a = 2a is even and a + b is even implies b + a is even. If a + b and b + c are even, then a + c = (a + b) + (b + c) 2b is also even. The equivalence classes are the set of even integers and the set of odd integers.
51. a belongs to the same subset as a. If a and b belong to the subset A and b and c belong to the subset B, then A = B, since the distinct subsets of P are disjoint. So, a and c belong to A.
52. Suppose that n is odd prime greater than 3 and n + 2 and n + 4 are also prime. Then n mod 3 = 1 or n mod 3 = 2. If n mod 3 = 1 then n + 2 mod 3 = 0 and so is not prime. If n mod 3 = 2 then n + 4 mod 3 = 0 and so is not prime.
CHAPTER 1
Introduction to Groups
1. Three rotations: 0◦ , 120◦ , 240◦ , and three reflections across lines from vertices to midpoints of opposite sides.
2. Let R = R120, R2 = R240, F be a reflection across a vertical axis, F ′ = RF , and F ′′ = R2F
3. a. V b. R270 c. R0 d. R0, R180, H, V, D, D′ e. none
4. Five rotations: 0◦ , 72◦ , 144◦ , 216◦ , 288◦ , and five reflections across lines from vertices to midpoints of opposite sides.
5. Dn has n rotations of the form k(360◦/n), where k = 0, . . . , n 1. In addition, Dn has n reflections. When n is odd, the axes of reflection are the lines from the vertices to the midpoints of the opposite sides. When n is even, half of the axes of reflection are obtained by joining opposite vertices; the other half, by joining midpoints of opposite sides.
6. A nonidentity rotation leaves only one point fixed – the center of rotation. A reflection leaves the axis of reflection fixed. A reflection followed by a different reflection would leave only one point fixed (the intersection of the two axes of reflection), so it must be a rotation.
7. A rotation followed by a rotation either fixes every point (and so is the identity) or fixes only the center of rotation. However, a reflection fixes a line.
8. In either case, the set of points fixed is some axis of reflection.
9. Observe that 1 1 = 1; 1( 1) = 1; ( 1)1 = 1; ( 1)( 1) = 1. These relationships also hold when 1 is replaced by a “rotation” and 1 is replaced by a “reflection.”
10. Reflection.
11. Thinking geometrically and observing that even powers of elements of a dihedral group do not change orientation, we note that each of a, b and c appears an even number of times in the expression. So, there is no change in orientation. Thus, the expression is a rotation. Alternatively, as in Exercise 9, we associate each of a, b and c with 1 if they are rotations and 1 if they are reflections and we observe that in the product a2 b4 ac5 a3 c the terms involving a represent six 1s or six 1s, the term b4 represents four 1s or four 1s, and the terms involving c represent six 1s or six 1s. Thus the product of all the 1s and 1s is 1. So the expression is a rotation.
12. n is even.
13. In D4 , HD = DV but H /= V
14. Dn is not commutative.
15. R0 , R180 , H, V
16. Rotations of 0◦ and 180◦; Rotations of 0◦ and 180◦ and reflections about the diagonals.
17. R0 , R180 , H, V
18. Let the distance from a point on one H to the corresponding point on an adjacent H be one unit. Then, a translations of any number of units to the right or left are symmetries; a reflection across the horizontal axis through the middle of the H’s is a symmetry; and a reflection across any vertical axis midway between two H’s or bisecting any H is a symmetry. All other symmetries are compositions of finitely many of those already described. The group is non-Abelian.
19. In each case the group is D6.
20. D28
21. First observe that X2 = R0. Since R0 and R180 are the only elements in D4 that are squares we have X2 = R180 Solving X2Y = R90 for Y gives Y = R270
22. Their only symmetry is the identity.
/ 360/n 360/n 360/n
23. The n rotations of Dn are R0, R360/n, R2 , . . . , Rn 1 . Suppose that n = 2k for some positive integer k Then Rk = R360k/2k = R180 Conversely, if k 360/n = R180 then 360k/n = 180 and therefore 2k = n
24. X2 = F has no solutions; the only solution to X3 = F is F .
25. Let p be any vertex on the n-gon and p ′ be the counterclockwise vertex adjacent to p. Let F be the reflection that passes through p and let F ′ be the reflection that passes through the midpoint point of the edge joining p and p ′ . Then, in the composition FF ′ , then reflection F fixes p and the reflection F ′ takes p to p ′ . So, FF ′ = R360/n.
26. Z4 , D5 , D4 , Z2
D4 , Z3 , D3 , D16
D7 , D4 , D5 , Z10
CHAPTER 2
Groups
1. c, d
2. c, d
3. none
4. a, c
5. 7; 13; n 1; 1 = 1 3+2 i = 3 + 2 i 3 2i 3 2 i 3+2i 13 13
6. a. 31 i b. 5 c. 1 2 3 d. 2 4 12 8 6
7. Let A = 2 0 Then A ∈ G1 and det A = 2 but det A2 = 0. So G1 is not closed under multiplication. Also A ∈ G2 but A 1 = 1/2 0 is not in G2 G3 is a group.
8. Say, x is the identity. Then, 0 x = 0. So, x = 0. But 0 1 /= 1.
9. If 5x = 3 multiply both sides by 4, we get 0 = 12. If 3x = 5 multiply both sides by 7, we get x = 15. Checking, we see that 3 15 = 5 mod 20.
11. One is Socks-Shoes-Boots.
12. The set does not contain the identity; closure fails.
10. 1, 3, 7, 9, 11, 13, 17, 19.1, 9, 11, and 19 are their own inverses; 3 and 7 are inverses of each other as are 11 and 13.
13. Under multiplication modulo 4, 2 does not have an inverse. Under multiplication modulo 5, 1, 2, 3, 4 is closed, 1 is the identity, 1 and 4 are their own inverses, and 2 and 3 are inverses of each other. Modulo multiplication is associative.
18. (ab)3 = ababab and (ab 2 c) 2 = ((ab 2 c) 1 )2 =
19. Observe that a5 = e implies that a 2 = a3 and b7 = e implies that b14 = e and therefore b 11 = b3 . Thus, a 2b 11 = a3b3 . Moreover, (a2b4) 2 = ((a2b4) 1)2 = (b 4a 2)2 = (b3a3)2 .
20. K = {R0, R180}; L = {R0, R180, H, V, D, D′}. 0 1
21. The set is closed because det (AB) = (det A)(det B). Matrix multiplication is associative. 1 0 is the identity. Since 0 1 a b 1 c d = d b c a its determinant is ad bc = 1.
22. 12 = (n 1)2 = 1.
23. Using closure and trial and error, we discover that 9 74 = 29 and 29 is not on the list.
24. All we need do is find an x with the property xab = bax. The solution is x = b.
25. For n 0, we use induction. The case that n = 0 is trivial. Then note that (ab)n+1 = (ab)nab = anbnab = an+1bn+1 For n < 0, note that e = (ab)0 = (ab)n (ab) n = (ab)n a n b n so that an bn = (ab)n . In a non-Abelian group (ab)n need not equal anbn
26. The “inverse” of putting on your socks and then putting on your shoes, is taking off your shoes then taking off your socks. Use D4 for the examples. (An appropriate name for the property (abc) 1 = c 1b 1a 1 is “Socks-Shoes-Boots Property.”)
27. Suppose that G is Abelian. Then by Exercise 26, (ab) 1 = b 1a 1 = a 1b 1 . If (ab) 1 = a 1b 1 then by Exercise 24 e = aba 1b 1 Multiplying both sides on the right by ba yields ba = ab.
28. By definition, a 1(a 1) 1 = e. Now multiply on the left by a.
29. The case where n = 0 is trivial. For n > 0, note that (a 1ba)n = (a 1ba)(a 1ba) (a 1ba) (n terms). So, cancelling the consecutive a and a 1 terms gives a 1bna For n < 0, note that e = (a 1 ba)n (a 1 ba) n = (a 1 ba)n (a 1 b n a) and solve for (a 1ba)n .
30. (a1a2 · · · an)(a 1a 1 · · · a 1a 1) = e
31. By closure we have {1, 3, 5, 9, 13, 15, 19, 23, 25, 27, 39, 45}.
32. f (x) = x for all x. See Theorem 0.8.
33. Suppose x appears in a row labeled with a twice. Say x = ab and x = ac. Then cancellation gives b = c. But we use distinct elements to label the columns.
34. Z105 ; Z40 , D20 , U (41)
35. Closure and associativity follow from the definition of multiplication; a = b = c = 0 gives the identity; we may find inverses by solving the equations a + a ′ = 0, b′ + ac ′ + b = 0, c ′ + c = 0 for a ′ , b′ , c ′ .
36. (ab)2 = a2b2 e abab = aabb e ba = ab (ab) 2 = b 2a 2 e b 1a 1b 1a 1 = b 1b 1a 1a 1 e a 1b 1 = b 1a 1 e ba = ab.
37. Since e is one solution, it suffices to show that nonidentity solutions come in distinct pairs. To this end, note that if xn = e and x = e, then (x 1)n = e and x = x 1 . So if we can find one nonidentity solution we can find a second one. Now suppose that a and a 1 are nonidentity elements that satisfy xn = e and b is a nonidentity element such that b = a and b = a 1 and bn = e. Then, as before, (b 1)n = e and b = b 1 . Moreover, b 1 = a and b 1 = a 1 Thus, finding a third nonidentity solution gives a fourth one. Continuing in this fashion, we see that we always have an even number of nonidentity solutions to the equation xn = e
38. Note that ( 1 , 1 ) = ( 2 , 1 ), but ( 1 , 1 ) corresponds to 2 whereas ( 2 , 1 ) corresponds to 3 2 3 4 3 2 3 5 4 3 7 . So, the correspondence is not a function from Q+ × Q+ to Q+ .
39. For D6 the subgroups of order 6 have the form {R0, R60, R120, R180, R240, R300} and R0, R120, R240, F, R120F, R240F where F is a reflection. For the general case where n > 3 is even, we have the set of n rotations is a subgroup of order n. For a second subgroup of order n, let R = R360/n and F be any reflection in Dn. Then H = R0, R2 , , R n 2 , F, R2F, , R n 2 F is a subgroup of D n of order n (Closure follows from the relation FR i = R iF .) Since H contains only half of the reflections in Dn, we can obtain a third subgroup of order n by replacing F in H by any reflection F ′ not in H. To see that these three subgroups are the only ones, note that any other subgroup K in D n (n 4) must have the same n/2 rotations and n/2 reflections. But every reflection was used in the two subgroups described above, so K would be contained in one of the two. So, the number of subgroups of order n 4 when n is even is 3.
40. Observe that F1F2 = F2F1 implies that (F1F2)(F1F2) = R0 Since F1 and F2 are distinct and F1F2 is a rotation it must be R180. Alternate proof. Observe that (F1 F2 ) 1 = F2 1F1 1 = F2 F1 = F1 F2 implies that (F1F2) is its own inverse. Since F1 and F2 are distinct and F1F2 is a rotation it must be R180 .
41. Since FR k is a reflection we have (FRk)(FRk) = R0 Multiplying on the left by F gives RkFRk = F .
42. Since FR k is a reflection, we have (FRk)(FRk) = R0. Multiplying on the right by R k gives FR k F = R k If Dn were Abelian, then FR360◦ /n F = R360◦ /n But (R360 ◦ /n ) 1 = R360 ◦(n 1)/n /= R360 ◦ /n when n ≥ 3.
43. Using Exercise 42 we obtain the solutions R and R 1 F .
44. Rβ α; Rα β
45. Since a2 = b2 = (ab)2 = e, we have aabb = abab Now cancel on left and right.
46. If a satisfies x5 = e and a = e, then so does a2 , a3 , a4 Now, using cancellation we have that a2 , a3 , a4 are not the identity and are distinct from each other and distinct from a. If these are all of the nonidentity solutions of x5 = e, we are done. If b is another solution that is not a power of a, then by the same argument b, b2 , b3 and b4 are four distinct nonidentity solutions. We must further show that b2 , b3 and b4 are distinct from a, a2 , a3 , a4 If b2 = ai for some i, then cubing both sides we have b = b6 = a3i , which is a contradiction. A similar argument applies to b3 and b4 . Continuing in this fashion, we have that the number of nonidentity solutions to x5 = e is a multiple of 4. In the general case, the number of solutions is a multiple of 4 or is infinite.
47. The matrix a b c d is in GL(2, Z2) if and only if ad /= bc. This happens when a and d are 1 and at least 1 of b and c is 0 and when b and c are 1 and at least 1 of a and d is 0. So, the elements are 1 0 1 1 1
1 1 and 1 0 do not commute.
48. If n is not prime, we can write n = ab, where 1 < a < n and 1 < b < n. Then, a and b belong to the set 1, 2, . . . , n 1 , but 0 = ab mod n does not. If n is prime, let c be any element in the set. Then by the Corollary of Theorem 0.2 there are integers s and t such that cs + nt = 1. So, mod n we have cs = 1.
49. Proceed as follows. By definition of the identity, we may complete the first row and column. Then complete row 3 and column 5 by using Exercise 33. In row 2 only c and d remain to be used. We cannot use d in position 3 in row 2 because there would then be two d’s in column 3. This observation allows us to complete row 2. Then rows 3 and 4 may be completed by inserting the unused two elements. Finally, we complete the bottom row by inserting the unused column elements.
In D4 , R0 = 1; R90 = R270 = 4; R180 = H = V = D = D′ = 2.
In each case, notice that the order of the element divides the order of the group.
2. In Q, ⟨1/2⟩ = {n(1/2)| n ∈ Z} = {0, ±1/2, ±1, ±3/2, } In Q∗ , ⟨1/2⟩ = {(1/2)n| n ∈ Z} = {1, 1/2, 1/4, 1/8, . . . ; 2, 4, 8, . . .}.
3. In Q, 0 = 1. All other elements have infinite order since x + x + + x = 0 only when x = 0.
4. Observe that an = e if and only if (an) 1 = e 1 = e and (an) 1 = (a 1)n The infinite case follows from the infinite case. Alternate solution. Suppose |a| = n and |a 1| = k. Then (a 1)n = (an) 1 = e 1 = e So k ≤ n Now reverse the roles of a and a 1 to obtain n ≤ k. The infinite case follows from the finite case.
5. By the corollary of Theorem 0.2 there are integers s and t so that 1 = ms + nt. Then a1 = ams + nt = ams ant = (am )s (an )t = (at)n
6. In Z, the set of positive integers. In Q, the set of numbers greater than 1.
7. In Z30, 2 + 28 = 0 and 8 + 22 = 0. So, 2 and 28 are inverses of each other and 8 and 22 are inverses of each other. In U (15), 2 8 = 1 and 7 13 = 1. So, 2 and 8 are inverses of each other and 7 and 13 are inverses of each other.
8. a. |6| = 2, |2| = 6, |8| = 3; b. |3| = 4, |8| = 5, |11| = 12 ; c. |5| = 12, |4| =
, |
| = 4. In each case |a + b| divides lcm(|a|, |b|).
9. (a4c 2b4) 1 = b 4c2a 4 = b3c2a2 .
10. aba2 = a(ba)a = a(a2b)a = a3(ba) = a5b.
11. For F any reflection in D6 , {R0 , R120 , R240 , F, R120 F, R240 F }.
12. Let g ∈ G, g /= e If |g| = pm, then |gm | = p.
13. If a subgroup of D4 contains R270 and a reflection F , then it also contains the six other elements R0, (R270)2 = R180, (R270)3 = R90, R270F, R180F and R90F If a subgroup of D4 contains H and D, then it also contains HD = R90 and DH = R270. But this implies that the subgroup contains every element of D4 If it contains H and V , then it contains HV = R180 and R0.
15. If n is a positive integer, the real solutions of xn = 1 are 1 when n is odd and ±1 when n is even. So, the only elements of finite order in R∗ are ±1.
16. 1 or 2.
17. By Exercise 29 of Chapter 2 we have e = (xax 1)n = xanx 1 if and only if an = e.
18. By Exercise 17, for every x in G |xax 1| = |a|, so that xax 1 = a or xa = ax.
19. Suppose G = H ∪ K. Pick h ∈ H with h /∈ K. Pick k ∈ K with k /∈ H. Then, hk ∈ G but hk /∈ H and hk /∈ K U (8) = {1, 3} ∪ {1, 5} ∪ {1, 7}
20. By the corollary of Theorem 0.2 we can write 1 = 2s + nt. Then a1 = a2s+nt = (a2)s(an)t = (b2)(bn)t = b2s+nt = b1
To prove that Uk(n) is a subgroup it suffices to show that it is closed. Suppose that a and b belong to Uk(n). We must show that in U (n), ab mod k = 1. That is, (ab mod n) mod k = 1. Let n = kt and ab = qn + r where 0 r < n Then (ab mod n) mod k = r mod k = (ab qn) mod k = (ab qkt) mod k = ab mod k = (a mod k)(b mod k) = 1 1 = 1. H is not a subgroup because 7 H but 7 7 = 9 is not 1 mod 3.
22. The possibilities are 1, 2, 3, and 6. 5 is not possible, for if a5 = e, then e = a6 = aa5 = a 4 is not possible for if a4 = e, then e = a6 = a2a4 = a2
23. Suppose that m < n and am = an Then e = ana m = an m This contradicts the assumption that a has infinite order.
24. Observe that (ab)n = e if and only if a 1(ab)na = a 1ea = e and that a 1 (ab)n a = (ba)n
Alternative solution. From Exercise 17, we have |ab| = |a 1(ab)a| = |ba|. For the second part, we have |abab| = |(aba)b| = |b(aba) = |baba| For the last part, observe that in D4, |R90HR90| = 2, whereas |HR90H| = 4.
25. det A = ±1
26. k = 4n 1
⟩ { } { }
⟩ { }
27. 3 = 3, 32 , 33 , 34 , 35 , 36 = 3, 9, 13, 11, 5, 1 = U (14) 5 = 5, 52 , 53 , 54 , 55 , 56 = 5, 11, 13, 9, 3, 1 = U (14). 11 = 11, 9, 1 = U (14). Since U (20) = 8, for U (20) = k for some k it must be the case that k = 8. But 11 = 1, 34 = 1, 74 = 1, 92 = 1, 112 = 1, 134 = 1, 174 = 1, and 192 = 1. So, the maximum order of any element is 4.
28. Let A be the subset of even members of Zn and B the subset of odd members of Zn. If x ∈ B, then x + A = {x + a|a ∈ A} ⊆ B, so |A| ≤ |B| Also, x + B = {x + b|b ∈ B} ⊆ A, so |B| ≤ |A|
29. By Exercise 30, either every element of H is even or exactly half are even. Since H has odd order the latter cannot occur.
30. Suppose that K is a subgroup of Dn that has at least one reflection F Denote the rotations of K by R1, R2, . . . , Rm. Then R1F, R2F, . . . , R m F are distinct reflections in K. If F ′ is any reflection in K, then F ′F = R i for some i. But then F ′ = R iF . Thus, K has exactly m reflections.
31. By Exercise 30, either every element of H is a rotation or exactly half are rotations. Since H has odd order the latter cannot occur.
32. Suppose that a and b are two elements of order 2 that commute. Then, e, a, b, ab is closed and therefore a subgroup.
33. Observe that by Exercise 32 we have that for any reflection F in Dn the set {R0, R180, F, R180F } is a subgroup of order 4.
34. ⟨2⟩
35. First observe that because 6 = 30 + 30 54 belongs to H, we know 6 is a subgroup of H. Let n be the smallest positive integer in H. Then the possibilities for n are 1, 2, 3, 4, 5, and 6. Because ⟨6⟩, ⟨3⟩ and ⟨2⟩ contain 12, 30 and 54, these cannot be excluded. We can exclude 1 because ⟨1⟩ = Z. The same is true for 5 because 6 5 = 1. Finally, if 4 is in H, then so is 6 4 = 2. So, our list is complete.
36. By the corollary to Theorem 0.2, H = Z.
37. Suppose that H is a subgroup of D3 of order 4. Since D3 has only two elements of order 2, H must contain R120 or R240. By closure, it follows that H must contain R0, R120, and R240 as well as some reflection F . But then H must also contain the reflection R120F
38. H ∩ K /= ∅, since e ∈ H ∩ K Now suppose that x, y ∈ H ∩ K Then, since H and K are subgroups, we know xy 1 ∈ H and xy 1 ∈ K. That is, xy 1 ∈ H ∩ K.
39. For D6 the subgroups of order 6 have the form {R0, R60, R120, R180, R240, R300} and R0, R120, R240, F, R120F, R240F where F is a reflection. For the general case where n > 3 is even, we have the set of n rotations is a subgroup of order n. For a second subgroup of order n, let R = R360/n and F be any reflection in Dn. Then H = R0, R2 , , R n 2 , R0, F, R2F, , R n 2 F is a subgroup of D n of order n (Closure follows from the relation FR i = R iF .) Since H contains only half of the reflections in Dn, we can obtain a third subgroup of order n by replacing F in H by any reflection F ′ not in H. To see that these three subgroups are the only ones, note that any other subgroup K in D n (n 4) must have the same n/2 rotations and n/2 reflections. But every reflection was use of in the two subgroups described above, so K would be contained in one of the two. So, the number of subgroups of order n > 3 is even is 3.
40. {R0, R180}
41. If x ∈ Z(G), then x ∈ C(a) for all a, so x ∈ a∈G C(a). If x ∈ a∈G C(a), then xa = ax for all a in G, so x ∈ Z(G).
42. Suppose x C(a). Then, xa = ax. So a 1(xa) = a 1(ax) = x. Thus, (a 1x)a = x, and therefore a 1x = xa 1 This shows x C(a 1). The other half follows by symmetry.
43. We proceed by induction. The case that k = 0 is trivial. Let x C(a). If k is positive, then by induction on k, xak+1 = xaak = axak = aakx = ak+1x. Since x ∈ C(a) implies that that x commutes with ak, we have ak ∈ C(x). But then a k = (ak) 1 ∈ C(x). The statement “If for some integer k, x commutes ak , then x commutes with a” is false as can be seen in the group D4 with x = H, a = R90 and k = 2.
44. If a2 = b2 and a3 = b3 , then a2a = b2b. Now cancel a2 and b2 .
45. Since S is nonempty, so is ⟨S⟩ If a = am 1 am 2 · · · anr and b = bn1 bn2 · · · bnt , belong to ⟨S⟩, then ab 1 = am 1 am 2 · · · am r b nt · · · b n2 b n1 belongs to ⟨S⟩ So, ⟨S⟩ is a subgroup of G that contains S. If K is any subgroup of G that contains S, then K also contains S by closure. Thus, H contains S . Moreover, because S is a subgroup of G that contains S, it is one of the terms in the intersection. So, H is contained in ⟨S⟩.
46. a. ⟨2⟩ b. ⟨1⟩ c. ⟨3⟩ d. ⟨gcd(m, n)⟩ e. ⟨3⟩
47. Since ea = ae, C(a) = . Suppose that x and y are in C(a). Then xa = ax and ya = ay. Thus, (xy)a = x(ya) = x(ay) = (xa)y = (ax)y = a(xy) and therefore xy C(a). Starting with xa = ax, we multiply both sides by x 1 on the right and left to obtain x 1xax 1 = x 1axx 1 and so ax 1 = x 1a. This proves that x 1 ∈ C(a). By the Two-Step Subgroup Test, C(a) is a subgroup of G.
48. Mimic the proof of Theorem 3.5.
49. No. In D4, C(R180) = D4. Yes. Elements in the center commute with all elements.
50. That C(a) ⊆ C(a3) is easy. To prove the other inclusion, observe that a6 = a so if x ∈ C(a ), then xa = xa6 = x(a3a3
For the second part of the exercise try D6.
51. Let H = x G xn = e . Since e1 = e, H = . Now let a, b H. Then an = e and bn = e So, (ab)n = anbn = ee = e and therefore ab H Starting with an = e and taking the inverse of both sides, we get (an) 1 = e 1 . This simplifies to (a 1)n = e. Thus, a 1 ∈ H By the Two-Step test, H is a subgroup of G In D4, {x| x2 = e} = {R0, R180, H, V, D, D′} This set is not closed because HD = R90
52. For any integer n 3, observe that the rotation R360/n in Dn has order n. Now in Dn let F be any reflection. Then F ′ = R360/nF is a reflection in Dn Also |F ′| = |F | = 2 and F ′F = R360 /n has order n.
53. Induction shows that for any positive integer n we have 1 1 n
So, when the entries are from R, 1 1 has infinite order. When the entries are 0 1 from Zp, the order is p.
54. |A| = 2, |B| = 2, |AB| = ∞.
55. First observe that (ad)n/d = an = e, so |ad| is at most n/d. Moreover, there is no positive integer t < n/d such that (ad)t = adt = e, for otherwise |a| /= n
56. In Exercise 54, let a = A 1 , b = AB Then, ab = A 1AB = B has finite order.
57. Let G be a group of even order. Observe that for each element x of order greater than 2, x and x 1 are distinct elements of the same order. So, because elements of order greater than 2 come in pairs, there is an even number of elements of order greater than 2 (possibly 0). This means that the number of elements of order 1 or 2 is even. Since the identity is the unique element of order 1, it follows that the number of order 2 is odd.
58. For the first part use induction; 6, ∞.
59. For√any positive integer n, a rotation of√3 60◦/n has order n. If we let R be a rotation of 2√d egrees then Rn i√ s a rotation of 2n degrees. This is never a multiple of 360◦ , for if 2n = 360k then 2 = 360k/n, which is rational. So, R has infinite order.
60. Let a = 2 and b = 2 1 .
61. Inscribe a regular n-gon in a circle. Then every element of Dn is a symmetry of the circle.
62. Let F and F √ ′ be reflections that intersect in an angle of √ 2 degrees. Then FF ′ is a rotation of 2 2 degrees, which has infinite order.
63. Let g = m and write m = nq + r where 0 r < n. Then gr = gm nq = gm (gn) q belongs to H So, r = 0.
64. a. 2, 2, 4 b. 4, 6, 24 c. 2, 4, 8 d. 2, 4, 8.
65. 1 H, so H = . Let a, b H. Then (ab 1)2 = a2(b2) 1, which is the product of two rationals. The integer 2 can be replaced by any positive integer.
66. {1, 9, 11, 19}
67. Let a = n and write m = nq + r where 0 r < n. Then e = am = anq+r = (an)qar = ar . But that forces r = 0.
68. Since |e| = 1, H is nonempty. Assume a, b ∈ H and Let |a| = m and |b| = n Then, (ab)mn = (am)n(bn)m = enem = ee = e. So, |ab| divides mn. Since mn is odd, so is |ab|.
69. In Z6, H = {0, 1, 3, 5} is not closed.
70. 1 ∈ H Let a, b ∈ H and let m and n be integers such that am = ±1 and bm = ±1. Then, (ab)mn = (am )n (bn ) m = (±1)n (±1) m = ±1.
71. a. Let xh1 x 1 and xh2 x 1 belong to xHx 1 . Then (xh1x 1)(xh2x 1) 1 = xh1h2 1 x 1 ∈ xHx 1 also.
72. The identity is in H. Let A, B H. Then, det A = m and det B = n are positive rationals. Thus, det (AB 1) = (det A)(det B) 1 = m/n, which is a positive rational.
73. Let a/b and c/d belong to the set. By observation, ac/bd and b/a have odd numerators and denominators. If ac/bd reduces to lowest terms to x/y, then x divides ac and y divides bd. So they are odd.
74. The identity has determinant 20 = 1. Let det A = 2m and det B = 2n . Then, det (AB) = 2m+n and det A 1 = 2 m .
75. If 2a and 2b ∈ K, then 2a(2b) 1 = 2a b ∈ K, since a b ∈ H.
76. The function i(x) = 1 where x = 0 is in H. Let f, g H. Then, (f g)(2) = f (2) g(2) = 1 1 = 1. Also, 1 = i(2) = (f f 1 )(2) = f (2) f 1 (2) = f 1 (2) so f 1 (2) = 1 and f 1(x) is in H. The 2 can be replaced by any nonzero number.
77. 2 0 1 1 0 0 2 = 2 1 is not in H.
78. Dn when n is odd; Dn 1 when n is even.
79. If a + bi and c + di ∈ H, then (a + bi)(c + di) 1 = a + bi c di = ( ac + bd )+( bc ad ) i = (ac + bd) + (bc ad)i. Moreover, c + di c di c2+d2 (ac + bd)2 + (bc ad)2 = a2c2 + 2acbd + b2d2 + b2c2 2bcad + a2d2 . Simplifying we obtain (a2 + b2)c2 + (a2 + b2)d2 = (a2 + b2)(c2 + d2) = 1 1 = 1. So, H is a subgroup. H is the unit circle in the complex plane.
80. a ∈ ⟨a, b⟩ Let aibj and as bt ∈ ⟨a, b⟩ Then, (aibj)(as bt) 1 = ai s bj t ∈ ⟨a, b⟩ |⟨a, b⟩| ≤ |a|b|
81. {1, 2n 1, 2n + 1, 4n 1}. This group is not cyclic.
82. Let F1 and F2 be distinct reflections in D3 and H = {R0, F1} and K = {R0, F2}. Then, HK = {R0, F2, F1, F1F2} but F2F1 is not in HK
83. In D10 let a be any reflection and b = R36.
84. First, observe that pn 1 is not divisible by p. For, if so, then pn (pn 1) = 1 would be divisible by p So, pn 1 U (pn). Moreover, (pn 1)2 = p2n 2pn + 1 = 1 mod pn, so pn 1 has order 2. Now suppose that x U (pn) and x2 = 1 mod pn . We will prove that x = 1 or x = pn 1. From x2 = 1 mod pn we have that x2 1 = (x 1)(x + 1) is divisible by pn . We claim that pn divides x 1 or pn divides x + 1. If this is not the case, then p divides both x 1 and x + 1, and therefore p divides (x + 1) (x 1) = 2, which is false. Since 1 x pn 1 we know that pn divides x 1 only when x = 1. Thus, if x = 1, we have x + 1 pn and pn divides x + 1. So, x + 1 = pn
85. First observe that 2n 1 and 2n 2 ± 1 are in U (2n) and satisfy x2 = 1. Now suppose that x ∈ U (2n), x /= 1, and x2 = 1 mod 2n . From x2 = 1 mod 2n we have that x2 1 = (x 1)(x + 1) is divisible by 2n . Since x 1 and x + 1 are even and n ≥ 3, we know that at least one of x 1 and x + 1 is divisible by 4. Moreover, it cannot be the case that both x 1 and x + 1 are divisible by 4 for then so would (x + 1) (x 1) = 2. If x 1 is not divisible by 4, then x + 1 is divisible by 2n 1 Thus x + 1 = k2n 1 for some integer k and k2n 1 = x + 1 ≤ 2n . So, k = 1 or k = 2. For k = 1, we have x = 2n 1 1. For k = 2, we have x = 2n 1. If x + 1 is not divisible by 4, then x 1 is divisible by 2n 1 Thus x 1 = k2n 1 for some integer k and k2n 1 = x 1 < 2n . So, k = 1 and x = 2n 1 + 1.
86. a. U (5) or in C∗ the subgroup {1, 1, i, i}; R∗
b. GF (2, Z3 ); GF (2, Q)
c. U (8) or U (12)
d. Z6
87. Since ee = e is in HZ(G) it is non-empty. Let h1z1 and h2z2 belong to HZ(G). Then h1z
)
HZ(G).
88. Observe that if a/b ∈ H and c/d ∈ K where a, b, c, d are nonzero intergers, then a = b(a/b) ∈ H, c = d(c/d) ∈ K and ac ∈ H ∩ K.
89. First note that if m/n = 0 is an element of H, then n(m/n) = m and m are also in H. By the Well Ordering Principle, H has a least positive integer t. Since t is not in K = 2h h H , K is a nontrivial proper subgroup of H (see Example 5). Alternatively, one can use Exercise 88.
/ { | ∈ } | | / /
90. Suppose that G is a group of order n > 2 and H is a subgroup of G with H = n 1. Let a be in G but a not in H and let b be in H and b = e. Then, ab = a and ab is not in H
CHAPTER 4
Cyclic Groups
1. For Z6, generators are 1 and 5; for Z8 generators are 1, 3, 5, and 7; for Z20 generators are 1, 3, 7, 9, 11, 13, 17, and 19.
2. For a , generators are a and a5 ; for b , generators are b, b3 , b5 , and b7 ; for c , generators are c, c3 , c7 , c9 , c11 , c13 , c17 , c19 .
11. By definition, a 1 ∈ ⟨a⟩ So, ⟨a 1⟩ ⊆ ⟨a⟩ By definition, a = (a 1) 1 ∈ ⟨a 1⟩ So, ⟨a⟩ ⊆ ⟨a 1⟩.
12. ⟨3⟩, ⟨ 3⟩; a3 , a 3
⟩ { ± ± }
⟩ { ± ± }
13. Observe that 10 = 0, 10, 20, . . . and 12 = 0, 12, 24, . . . . Since the intersection of two subgroups is a subgroup, according to the proof of Theorem 4.3, we can find a generator of the intersection by taking the smallest positive integer that is in the intersection. So, 10 12 = 60 For m and n we have m = 0, m, 2m, . . . and n = 0, n, 2n, ................... Then the smallest positive integer in the intersection is lcm(m, n).
⟩ { ± ± } ⟨ ⟩ { ± ± }
For the case ⟨am⟩ ∩ ⟨an⟩, let k = lcm(m, n). Write k = ms and k = nt. Then ak = (am )s ∈ ⟨am ⟩ and ak = (an)t ∈ ⟨an⟩. So, ⟨ak⟩ ⊆ ⟨am ⟩ ∩ ⟨an⟩. Now let ar be any element in ⟨am ⟩ ∩ ⟨an⟩. Then r is a multiple of both m and n. It follows that r is a multiple of k (see Exercise 12 of Chapter 0). So, ar ∈ ⟨ak⟩.
14. 49. First note that the group is not infinite since an infinite cyclic group has infinitely many subgroups. Let |G| = n. Then 7 and n/7 are both divisors of n. If n/7 /= 7, then G has at least 4 divisors. So, n/7 = 7. When 7 is replaced by p , |G| = p2 .
15. g divides 12 is equivalent to g12 = e So, if a12 = e and b12 = e, then (ab 1)12 = a12(b12) 1 = ee 1 = e. The same argument works when 12 is replaced by any integer (see Exercise 51 of Chapter 3).
16. a. |a| = |a2| if and only if |a| is odd or infinite. To see this note, that if |a| = ∞ , then |a2| cannot be finite, and if |a| = n, by Theorem 4.2, we have n = |a2| = n/gcd(n, 2) and therefore gcd(n, 2) = 1. b. |a2| = |a12| if and only if |a| = ∞ or |a| is finite and gcd(|a|, 2) = gcd(|a|, 12). c. Both i and j are 0 or both are not 0. d. i = ±j
17. By Theorem 4.2, we have |⟨a6⟩| = n/gcd(n, 6). Since n is odd and ⟨a6⟩ is a proper subgroup, we have gcd(n, 6) = 3. So, |⟨a6⟩| = n/3.
18. By Theorem 4.2, we have gcd(n, 10) = gcd(n, 5). If n is even and gcd(n, 5) = 5, then gcd(n, 10) = 10, which contradicts that gcd(n, 10) = gcd(n, 5).
19. If |a2| = 3, |a| is 3 or 6. If |a2| = 4, |a| = 8.
20. For Dpn there are pn cyclic subgroups of order 2. Since the rotations form a cyclic subgroup of order pn there is exactly one subgroup for each of the orders p0 , p1 , p2 , . . . , pn and no others. So, the total for Dpn is pn + n + 1. For Dpq there are pq cyclic subgroups of order 2. Since the rotations form a cyclic subgroup of order pq, there is exactly one cyclic subgroup for each of the orders pq, p, q and 1. So, the total for Dpq is pq + 4.
21. For every a and b we have ab = (ab) 1 = b 1a 1 = ba. Alternate solution. Let a and b belong to G Observe that aabb = a2b2 = ee = e = (ab)2 = abab. By cancellation we have ab = ba.
23. Let a = m, b = n, ab = k and gcd(m, n) = d. Then lcm(m, n) = mn/d and (ab)mn/d = (am )n/d (bn/d )m = ee = e so k divides lcm(m, n). So, if d > 1, then k < mn. If d = 1, then a b = e because a b divides both a and b . We also have e = (ab)k = akbk and therefore ak = b k a b = e . This means that both m and n and therefore mn are divisors of k
24. First suppose G is infinite. Let x ∈ G, x /= e. Then, G = ⟨x⟩ and ⟨x2⟩ contradicts the hypothesis. Next, assume G is finite and e /= x ∈ G Then, ⟨x⟩ = G (otherwise ⟨x⟩ is nontrivial and proper). By Theorem 4.3, G has a subgroup for each divisor of |G| and since the only subgroups of G have orders |G| and 1 we have that only divisors of |G| are |G| and 1. So, |G| is prime.
25. Exercise 31 in Chapter 3 tells us that H is a subgroup of the cyclic group of n rotations in Dn. So, by Theorem 4.3, H is cyclic.
26. Z3n; D3n These generalize to the p odd case.
27. 1 (the identity). To see this, note that we can let the group be a where a is infinite. If some element ai has finite order n then (ai)n = e. But then ain = e, which implies that a has finite order. This contradicts our assumption.
28. By Corollary 2 of Theorem 4.1, a nonidentity element of G must have order 5, 7 or 35. We may assume that G has no element of order 35. Since 34 is not a multiple of φ(5) = 4, not all of the nonidentity elements can have order 5. Similarly, not all of them can have order 7. So, G has elements of orders both 5 and 7. Say, a = 5 and b = 7. Then, since (ab)5 = b5 = e and (ab)7 = a7 = a2 = e, we must have ab = 35, a contradiction.
| | | | | |
29. a. a divides 12. b. a divides m. c. By Theorem 4.3, a = 1, 2, 3, 4, 6, 8, 12, or 24. If a = 2, then a8 = (a2)4 = e4 = e A similar argument eliminates all other possibilities except 24.
30. Let G = {e, a, b}. Cancellation shows ab must be e. Thus, G = {e, a, a 1} = ⟨a⟩.
31. Yes, by Theorem 4.3. The subgroups of Z are of the form ⟨n⟩ = {0, ±n, ±2n, ±3n, . . .}, for n = 0, 1, 2, 3, The subgroups of ⟨a⟩ are of the form ⟨an⟩ for n = 0, 1, 2, 3, . . .
32. Certainly, a ∈ C(a). Thus, ⟨a⟩ ⊆ C(a).
33. Dn has n reflections, each of which has order 2. Dn also has n rotations that form a cyclic group of order n So, according to Theorem 4.4, there are φ(d) rotations of order d in Dn. If n is odd, there are no rotations of order 2. If n is even, there is φ(2) = 1 rotation of order 2. (Namely, R180.) So, when n is odd, Dn has n elements of order 2; when n is even, Dn has n + 1 elements of order 2.
34. 1 and 1 are the only generators of Z. Suppose that ak generates ⟨a⟩. Then there is an integer t so that (ak )t = a By Theorem 4.1, we conclude that kt = 1. So, k = ±1.
35. See Example 16 of Chapter 2.
36. Let a and b to H. Then, a ∈ H i and b ∈ H j for some i and j and we may assume that i < j Then, ab 1 ∈ H j ⊂ H
37. 1000000, 3000000, 5000000, 7000000. By Theorem 4.3, 1000000 is the unique subgroup of order 8, and only those on the list are generators; a1000000 , a3000000 , a5000000 , a7000000 By Theorem 4.3, a1000000 is the unique subgroup of order 8, and only those on the list are generators.
38. Let H be the unique maximal subgroup of G and suppose that G is not cyclic. Let a be any element in G not in H Then H1 = ⟨a⟩ is not G Since H is the unique maximal subgroup of G, H1 cannot be maximal and is therefore properly contained some proper subgroup H2 of G. Because a ∈ H2, H2 cannot be maximal. Since G is finite, we can repeat this argument to reach a contradiction. Thus, G = ⟨a⟩. Now suppose that |a| = n and n is divisible by two distinct primes p and q. Then, ⟨am/p⟩ and ⟨am/q⟩ are distinct maximal subgroups of G, which is a contradiction.
39. Let G = {a1 , a2 , . . . , ak }. Now let |ai| = ni and n = n1n2 nk . Then an = e for all i since n is a multiple of ni.
40.
41. The lattice is a vertical line with successive terms from top to bottom ⟨p0⟩, ⟨p1⟩, ⟨p2⟩, . . . , ⟨pn 1⟩, ⟨0⟩.
42. First suppose that G is the union of proper subgroups. If G were cyclic, say G = ⟨a⟩, and G was the union of proper subgroups H1, H2, . . . , Hn, then a must be in one of Hi, since the union contains every element. But if a belongs to Hi, then G = ⟨a⟩ is a subgroup of the proper subgroup Hi. This is a contradiction. Now suppose that G is not cyclic. Then, for every g ∈ G, ⟨g⟩ is a proper subgroup and G = ∪g∈ G⟨g⟩
43. Suppose that Q+ is cyclic. Because a/b = b/a we may assume that a/b > 1. Let p be any prime that does not divide a. Then there is a positive integer such that (a/b)n = p Thus an = pbn But this contradicts Theorem 0.3. Alternate solution. Suppose that r is a generator of Q+ . Since r = r 1 , we may assume that r > 1. Then there are positive integers m and n such that rm = 2 and rn = 3. Then rmn = (rm )n = 2n and rmn = (rn)m = 3m . This implies that 2n = 3m . But 2n is even and 3m is odd. This proves the group of nonzero rationals under multiplication is not cyclic, for otherwise, its subgroups would be cyclic.
44. 4 8 12 16 4 8 12 16 16 12 8 4 12 4 16 8 8 16 4 12 4 8 12 16 The identity is 16. The group is generated by 8 and by 12.
45. For 7, use Z2 6 . For n, use Z2 n 1 .
46. |ab| could be any divisor of lcm(|a|, |b|).
47. Suppose that |ab| = n. Then (ab)n = e implies that bn = a n ∈ ⟨a⟩, which is finite. Thus, |b| is finite.
48. See Example 5 in Chapter 3 for showing G is a subgroup. For the general case, observe by Theorem 4.2 for any a ∈ G we have ⟨ak⟩ = ⟨agcd(n,k)⟩ so that Gk = Ggcd(n,k)
49. Since gcd(100, 98) = 2 and gcd(100, 70) = 10 we have |a98| = |a2| = 50 and |a70| = |a10| = 10.
50. Since FF ′ is a rotation other than the identity and the rotations of D21 form a cyclic subgroup of order 21, we know by Theorem 4.3 that FF ′ is a divisor of 21. Moreover, FF ′ cannot be the identity for then FF ′ = FF , which implies that F ′ = F . So, |FF ′| = 3, 7 or 21.
51. Because H is cyclic, we know that a6 divides 10. So, a60 = e. Thus a can be any divisor of 60.
52. Using the corollary to Theorem 4.4 we get 21600
53. The argument given in the proof of the corollary to Theorem 4.4 shows that in an infinite group, the number of elements of finite order n is a multiple of φ(n) or there is an infinite number of elements of order n.
54. By Corollary 1 of Theorem 4.2, a = m divides n. Say n = mq. Then, an = (am)q = em = e
55. It follows from Example 16 in Chapter 2 and Example 15 in Chapter 0 that the group H = cos(360◦/n) + i sin(360◦/n) is a cyclic group of order n and every member of this group satisfies xn 1 = 0 Moreover, since every element of order n satisfies xn 1 = 0 and there can be at most n such elements, all complex numbers of order n are in H Thus, by Theorem 4.4, C∗ has exactly φ(n) elements of order n
56. Clearly 0 is in H. If m and n are in H, then m has the forms 8m1 and 10m2 and n has the forms 8n1 and 10n2 Then, m n has the forms 8m1 8n1 and 10m1 10n1. So, m n is in H and H is a subgroup of Z. If the condition is changed to “divisible by 8 or 10,” H is not a subgroup since 8 and 10 would belong to H but 10 8 = 2 would not.
57. Let x ∈ Z(G) and |x| = p where p is prime. Say y ∈ G with |y| = q where q is prime. Then (xy)pq = e and therefore |xy| = 1, p or q. If |xy| = 1, then x = y 1 and therefore p = q. If |xy| = p, then e = (xy)p = yp and q divides p. Thus, q = p. A similar argument applies if |xy| = q.
58. If an infinite group had only a finite number of subgroups, then it would have only a finite number of cyclic subgroups. If each of these cyclic subgroups is finite then the group would be finite since every element is in the cyclic subgroup generated by itself. So the group contains at least one infinite cyclic subgroup, call it ⟨a⟩. Then the subgroups ⟨a⟩, ⟨a2 ⟩, ⟨a3 ⟩, . . . are distinct subgroups since the least positive power of a in ⟨ai⟩ is ai and ai is not the least positive power of a in any subgroup ⟨aj ⟩ for j /= i.
59. An infinite cyclic group does not have an element of prime order. A finite cyclic group can have only one subgroup for each divisor of its order. A subgroup of order p has exactly p 1 elements of order p. Another element of order p would give another subgroup of order p.
62. In a group, the number of elements order d is divisible by φ(d) or there are infinitely many elements of order d
63. D33 has 33 reflections, each of which has order 2 and 33 rotations that form a cyclic group. So, according to Theorem 4.4, for each divisor d of 33 there are φ(d) rotations of order d in Dn. This gives one element of order 1; φ(3) = 2 elements of order 3; φ(11) = 10 elements of order 11; and φ(33) = 20 elements of order 33.
64. Since U (25) = 20, by Corollary 1 of Theorem 4.2, we know that |2| must divide 20. So, |2| = 1, 2, 4, 5, 10, or 20. But 210 /= 1 implies that |2| /= 1, 2, 5 or 10 and 24 /= 1 implies that |2| /= 4.
65. Let a = 4 and b = 5. Since (ab)20 = (a4 )5 (b5 )4 = e e = e, we know that ab divides 20. Noting that (ab)4 = b4 = e we know that ab = 1, 2 or 4. Likewise, (ab)10 = a2 = e implies that ab = 5 or 10. So, ab = 20. Then, by Theorem 4.3, ab has subgroups of orders 1, 2, 4, 5, 10 and 20. In general, if an Abelian group contains cyclic subgroups of order m and n where m and n are relatively prime, then it contains subgroups of order d for each divisor d of mn.
66. 1, 2, 3, 12. In general, if an Abelian group contains cyclic subgroups of order m and n, then it contains subgroups of order d for each divisor d of the least common multiple of m and n.
67. Say a and b are distinct elements of order 2. If a and b commute, then ab is a third element of order 2. If a and b do not commute, then aba is a third element of order 2.
68. φ(42) = 12
69. By Exercise 38 of Chapter 3, ⟨a⟩ ∩ ⟨b⟩ is a subgroup. Also, ⟨a⟩ ∩ ⟨b⟩ ⊆ ⟨a⟩ and ⟨a⟩ ∩ ⟨b⟩ ⊆ ⟨b⟩ So, by Theorem 4.3, |⟨a⟩ ∩ ⟨b⟩| is a common divisor of 10 and 21. Thus, |⟨a⟩ ∩ ⟨b⟩| = 1 and therefore ⟨a⟩ ∩ ⟨b⟩ = {e}.
70. Mimic Exercise 69.
71. |⟨a⟩ ∩ ⟨b⟩| must divide both 24 and 10. So, ⟨a⟩ ∩ ⟨b⟩| = 1 or 2.
72. By Exercise 38 of Chapter 3, a b is a subgroup a and b . So, a b divides 12 and 22. It follows that a b = 1 or 2, and since a b = e , we have that a b = 2. Because a6 is the only subgroup of a of order 2 and b11 is the only subgroup of b of order 2, we have a b = a6 = b11 , and therefore a6 = b11 .
73. Suppose that G has 14 elements of order 3. Let a G and a = e. Let b G and b a Then, by cancellation, H = aibj i, j are 0, 1, 2 has exactly nine elements and is closed and therefore is a subgroup of G. Let c G and c H. Then, by cancellation, the nine expressions of the form aibjc where i, j are 0, 1, 2 are distinct and have no overlap with the nine elements of H. But that gives 18 elements in G.
74. First note that if k is a generator then so is k. Thus, it suffices to show that k /= k. But k = k implies that 2k = 0 so that n = |k| = 1 or 2.
75. Observe that |a5| = 12 implies that e = (a5)12 = a60 , so |a| divides 60. Since ⟨a5 ⟩ ⊆ ⟨a⟩, we know that |⟨a⟩| is divisible by 12. So, |⟨a⟩| = 12 or 60. If |a4 | = 12, then |a| divides 48. Since ⟨a4 ⟩ ⊆ ⟨a⟩, we know that |⟨a⟩| is divisible by 12. So, |⟨a⟩| = 12, 24, or 48. But |a| = 12 implies |a4| = 3 and |a| = 24 implies |a | = 6. So, |a| = 48.
76. By Theorem 4.3, it suffices to find necessary and sufficient conditions so that |xr| divides |xs| By Theorem 4.2, we obtain gcd(n, s) divides gcd(n, r).
10. (13)(1245)(13) = (3245); (24)(13456)(24) = (13256). In general, for any cycle α we have (ij)α(ij) is the same as α with i replaced by j. In both cases, the element of the 2-cycle that appears in all three cycles is replaced by the other element of the 2-cycle.
11. An n-cycle is even when n is odd, since we can write it as a product of n 1 2-cycles by successively pairing up the first element of the cycle with each of the other cycle elements starting from the last element of the cycle and working toward the front. The same process shows that when n is odd we get an even permutation.
12. If α1, α2, . . . , αn are 2-cycles and α = α1 αn, then α 1 = αnαn 1 α2α1.
13. To prove that α is 1 1, assume α(x1) = α(x2). Then x1 = α(α(x1)) = α(α(x2)) = x2 To prove that α is onto, note that for any s in S, we have α(α(s)) = s.
14. (n 3)! in Sn; (n 3)!/2 in An
15. Suppose that α can be written as a product on m 2-cycles and β can be written as product of n 2-cycles. Then juxtaposing these 2-cycles we can write αβ as a product of m + n 2-cycles. Now observe that m + n is even if and only if m and n are even or both odd.
16. (+1) (+1) = (+1) ( 1) ( 1) = +1 even · even = even odd · odd = even (+1) · ( 1) = ( 1) ( 1) · (+1) = ( 1) even · odd = odd odd · even = odd
17. n is odd
18. even; odd.
19. If α is the product of m 2-cycles and β is the product of n 2-cycles, then α 5βα3 is the product of 8m + n and 8m + n is odd if and only if n is odd.
20. Say β can be written with m 2-cycles and α with n 2-cycles. Then, β 1αβ can be written with 2m + n 2-cycles.
21. even
22. 10; 12
23. We find the orders by looking at the possible products of disjoint cycle structures arranged by longest lengths left to right and denote an n-cycle by (n). (6) has order 6 and is odd; (5)(1) has order 5 and is even; (4)(2) has order 4 and is even; (4)(1)(1) has order 4 and is odd; (3)(3) has order 3 and is even; (3)(2)(1) has order 6 and is odd; (3)(1)(1)(1) has order 3 and is even; (2)(2)(2) has order 2 and is odd; (2)(2)(1)(1) has order 2 and is even; (2)(1)(1)(1)(1) has order 2 and is odd. So, for S6 , the possible orders are 1, 2, 3, 4, 5, 6; for A6 the possible orders are 1, 2, 3, 4, 5. We see from the cycle structure of S7 shown in Example 4 that in A7 the possible orders are 1, 2, 3, 4, 5, 6, 7.
27. If all members of H are even we are done. So, suppose that H has at least one odd permutation σ For each odd permutation β in H observe that σβ is even and, by cancellation, different βs give different σβs. Thus, there are at least as many even permutations as there are odd ones. Conversely, for each even permutation β in H observe that σβ is odd and, by cancellation, different βs give different σβs. Thus, there are at least as many odd permutations as there are even ones.
28. By Exercise 27, either every element of H is even or half are even and half are odd. In the latter case, H would have even order.
29. The identity is even; the set is not closed.
30. If α can be written as the product of m 2-cycles and β can be written as a product n 2-cycles, then α 1β 1αβ can be written as the product of n + m + n + m = 2(m + n) 2-cycles.
33. In A6 , elements of order 2 in disjoint cycle form must be the product of two 2-cycles. So the number of elements of order 2 is 6 5 4 3/(2 2 2).
34. Suppose that α Z(A n ) and α = ǫ. Pick i 1, 2, . . . , n such that α (i) = j = i and let α (j) = k. Let t 1, 2, . . . , n and t = i, j, or k. If k = i, then (jti)α = α (jti). But the left side sends i to t, whereas the right side sends i to i, which is a contradiction. If k = i, then the left side of (jti)α = α(jti) sends i to t, whereas the right side sends i to k, which is a contradiction.
{ } / | | | |
35. Since x5 = 5, we know x = 25. One solution is (1, 6, 7, 8, 9, 2, 10, 11, 12, 13, 3, 14, 15, 16, 17, 4, 18, 19, 20, 21, 5, 22, 23, 24, 25). The number of solutions is 20!
/ /
36. Since αm = (1, 3, 5, 7, 9)m (2, 4, 6)m (8, 10)m and the result is a 5-cycle, we deduce that (2, 4, 6)m = ǫ and (8, 10)m = ǫ. So, 3 and 2 divide m. Since (1, 3, 5, 7, 9)m = ǫ we know that 5 does not divide m. Thus, we can say that m is a multiple of 6 but not a multiple of 30.
37. An odd permutation of order 4 must be of the form (a1a2a3a4). There are 6 choices for a1, 5 for a2, 4 for a3, and 3 for a4. This gives 6 · 5 · 4 · 3 choices. But since for each of these choices the cycles (a1a2a3a4) = (a2a3a4a1) = (a3a4a1a2) = (a
) give the same group element, we must divide 6 · 5 · 4 · 3 by 4 to obtain 90. An even permutation of order 4 must be of the form (a1a2a3a4)(a5a6). As before, there are 90 choices (a1a2a3a4). Since (a5a6) = (a6a5) there are 90 elements of the form (a1a2a3a4)(a5a6). This gives 180 elements of order 4 in S6.
A permutation in S6 of order 2 has three possible disjoint cycle forms: (a1a2), (a1a2)(a3a4) and (a1a2)(a3a4)(a5a6). For (a1a2) there are 6 · 5/2 =15 distinct elements; for (a1a2)(a3a4) there are 6 · 5 · 4 · 3 choices for the four entries but we must divide by 2 · 2 · 2 since (a1a2) = (a2a1), (a3a4) = (a4a3) and (a1 a2 )(a3 a4 ) = (a4 a3 )(a1 a2 ). This gives 45 distinct elements. For (a1 a2 )(a3 a4 )(a5 a6 ) there are 6! choices for the six entries but we must divide by 2 2 2 3! since each of the three 2-cycles can be written 2 ways and the three 2-cycles can be permuted 3! ways. This gives 15 elements. So, the total number of elements of order 2 is 75.
38. Any product of 3-cycles is even, whereas (1234) is odd. In general, no odd permutation can be written as a product of 3-cycles.
39. Since β28 = (β4 )7 = ǫ, we know that β divides 28. But β4 = ǫ so β = 1, 2, or 4. If β = 14, then β written in disjoint cycle form would need at least one 7-cycle and one 2-cycle. But that requires at least 9 symbols and we have only 7. Likewise, β = 28 requires at least one 7-cycle and one 4-cycle. So, β = 7. Thus, β = β8 = (β4 )2 = (2457136). In S9 , β = (2457136) or β = (2457136)(89).
40. Observe that β = (123)(145) = (14523) so that β99 = β4 = β 1 = (13254).
41. Since (a1a2a3a4)(a5a6) = 4, such an x would have order 8. But the elements in S10 of order 8 are 8-cycles or the disjoint product of an 8-cycle and a 2-cycle. In both cases the square of such an element is the product of two 4-cycles.
42. If α and β are disjoint 2-cycles, then αβ = lcm(2,2) = 2. If α and β have exactly one symbol in common we can write α = (ab) and β = (ac). Then αβ = (ab)(ac) = (acb) and |αβ| = 3.
43. Let α, β stab(a). Then (αβ)(a) = α(β(a)) = α(a) = a. Also, α(a) = a implies α 1(α(a)) = α 1(a) or a = α 1(a).
44. Let β, γ ∈ H. Then (βγ)(1) = β(γ(1)) = β(1) = 1; (βγ)(3) = β(γ(3)) = β(3) = 3. So, by Theorem 3.3, H is a subgroup. |H| = 6. The proof is valid for all n ≥ 3. In the general case, |H| = (n 2)!. When Sn is replaced by An, |H| = (n 2)!/2
45. ⟨(1234)⟩; {(1), (12), (34), (12)(34)}
46. αk has k n/k-cycles.
47. This follows directly from Corollary 3 of Theorem 4.2.
48. Let α = (12) and β = (13).
49. Let α = (123) and β = (145).
50. R0 = (1)(2)(3); R120 = (123); R240 = (132); the reflections are (12), (13), (23).
51. Observe that (12) and (123) belong to Sn for all n ≥ 3 and they do not commute. Observe that (123)(124) and (124)(123) belong to An for all n 4 and they do not commute.
52. For any permutation α in S7, α2 is even, whereas (1234) is odd. For the second part note that because x3 is a subgroup of x and has order 4, we know 4 divides x . So we see from Example 5 that the possible values for x are 4 and 12. If x = 4, then (1234) = x3 = x 1 and therefore x = (4321). If x = 12, then x = (4321)(567) or x = (4321))(576) since no other choices yield x3 = (1234).
53. An even number of 2-cycles followed by an even number of 2-cycles gives an even number of 2-cycles in all. So the Finite Subgroup Test is verified.
54. Theorem 5.2 shows that disjoint cycles commute. For the other half, assume that a = c. Since (ab) = (cd), we know b = d. If (ab)(cd) = (cd)(ab), then (ab)(cd) = (ab)(ad) = (adb) and (cd)(ab) = (ad)(ab) = (abd). This means that a maps to both b and d, which are distinct. But permutations are 1-1 mappings.
55. Observe that H = β Sn β( 1, 2 ) = 1, 2 . So if α, β H, then (αβ)( 1, 2 ) = α(β( 1, 2 ) = α( 1, 2 ) = 1, 2 So H is a subgroup. To find H , observe that for elements of H there are two choices for the image of 1, then no choice for the image of 2, and (n 2)! choices for the remaining n 2 images. So, |H| = 2(n 2)!
58. 216◦ rotation; reflection about the axis joining vertex 1 to the midpoint of the opposite side.
59. Labeling consecutive vertices of a regular 5-gon 1, 2, 3, 4, 5, the even permutation (14)(23)(5) is the reflection that fixes 5 and switches vertices 1 and 4 and 2 and 3. Multiplying the n rotations by this reflection yields all n reflections. There is no reflection in D7 since their disjoint cycle form is a 1-cycle and three 2-cycles, which is an odd permutation.
60. Any element from An is expressible as a product of an even number of 2-cycles. For each pair of 2-cycles there are two cases. One is that they share an element in common (ab)(ac), and the other is that they are disjoint (ab)(cd). Now observe that (ab)(ac) = (abc) and (ab)(cd) = (cbd)(acb).
61. Since (1234)2 is in Bn, it is non-empty. If α = α1α2 · · · α i and β = β1β2 · · · βj where i and j are even and all the α’s and β’s are 4-cycles, then αβ = α1α2 αiβ1β2 βj is the product of i + j 4-cycles and i + j is even. So, by the finite subgroup test, B n is a subgroup. To show that B n is a subgroup of An, note that 4-cycles are odd permutations and the product of any two odd permutations is even. So, for the product of any even number of 4-cycles the product of the first two 4-cycles is even, then the product of the next two 4-cycles is even, and so on. This proves that B n is a subgroup of An
62. In Exercise 43 let G be A5 . Then stab(1) is the subgroup of A5 consisting of the 24 even permutations of the set 2, 3, 4, 5 Similarly, stab(2), stab(3), stab(4), stab(5) are subgroups of order 24.
63. The product of an element from Z(A4) of order 2 and an element of A4 of order 3 would have order 6. But A4 has no element of order 6.
64. α1 α2 α3 is odd and ǫ is even. The product of an odd number of 2-cycles cannot ǫ.
65. Cycle decomposition shows any nonidentity element of A5 is a 5-cycle, a 3-cycle, or a product of a pair of disjoint 2-cycles. Then, observe there are (5 4 3 2 1)/5 = 24 group elements of the form (abcde), (5 4 3)/3 = 20 group elements of the form (abc) and (5 4 3 2)/8 = 15 group elements of the form (ab)(cd). In this last case we must divide by 8 because there are 8 ways to write the same group element (ab)(cd) = (ba)(cd) = (ab)(dc) = (ba)(dc) = (cd)(ab) = (cd)(ba) = (dc)(ab) = (dc)(ba).
66. One possibility for a cyclic subgroup is ⟨(1234)(5678)⟩ One possibility for a noncyclic subgroup is {(1), (12)(34), (56)(78), (12)(34)(56)(78)}.
67. If α has odd order n and α is an odd permutation, then ǫ = αn would be an odd permutation.
68. Using the notation in Table 5.1, α2 , α3 , and α4 have order 2; α5 , α6 , . . . , α12 have order 3. The orders of the elements divide the order of the group.
69. Since |α| is the least common multiple of the disjoint cycle lengths of α, let the distinct disjoint cycle lengths be α be n1 , n2 , , nk Because n1 , n2 , , nk are integers between 1 and n, each of them appears as a term in n!.
70. Suppose α /= ε and α ∈ Z(Sn). Write α in disjoint cycle form (a1a2 ) , where a1 /= a2. Let a3 be different from a1 and a2 and let β = (a1)(a2a3). Then, (αβ)(a1) = a2, while (βα)(a1) = a3.
71. That a ∗ σ(b) /= b ∗ σ(a) is done by examining all cases. To prove the general case, observe that σi(a) ∗ σi+1(b) /= σi(b) ∗ σi+1(a) can be written in the form
∗ / ∗ σi(a) σ(σi(b)) = σi(b) σ(σi(a)), which is the case already done. If a transposition were not detected, then
which implies
72. 5
73. By Theorem 5.4 it is enough to prove that every 2-cycle can be expressed as a product of elements of the form (1k). To this end observe that if a = 1, b = 1, then (ab) = (1a)(1b)(1a).
74. By case-by-case, analysis H is a subgroup for n = 1, 2, 3, and 4. For n 5, observe that (12)(34) and (12)(35) belong to H but their product does not.
75. ADVANCE WHEN READY
76. TAAKTPKSTOOPEDN
CHAPTER 6
Isomorphisms
1. Property 1: Because φ is 1-1 and onto, so is φ 1 . Moreover, φ 1(φ(a)φ(b)) = φ 1(φ(ab)) = ab = φ 1(φ(a))φ 1(φ(b)).
Property 2: φ(e) = e is in φ(K), so φ(K) is not empty. Also, φ(a), φ(b) ∈ φ(K) implies that φ(a)φ(b) 1 = φ(a)φ(b 1)φ(ab 1 ∈ φ(K).
2. An automorphism of a cyclic group must carry a generator to a generator. Thus 1 → 1 and 1 → 1 are the only two choices for the image of 1. So let α : n → n and β : n → n. Then Aut(Z) = {α, β}. The same is true for Aut(Z6).
3. √ φ is o n√ t o since any positive real number r is the image√of √ r. φ is one-to-one since a = b implies that a = b Finally, φ(xy) = √ xy = x √ y = φ(x)φ(y).
4. U (8) is not cyclic while U (10) is. Define φ from U (8) to U (12) by φ(1) = 1; φ(3) = 5; φ(5) = 7; φ(7) = 11. To see that φ is operation preserving we observe that φ(1a) = φ(a) = φ(a) 1 = φ(a)φ(1) for all a; φ(3 5) = φ(7) = 11 = 5 7 = φ(3)φ(5); φ(3 · 7) = φ(5) = 7 = 5 · 11 = φ(3)φ(7); φ(5 · 7) = φ(3) = 5 = 7 · 11 = φ(5)φ(7).
5. The mapping φ(x) = (3/2)x is an isomorphism from G onto H. Multiplication is not preserved. When G = m and H = n the mapping φ(x) = (n/m)x is an isomorphism from G onto H
6. The identity is an isomorphism from G onto G If β is an isomorphism from G onto H, then β 1 is an isomorphism from H onto G (see Theorem 6.3). If β is an isomorphism from G onto H, and α is an isomorphism from H onto K, then αβ is an isomorphism from G onto K. That αβ is one-to-one and onto is done in Theorem 0.8. If a, b G, then (αβ)(ab) = α(β(ab)) = α(β(a)β(b)) =
7. D12 has an element of order 12 and S4 none; D12 has and element of order 6 and S4 none; D12 has 2 elements of order 3 and S4 has 8; D12 has 13 elements of order 2 and S4 has 9.
8. Properties of real numbers assure that the mapping is one-to-one and onto R and that log10(ab) = log10(a)+ log10(b) is a property of logs.
9. Since Te(x) = ex = x for all x, Te is the identity. For the second part, observe that Tg ◦ (Tg ) 1 = Te = Tgg 1 = Tg ◦ Tg 1 and cancel.
10. φ(na) = nφ(a).
11. 3 a 2b
12. Suppose α is an automorphism of G. Then, α(ab) = (ab) 1 and α(ab) = α(a)α(b) = a 1b 1 So, a 1b 1 = (ab) 1 = b 1a 1 for all a and b in G Taking the inverse of both sides proves that G is Abelian. If G is Abelian, then for all a and b in G, we have (ab) 1 = (ba) 1 = a 1b 1 . Thus, α(ab) = α(a)α(b). That α is one-to-one and onto follows directly from the definitions.
13. For any x in the group, we have (φgφh)(x) = φg(φh(x)) = φg(hxh 1) = ghxh 1g 1 = (gh)x(gh) 1 = φgh(x).
14. In U (16), |7| = |9| = 2, so by Theorem 4.4, U (16) is not cyclic.
15. φR0 and φR90 disagree on H; φR0 and φH disagree on R90; φR0 and φD disagree on R90; φR90 and φH disagree on R90; φR90 and φD disagree on R90; φH and φD disagree on D.
16. Aut(Z2) ≈ Aut(Z1) ≈ Z1; Aut(Z6) ≈ Aut(Z4) ≈ Aut(Z3) ≈ U (6) ≈ Z2; Aut(Z10) ≈ Aut(Z5) ≈ Z4 (see Example 4 and Theorem 6.4); Aut(Z12) ≈ Aut(Z8) (see Exercise 4 and Theorem 6.4).
17. We must show that Aut(G) has an identity, Aut(G) is closed, the composition of automorphisms is associative, and the inverse of every element in Aut(G) is in Aut(G). Clearly, the identity function ǫ(x) = x is 1-1, onto, and operation preserving. For closure let α, β Aut(G). That αβ is 1-1 and onto follows from Theorem 0.8. For a, b G, we have (αβ)(ab) = α(β(ab)) = α(β(a)β(b)) = (α(β(a))(α(β(b)) = (αβ)(a)(αβ)(b). Associativity follows from properties of functions (see Theorem 0.8). Let α Aut(G). Theorem 0.8 shows that α 1 is 1-1 and onto. We must show that α 1 is operation preserving: α 1 (xy) = α 1 (x)α 1 (y) if and only if α(α 1 (xy)) = α(α 1 (x)α 1 (y)). That is, if and only if xy = α(α 1 (x))α(α 1 (y)) = xy. So α 1 is operation preserving.
To prove that Inn(G) is a group, we may use the subgroup test. Exercise 13 shows that Inn(G) is closed. From φe = φgg 1 = φgφg 1 we see that the inverse of φg is in Inn(G). That Inn(G) is a group follows from the equation φgφh = φgh
18. Let φ be an isomorphism from G to H For any β in Aut(G), define a mapping from Aut(G) to Aut(H) by Γ(β) = φβφ 1 . Then, Γ is 1-1 and operation preserving. (See Theorem 0.8 and Exercise 6). To see that Γ is onto, observe that for any γ in Aut(H), Γ(φ 1γφ) = γ
19. Note that for n > 1, (φa)n = (φa)n 1φa, so an induction argument gives (φa)n = (φn 1)φa = φan 1 φa. Thus (φan 1 φa)(x) =
φan 1 (axa 1) = an 1(axa 1)(an 1) 1 =
axa 1
a n+1) = anxa n = φan (x). To handle the case where n is negative, we note that φe = φa n a n = φa n φa n = φa n (φa ) n (because n is positive). Solving for φan we obtain φa n φa n = φa n = (φa )n
20. This follows from Exercise 19.
21. Since b = φ(a) = aφ(1) it follows that φ(1) = a 1b and therefore φ(x) = a 1bx. (Here a 1 is the multiplicative inverse of a mod n, which exists because a ∈ U (n).)
22. Note that φ must take R360 /n to an element of order n. Since the reflections have order 2, we know φ(R360 /n ) is a rotation. Thus, φ(H) = φ(⟨R360 /n ⟩) ⊆ K.
23. Note that both H and K are isomorphic to the group of all permutations of four symbols, which is isomorphic to S4. The same is true when 5 is replaced by n since both H and K are isomorphic to Sn 1
24. Observe that ⟨2⟩, ⟨3⟩, are distinct and each is isomorphic to Z
25. Recall when n is even, Z(Dn) = {R0, R180}. Since R180 and φ(R180) are not the identity and belong to Z(Dn) they must be equal.
26. Note that φ(e) = e = e4 so H is not empty. Let a and b belong to H Then,
φ(ab 1) = φ(a)φ(b 1) = a4(b 1)4 = (ab 1)4 . This proof works for any integer n. If φ(x) = x 1 and a ∈ H, then φ(a) = a 1 = a4 , so a5 = e. Thus |a| = 1 or 5.
27. Z60 contains cyclic subgroups of orders 12 and 20 and any cyclic group that has subgroups or orders 12 and 20 must be divisible by 12 and 20. So, 60 is the smallest order of any cyclic group that has subgroups isomorphic to Z12 and Z60 .
28. φ(5) = 5 mod 20 is the same as 5φ(1) = 5, 25, 45, 65, 85 in Z. But we also need φ(1) = k = 20. So, we need gcd(n, k) = 1. This gives us φ(x) = x; φ(x) = 9x; φ(x) = 13x; φ(x) = 17x.
29. See Example 16 of Chapter 2
30. Since 3 is an element of Z20 with order 4, the mapping φ(x) = 3x is in Aut(20) and applying it to 1 four times sends 1 to 1. So, for any k in Z20 , φ(k) = φ(k 1) = kφ(1) = k.
31. That α is one-to-one follows from the fact that r 1 exists module n Onto follows from Exercise 13 in Chapter 5. The operation preserving condition is Exercise 11 of Chapter 0.
32. The mapping 1 a 0 1 → a is an isomorphism to Z when a ∈ Z and to R when a ∈ R.
33. By Part 1 of Theorem 6.2, we have φ(an) = φ(a)n = γ(a)n = γ(an) thus φ and γ agree on all elements of ⟨a⟩.
34. Observe that φ(7) = 7φ(1) = 13 and since 7 is relatively prime to 50, 7 1 exists modulo 50. Thus, we have φ(1) = 7 1 · 13 = 43 · 13 = 9 and φ(x) = φ(x · 1) = xφ(1) = 9x
35. First observe that because 25 = 10 = 1 we have 2 = 10. So, by parts 4 and 2 of Theorem 6.1, the mapping that takes φ(x) = 2x is an isomorphism.
36. For all automorphisms φ of Q∗ we know that φ(1) = 1 and φ( 1) = 1. For any rational a/b = pm 1 pm 2 · · · pm s /qn1 qn2 · · · qnt we have φ(a/b) = φ(p1)m1 φ(p2)m2 · · · φ(ps)
37. Tg(x) = Tg(y) if and only if gx = gy or x = y. This shows that Tg is a one-to-one function. Let y ∈ G. Then Tg(g 1y) = y, so that Tg is onto.
38. U (20) has three elements of order 2, whereas U (24) has seven.
39. To prove that φ is 1-1, observe that φ(a + bi) = φ(c + di) implies that a bi = c di. From properties of complex numbers this gives that a = c and b = d Thus a + bi = c + di. To prove φ is onto, let a + bi be any complex number. Then
φ(a bi) = a + bi. To prove that φ preserves addition and multiplication, note that
φ((a + bi) + (c + di)) = φ((a + c) + (b + d)i) = (a + c) (b + d)i = (a bi) + (c di) = φ(a + bi) + φ(c + di). Also,
φ((a + bi)(c + di) = φ((ac bd) + (ad + bc)i) = (ac bd) (ad + bc)i and
φ(a + bi)φ(c + di) = (a bi)(c di) = (ac bd) (ad + bc)i
40. Map a + b √ 2 → a 2b This map preserves both addition and multiplication.
a
41. First observe that Z is a cyclic group generated by 1. By property 3 of Theorem 6.2, it suffices to show that Q is not cyclic under addition. By way of contradiction, suppose that Q = ⟨p/q⟩ But then p/2q is a rational number that is not in ⟨p/q⟩
42. S8 contains ⟨(12345)(678)⟩ which has order 15. Since |U (16)| = 8, by Cayley’s Theorem, S8 contains a subgroup isomorphic to U (16). The elements of D8 can be represented as permutations of the 8 vertices of a regular 8-gon.
43. The notation itself suggests that φ(a + bi) = a b is the appropriate isomorphism. To verify this, note that φ((a + bi) + (c + di)) = a + c (b + d) = (b + d) a + c a b + c d = φ(a + bi) + φ(c + di). Also, φ((a + bi)(c + di)) = φ((ac bd) + (ad + bc)i) = (ac bd) (ad + bc)) = a b c d = (ad + bc) ac bd) b a d c
b a d c
φ(a + bi)φ(c + di)
44. φ((a1, . . . , an) + (b1, . . . , bn)) = ( a1, . . . , an) = ( b1, . . . , bn) implies (a1, . . . , an) = (b1, . . . , bn)) so that φ is 1-1. For any (a1, . . . , an), we have φ( a1, , an) = (a1, , an) so φ is onto.
φ((a1 + b1, . . . , an + bn)) = ( (a1 + b1), . . . , (an + bn)) = ( a1 , , an )( b1 , , bn ) = φ((a1 , , an )) + φ((b1 , , bn)). φ reflects each point through the origin.
45. Yes, by Cayley’s Theorem.
46. (ab)2 = a2b2 shows that the mapping is O.P. To show it is 1-1, note that a2 = b2 , implies e = a2b 2 = (ab 1)2 so that ab 1 = 1 or 2. Thus, a = b. Since G is finite, 1-1 implies onto. For Z under addition, g 2g is not onto but is 1 1 and operation preserving.
47. Observe that φg (y) = gyg 1 and φzg (y) = zgy(zg) 1 = zgyg 1 z 1 = gyg 1 , since z ∈ Z(G). So, φg = φzg.
48. In R under addition every nonzero element has infinite order. In R∗ under multiplication 1 has order 2.
49. φg = φh implies gxg 1 = hxh 1 for all x This implies h 1 gx(h 1 g) 1 ) = x, and therefore h 1 g ∈ Z(G). φg = φh if and only if h 1 g ∈ Z(G).
50. α(x) = (12)x(12) and β(x) = (123)x(123) 1 .
51. By Exercise 49, φα = φβ implies β 1α is in Z(Sn) and by Exercise 70 in Chapter 5, Z(Sn) = {ǫ}, which implies that α = β.
52. Since 2 is not in the image, the mapping is not onto.
53. Since both φ and γ take e to itself, H is not empty. Assume a and b belong to H. Then φ(ab 1) = φ(a)φ(b 1) = φ(a)φ(b) 1
). Thus ab 1 is in H.
54. G is Abelian.
55. (12)H(12) and (123)H(123) 1 .
56. Since R45 = 8, it must map to elements of order 8. Since the integers between 1 and 8 relatively prime to 8 are 1, 3, 5, 7, the elements of order 8 are R45, R 3 , R 5 , R 7 . 45 45 45
57. Since 1 is the unique element of C∗ of order 2, φ( 1) = 1. Since i and i are the only elements of C∗ of order 4, φ(i) = i or i
58. If a is an element not in Z(G), then φa is not the identity.
59. Z120, D60, S5 Z120 is Abelian, the other two are not. D60 has an element of order 60 and S5 does not.
60. Using Exercise 25 we have φ(V ) = φ(R180 H) = φ(R180 )φ(H) = R180 D = D′ .
61. For the first part, observe that φ(D) = φ(R90 V ) = φ(R90 )φ(V ) = R270 V = D′ and φ(H) = φ(R90 D) = φ(R90 )φ(D) = R270 D′ = H For the second part, we have that φ(D) = φ(R90 V ) = φ(R90 )φ(V ) = R90 V = D and φ(H) = φ(R180V ) = (φ(R90))2φ(V ) = R 2V = R180V = H.
64. In the general case, because φ is 1-1, G has no element whose order divides order n.
65. The first statement follows from the fact that every element of Dn has the form i 360/n i 360/n F . Because α must map an element of order n to an element of order n, R360 /n must map to Ri where i ∈ U (n). Moreover, F must map to a reflection (see Exercise 20). Thus we have at most n|U (n)| choices.
R or R
66. By the previous exercise, the mapping from Dn to Inn(Dn) that takes a to the inner automorphism induced by φa(x) = axa 1 is 1-1 when n is odd, and 2-1 when n is even.
67. In both cases H is isomorphic to the set of all even permutations of the set of four integers, so it is isomorphic to A4
68. By part 2 of Theorem 6.1, nφ(1/n) = φ(1) so that φ(1/n) = (1/n)φ(1). Also, by Part 2 of Theorem 6.2, φ(m/n) = mφ(1/n) = m(1/n)φ(1) = (m/n)φ(1).
69. The mapping φ(x) = x2 is one-to-one from Q+ to Q+ since a2 = b2 implies a = b when both a and b are positive. Moreover, φ(ab) = φ(a)φ(b) for all a and b However, φ is not onto since there is no rational whose square is 2. So, the image of φ is a proper subgroup of Q+ .
70. The argument given in Exercise 68 shows that an isomorphic image of Q has the form aQ where a = φ(1). But aQ = Q.
71. Suppose t √ h a t√φ is an a√ u t o m o√ r ph is m o√ f R∗ and a is positive. Then φ(a) = φ( a a) = φ( a)φ( a) = φ( a)2 > 0. Now suppose that a is negative but φ(a) = b is positive. Then, by the case we just did, a = φ 1(φ(a)) = φ 1(b) is positive. This is a contradiction. Here is an alternate argument for the case that a is negative and φ(a) is positive. Because 1 is the only real number of order 2 and the first case, we know that 0 < φ( a) = φ( 1)φ(a) = φ(a), which is a contradiction.
72. If φ were an isomorphism, then 0 = φ(1) = φ( 1 · 1) = φ( 1) + φ( 1) = 2φ( 1) implies that φ( 1) = 0. But then, φ is not 1 1.
73. Say φ is an isomorphism from Q to R + and φ takes 1 to a. It follows that the integer r maps to ar . Then a = φ(1) = φ(s 1 ) = φ( 1 + + 1 ) = φ( 1 )s and therefore s s a 1 = φ( 1 ). Thus, the rational r/s maps to ar/s . But a number r/s. s s r/s /= aπ for any rational
s s
74. In Q∗ | 1| = 2, whereas the only element of finite order in Q+ is 1.
75. Send each even permutation in Sn to itself. Send each odd permutation α in Sn to α(n + 1, n + 1). This does not contradict Theorem 5.5, because the subgroup is merely isomorphic to An+2, not the same as An+2. In particular, this example shows that an isomorphism from one permutation group to another permutation group need not preserve oddness.
CHAPTER 7
Cosets and Lagrange’s Theorem
1. H, 1 + H, 2 + H. To see that there are no others, notice that for any integer n we can write n = 3q + r where 0 ≤ r < 3. So, n + H = r + 3q + H = r + H, where r = 0, 1 or 2. For the second part there are n left cosets: 0 + ⟨n⟩, 1 + ⟨n⟩, , n 1 + ⟨n⟩
2. b a ∈ H.
3.
a. 11 + H = 17 + H because 17 11 = 6 is in H;
b. 1 + H = 5 + H because 5 ( 1) = 6 is in H;
c. 7 + H /= 23 + H because 23 7 = 16 is not in H.
4. Since 8/2 = 4, there are four cosets. Let H = 1, 11 The cosets are H, 7H, 13H, 19H.
5. Five: ⟨a5⟩, a⟨a5⟩, a2⟨a5⟩, a3⟨a5⟩,
⟩. Since ⟨
4⟩ = ⟨a2⟩ there are two cosets: ⟨a4 ⟩, a⟨a4⟩
6. Let F and F ′ be distinct reflections in D3. Then, take H = {R0, F } and K = {R 0 , F ′}.
7. Suppose that H /= ⟨3⟩. Let a ∈ H but a not in ⟨3⟩. Then a + ⟨3⟩ = 1 + ⟨3⟩ or a + ⟨3⟩ = 2 + ⟨3⟩.
8. Observe that aH = bK implies that b 1aH = K Thus, b 1aH is a subgroup of G From part 9 of the lemma in Chapter 7, we have that b 1a is in H and therefore b 1aH = H
9. Let ga belong to g(H ∩ K) where a is in H ∩ K Then by definition, ga is in gH ∩ gK. Now let x ∈ gH ∩ gK. Then x = gh for some h ∈ H and x = gk for some k ∈ K. Cancellation then gives h = k. Thus x ∈ g(H ∩ K).
10. By Lagrange’s Theorem, H = 5, 31 or 155. But H = 5 implies that all non-identity elements in H have order 5 and H = 31 implies that all non-identity elements in H have order 31.
11. Suppose that h H and h < 0. Then hR+ hH = H. But hR+ is the set of all negative real numbers. Thus H = R∗
12. The coset containing c + di is the circle with center at the origin and radius √ c2 + d2
13. By Lagrange’s Theorem the possible orders are 1, 2, 3, 4, 5, 6, 10, 12, 15, 20, 30, 60. 14. 84 or 210.
15. By Lagrange’s Theorem, the only possible orders for the subgroups are 1, p and q. By Corollary 3 of Lagrange’s Theorem, groups of prime order are cyclic. The subgroup of order 1 is ⟨e⟩
16. Note that φ(n) = U (n) , then use Corollary 4 of Lagrange’s Theorem and mimic the proof of Corollary 5 of Lagrange’s Theorem.
17. By Exercise 16 we have 56 mod 7 = 1. So, using mod 7 we have 515 = 56 56 52 5 = 1 1 4 5 = 6; 713 mod 11 = 2.
18. Note that n 1 ∈ U (n) and has order 2.
19. By Corollary 4 of Theorem 7.1, gn = e. Then since gm = e and gn = e we know by Corollary 2 of Theorem 4.1 that |g| is a common divisor of both m and n So, |g| = 1.
20. Since |H ∩ K| must divide 12 and 35, |H ∩ K| = 1. If H and K are relatively prime, |H ∩ K| = 1.
21. First observe that for all n ≥ 3 the subgroup of rotations of Dn is isomorphic to Zn . If n is even let F be any reflection in Dn. Then the set {R0, R180, F, FR180 } is closed and therefore a subgroup of order 4. Now suppose that Dn has a subgroup K of order 4. By Lagrange, |Dn| = 2n = 4k and therefore n = 2k.
22. Since |H ∩ K| must divide both 12 and 18, it must divide 6. The case where |H ∩ K| = 1 is trivial. The case where |H ∩ K| = p follows from Corollary 3 of Theorem 7.1. The case where |H ∩ K| = 2p follows from Theorem 7.3.
23. Since G has odd order, no element can have order 2. Thus, for each x /= e, we know that x /= x 1 So, because G is Abelian, we can write the product of all the elements in the form ea1a1 1 a2a2 1 · · · a n a n 1 = e.
24. Let G be a group of order 4. If G has an element of order 4, then G is cyclic. So, every element in G has order 1 or 2. Then, for all a, b G, we have ab = (ab) 1 = b 1a 1 = ba.
25. For |G| = pn , the group is cyclic or ap n 1 = e for all a in G.
26. Since a is not in Z(G) we know C(a) is a proper subgroup of G. Thus, C(a) G /2. But then the numbers of elements not in C(a) is at least G /2. If G has odd order, then at least 2/3 of the elements in G do not commute with a.
27. Let G be a group of order 77. If G is cyclic, then by Theorem 4.3, G has exactly one subgroup of order 7 and exactly one of order 11. If G is not cyclic, by Lagrange, its 76 non-identity elements have orders 7 or 11. By Corollary of Theorem 4.4, G has at most 72 elements of order 6 and therefore must have elements of order 11, which come 10 at a time. Theorem 7.2 shows that G cannot have more that one subgroup of order 11, so there are exactly 66 elements of order 7.
28. Since G is non-Abelian, it has no element of order 8. By Lagrange, any nonidentity element must have order 2 or 4. Finally, by Exercise 45 of Chapter 2, not all nonidentity elements have order 2.
29. If the group is cyclic, Theorem 4.3 says that it has exactly one subgroup of order 5. So, assume the group is not cyclic. Not all of the 54 nonidentity elements can have order 5 because the number of elements of order 5 is a multiple of φ(5) = 4. So the group has an element of order 11. Also, since φ(11) = 10, the number of elements of order 11 is a multiple of 10. If there were more than 10, the group would have distinct subgroups H and K of order 11. But then HK = H K / H K = 121. So, excluding the subgroup of order 11, there are 44 elements remaining and each has order 5. That gives us exactly 11 subgroups of order 5.
30. Suppose that G has no element of order 105. Then, the possible orders of the elements are 1, 3, 5, 7, 15, 21 and 35. Since there is a unique subgroup of each of these orders, all the elements of order 3 occur in the unique subgroup of order 3, and likewise for the other orders. Now note that for any order k in this problem the number of elements of order k is φ(k). So, there is 1 element of order 1, 2 of order 3, 4 of order 5, 6 of order 7, 8 of order 15, 12 of order 21, and 24 of order 35. But this totals to fewer that 105, which is a contradiction.
Alternate proof: Again, assume that there is no element of order 105. Let H be the unique subgroup of order 35 and K the unique subgroup of order 15. Observe that H must have an element of order 35, for otherwise all 34 non-identity elements of H would have order 5 or 7. But G has only 4 elements of order 5 and 6 of order 7. So, H contains all the elements of G of orders 1, 5, 7 and 35. Likewise, K contains all the elements of G of orders 1, 3, 5 and 15. Thus, H K contains every element of G. But H K has less than 51 elements (the exact number is 45, since the elements of orders 1 and 5 appear in both H and K).
31. By Lagrange’s theorem every element in G has an order that is a divisor of n. So, we can partition the n elements of G according to their orders. For each divisor d of n, let md be the number of elements in G of order d By our assumption, md is φ(d) where φ is the Euler phi function. (If there were more than φ(d) elements of order d in G, then G would have at least 2 subgroups of order d.) So n = Σmd where d ranges over all divisors d of n We also have from Exercise 87 of Chapter 4 that n = Σφ(d) where d ranges over all divisors d of n. This proves that each nd = φ(d). In particular, mn /= 0.
32. By Theorem 7.3, G is isomorphic to Z2p or Dp. So the number of elements of order 2 is 1 or p + 1.
33. Suppose that H and K are distinct subgroups of order m. Then | HK| = | H|| K| = ≤ 2m and therefore ≤ |H ∩ K|. Since m is odd and H |H ∩K| m 2 and K are distinct, we know that This is impossible. 2 < |H ∩ K| < m and that |H ∩ K| divides m
34. For any positive integer n let ωn = cos 2π + i sin 2π . The finite subgroups of C∗ are those of the form ωn . To verify this, let H denote any finite subgroup of C∗ of order n Then every element of H is a solution to xn = 1. But the solution set of xn = 1 in C∗ is ⟨ωn⟩.
36. By Exercise 35, p = [G : H] = [G : K][K : H] so [G : H] = 1 or [K : H] = 1.
37. Cyclic subgroups of order 12 and 20 in Dn must be in the subgroup of rotations. So, n must be the smallest positive integer divisible by 12 and 20, which is 60.
38. Since |g| must divide both 14 and 21, |g| = 1 or 7.
39. Let a have order 3 and b be an element of order 3 not in a . Then a b = aibj i = 0, 1, 2; j = 0, 1, 2 is a subgroup of G of order 9. Now use Lagrange’s Theorem.
40. Suppose that G / Z(G = p where p is prime. Let a G, but a Z(G). Since the subgroup C(a) contains both a and Z(G) we know that Z(G) is proper subgroup of
C(a). Since p = (|G|/|C(a)|)(|C(a)|/|Z(G)|) we have |G|/|C(a)| = 1 and therefore C(a) = G and a ∈ Z(G).
41. By Corollary 5 of Theorem 7.1, the statement is true for n = 1. For the sake of induction assume that ap k = a. Then ap k+1 = ap k ap = ap = a.
42. Suppose G is a group of order 63. Let a be any non-identity element in G By Lagrange’s Theorem, a = 63, 21, 9, 7, or 3. If a = 3, we are done. If a = 63, then a21 = 3; if a = 21, then a7 = 3; and if a = 9, then a3 = 3. So, if any of these cases occur we are done. Thus, we may assume that all 62 non-identity elements in G have order 7. But, by the Corollary to Theorem 4.4, the number of elements of order 7 must be a multiple of 6.
| | | | | | | | |
43. Let a ∈ G and |a| = 5. Then by Theorem 7.2 we know that the set ⟨a⟩H has exactly 5 · |H|/|⟨a⟩ ∩ H| elements and |⟨a⟩ ∩ H| divides |⟨a⟩| = 5. It follows that |⟨a⟩ ∩ H| = 5 and therefore ⟨a⟩ ∩ H = ⟨a⟩.
44. Suppose the ak = bk By the Corollary to Theorem 0.2, there are integers s and t such that 1 = ns + kt. Then, by Corollary 4 of Lagrange’s Theorem ,we have a = ans+kt = (an)s(ak)t = (ak)t = (bn)s(bk)t = bns+kt = b To prove that the mapping is an automorphism when the group is also Abelian, note that by Exercise 10 of Chapter 5 a 1-1 mapping from a finite set to itself is onto. Lastly, observe that (ab)k = akbk .
45. First observe that by Corollary 2 of Lagrange’s Theorem every positive integer k with the property that xk = e for all x in G is a common multiple of orders of all the elements in G. So, d is the least common multiple of the orders of the elements of G. Since |G| is a common multiple of the orders of all the elements of G it follows directly from the division algorithm (Theorem 0.1) that |G| is divisible by d.
46. Since reflections have order 2, the subgroup must consist entirely of rotations and the subgroup of all rotations is cyclic.
47. Let G be a finite Abelian group. The case when G has 0 or 1 element of order 2 corresponds to the cases n = 0 and n = 1. Let H = x G x2 = e . Then H is a subgroup of G that consists of the identity and all elements of order 2. It suffices to prove |H| = 2n . If G has at least two elements of order 2, say a1 and b, then H1 = {e, a1, b, a1b} is a subgroup of order 22 If H1 = H, we are done. If not, let a2 ∈ H but not in H1. Then H2 = H1 ∪ ⟨a2⟩H1 is a subgroup of H of order 2|H1| = 23 . If H2 = H we are done. If not, let a3 ∈ H but not in H2 and let H3 = H2 ∪ ⟨a3⟩H2. We can continue this argument until we reach H.
48. Suppose that an Abelian group G has order 15 and is not cyclic. Then, G has 14 elements of order 3 or 5. Since 14 is not a multiple φ(5) = 4, G must have an element a of order 3. It follows from Example 7 of Chapter 3 and Theorem 7.2 that G cannot have more than one subgroup of order 3 or 5. Since the union of the unique subgroups of order 3 and 5 has only 7 elements, the remaining 8 must have order 15.
49. Let G = 2k + 1. Observe a = ae = aa2k+1 = a2k+2 = (ak+1)2 . To prove uniqueness suppose the x2 = a = y2 . Then x(x2)k = x2k+1 = e = y2k+1 = y(y2)k . So, we can cancel the terms (x2)k and (y2)k to obtain x = y
50. Let G be Abelian and G = 4pn , where p is a prime. If G has an element a with a of the form pm, then am = p So, we may assume that all nonidentity elements of G have order 2 or 4. If G has an element a of order 4, then for any element b not in
a , we have by Theorem 7.2 that a b is a subgroup of G of order 8 or 16, which contradicts Lagrange’s Theorem. Lastly, we may assume that every non-identity element of G has order 2. In this case, let a and b be distinct elements of G of order 2. Then H = {e, a, b, ab} is a subgroup of G of order 4 and for any element c in G that is not in H we have H⟨c⟩ is a subgroup of order 8.
51. If H and K are distinct subgroups of order pm , then npm = |G| ≥ |HK| = |H||K|/|H ∩ K| ≥ pm pm /pm 1 = ppm , which is obviously false.
52. If G has an element of order qk or pqk , then by the Fundamental Theorem of Cyclic Groups, it has an element of order q So we may assume that every nonidentity element of G has order p. Let a /= e belong to G and b belong to G with b not in ⟨a⟩. Then, since ⟨a⟩ ∩ ⟨b⟩ = {e} we have by Theorem 7.2, that pqn = |G| ≥ |⟨a⟩⟨b⟩|/|⟨a
∩ ⟨b⟩| = p2 But then qn ≥ p
53. Let G be the group and H the unique subgroup of order q. We must show that G has an element of order pq. Let a belong to G but not in H. By Lagrange, |a| = p or pq. If |a| = pq we are done. So, we may assume that a has order p and we let K = ⟨a⟩ Then H ∪ K accounts for q + p 1 elements (the identity appears twice). Pick b ∈ G but b not one of the elements in H ∪ K. Then L = ⟨b⟩ is a subgroup of G of order p different than K. Then K ∩ L = {e} because |K ∩ L| must divide p and is not p. By Theorem 7.2 |KL| = |K||L|/|K ∩ L| = (p · p)/1 = p2 . But a group of order pq with q < p cannot have p2 elements. This shows that b cannot have order p. So |b| = pq.
54. |H| = 1 or p where p is a prime. To see this, suppose that |H| > 1 and let a ∈ H and a /= e. Let |a| = pm where p is a prime. Then |⟨am ⟩| = p and H ⊆ ⟨am ⟩, so |H| = p. An example, where |H| = p, is G = Zp k where p is prime and k ≥ 1.
55. Let H and K be distinct subgroups of order 5. Then by Theorem 7.2 the subset HK has order 25. In the statement of the exercise, replace 5 with any prime p and 25 by p2 .
56. Say |a| = n Then φ n(x) = anxa n = x, so that φ n is the identity. For example, a a take a = R90 in D4 Then |φa| = 2.
57. Since the order of G is divisible by both 10 and 25, it must be divisible by 50. But the only number less than 100 that is divisible by 50 is 50.
58. If Z(A4) > 1, then A4 would have an element of order 2 or order 3 that commutes with every element. But any subgroup generated by an element of order 2 and an element of order 3 that commute has order 6. This contradicts the fact shown in Example 5 that A4 has no subgroup of order 6.
59. Let K be the set of all even permutations in H Since K is closed, it is a subgroup of H. If K = H, we are done. If not, let α be an element in H that is odd. Then αK must be the set of all odd permutations in H, for if β is any odd permutation in H, we have α 1 β ∈ H, which means β ∈ H. Thus |H| = |K ∪ αK| = 2|K|.
60. Observe that |orbA5 (5)| = 5. Now use the Orbit-Stabilizer Theorem to show that stabA5 (5) = 12. Note that the same argument applies to stabA5 (i) for i = 1, 2, 3, and 4.
61. Suppose that H is a subgroup of A5 of order 30. We claim that H contains all 20 elements of A5 that have order 3. To verify this, assume that there is some α in A5 of order 3 that is not in H. Then A5 = H ∪ αH. It follows that α2H = H or
α2 = αH. Since the latter implies that α H, we have that α2H = H, which implies that α2 H But then α = α2 H, which is a contradiction of our assumption that α is not in H. The same argument shows that H must contain all 24 elements of order 5. Since |H| = 30 we have a contradiction.
62. Suppose that H is a subgroup of A5 of order 20. We claim that H contains all 24 elements of A5 that have order 5. To verify this, assume that there is some α in A5 of order 5 that is not in H. Then A5 = H αH α2H. To see that the coset α2H is not the same as H, note that α2H = H implies that α2 H and α = α2 . Moreover, α2H is not the same as αH, for then α H. It follows that α3H is equal to one of the cosets H, αH or α2H If α3H = H, then α3 H and therefore α = α3 H, which contradicts the assumption that α is not in H. If α3H = αH, then α2 H and therefore α = α2 H, which contradicts the assumption that α is not in H. If α3H = α2H, then α H, which contradicts the assumption that α is not in H. The same argument shows that H must contain all 24 elements of order 5. Since H = 20, we have a contradiction. An analogous argument shows that A5 has no subgroup of order 15.
63. Say H is a subgroup of order 30. By Exercise 61, H is not a subgroup of A5 and by Exercise 27 of Chapter 6, H A5 is a subgroup of A5 of order 15. But this contradicts Exercise 52.
64. For any β H we have (αβ)(1)α(β(1)) = α(1) = 3. So, αH is a subset of the set all elements of Sn that send 1 to 3. Let γ(1) = 3. Then α 1γ(1) = α 1(3) = 1. So, α 1γ H and therefore γ αH. In the general case, for α Sn (n 3) and i and j are integers between 1 and n with α(i) = j and H = stab(i), then αH is a subset of the set all elements of Sn that send i to j
65. If H is a subgroup of S5 of order 60 other than A5, then it follows from Theorem 7.2 that |A5 ∩ H| = 30, which contradicts Exercise 61.
66. n = 1, 2, 3. To see that there are no others, note that |S4| = 24 does not divide 120, S5 does not have an element of order 60 and D60 does, and for n > 5, |Sn| > 120.
67. Certainly, a orbG(a). Now suppose c orbG(a) orbG(b). Then c = α(a) and c = β(b) for some α and β, and therefore (β 1α)(a) = β 1(α(a)) = β 1(c) = b So, if x ∈ orbG(b), then x = γ(b) = γ(β 1α)(a)) = (γβ 1α)(a). This proves orbG(b) ⊆ orbG(a). By symmetry, orbG(a) ⊆ orbG(b).
70. Think of a cube as sitting on a table top with one face perpendicular to your line of sight. The four lines that join the upper corner of a cube to the midpoint of diametrically opposite vertical edge are axes of rotational symmetry of 120 degrees. The four lines that join the upper corner of a cube to the midpoint of lower horizontal edge at the maximum distance from the starting corner are also axes of rotational symmetry of 120 degrees.
71. Suppose that B ∈ G and det B = 2. Then det (A 1B) = 1, so that A 1B ∈ H and therefore B ∈ AH Conversely, for any Ah ∈ AH we have det Ah = (det (A))(det (h)) = 2 · 1 = 2.
72. The circle passing through Q, with center at P.
73. It is the set of all permutations that carry face 2 to face 1.
74. The order of the symmetry group would have to be 6 · 20 = 120.
∈
75. If aH = bH, then b 1a H. So det (b 1 a) = (det b 1 )(det a) = (det b 1 )(det a) = (det b) 1 (det a) = 1. Thus det a = det b Conversely, we can read this argument backwards to get that det a = det b implies aH = bH.
76. a. 12 b. 24 c. 60 d. 60
77. To prove that the set is closed, note that αβ2 = (13) = β2α3 , α2 β2 = (14)(23) = β2 α2 , and α3 β2 = (24) = β2 α.
CHAPTER 8
External Direct Products
1. Closure and associativity in the product follows from the closure and associativity in each component. The identity in the product is the n-tuple with the identity in each component. The inverse of (g1, g2, , gn) is (g1 1 , g2 1 , , g n 1 )
2. In general, (1, 1, . . . , 1) is an element of largest order in Zn1 Zn2 Znt . To see this, note that because the order of the 1 in each component is the order of the group in that component, |(1, 1, . . . , 1)| = lcm(n1, n2, . . . , nt) and the order of every element in the product must divide lcm(n1, n2, , nt).
3. The mapping φ(g) = (g, eH) is an isomorphism from G to G eH . To verify that φ is one-to-one, we note that φ(g) = φ(g ′) implies (g, eH ) = (g ′ , eH) which means that g = g ′ . The element (g, eH) G eH is the image of g. Finally, φ((g, eH )(g ′ , eH )) = φ((gg ′ , eHeH )) = φ((gg ′ , eH )) = gg ′ = φ((g, eH ))φ((g ′ , eH)). A similar argument shows that φ(h) = (eG, h) is an isomorphism from H onto {eG} ⊕ H.
4. (g, h)(g ′ , h′) = (g ′ , h′)(g, h) for all g, g ′ , h, h′ if and only if gg ′ = g ′ g and hh′ = h′h, that is, if and only if G and H are Abelian. A corresponding statement holds for the external direct product of any number of groups.
5. If Z ⊕ Z = ⟨(a, b)⟩ then neither a nor b is 0. But then (1, 0) /∈ ⟨(a, b)⟩. Z ⊕ G is not cyclic when |G| > 1.
6. Z8 ⊕ Z2 contains elements of order 8, while Z4 ⊕ Z4 does not.
7. Define a mapping from G1 ⊕ G2 to G2 ⊕ G1 by φ(g1, g2) = (g2, g1). To verify that φ is one-to-one, we note that φ((g1, g2)) = φ((g1 ′ , g2 ′ )) implies (g2, g1) = (g2 ′ , g1 ′ ). From this we obtain that g1 = g1 ′ and g2 = g2 ′ The element (g2, g1) is the image on (g1, g2) so φ is onto. Finally, φ((g1, g2)(g1 ′ , g2 ′ )) = φ((g1g1 ′ , g2g2 ′ )) = (g2g2 ′ , g1g1 ′ ) = (g2, g1)(g2 ′ , g1 ′ ) = φ((g1, g2))φ((g1 ′ , g2 ′ )). In general, the external direct product of any number of groups is isomorphic to the external direct product of any rearrangement of those groups.
8. No, Z3 ⊕ Z9 does not have an element of order 27. See also Theorem 8.2.
9. In Zpm ⊕ Zp take ⟨(1, 0)⟩ and ⟨(1, 1)⟩.
10. Z9 has 6 elements of order 9 (the members of U (9)). Any of these together with any element of Z3 gives an ordered pair whose order is 9. So Z3 Z9 has 18 elements of order 9.
11. In both Z4 Z4 and Z8000000 Z400000, (a, b) = 4 if and only if a = 4 and b = 1, 2 or 4 or if b = 4 and a = 1 or 2 (we have already counted the case that a = 4). For the first case, we have φ(4) = 2 choices for a and φ(4) = φ(2) + φ(1) = 4 choices for b to give us 8 in all. For the second case, we have φ(4) = 2 choices for b and φ(2) + φ(1) = 2 choices for b This gives us a total of 12.
In the general case, observe that by Theorem 4.4 as long as d divides n, the number of elements of order d in a cyclic group depends only on d.
14. The group of rotations is Abelian and a group of order 2 is Abelian; now use Exercise 4.
15. Define a mapping φ from C to R R by φ(a + bi) = (a, b). To verify that φ is one-to-one, note that φ(a + bi) = φ(a ′ + b′i) implies that (a, b) = (a ′ , b′). So, a = a ′ and b = b′ and therefore a + bi = a ′ + b′i. The element (a, b) in R R is the image of a + bi so φ is onto. Finally, φ((a + bi) + (a ′ + b′i)) = φ((a + a ′) + (b + b′)i) = (a + a ′ , b + b′) = (a, b) + (a ′ , b′) = φ(a + bi) + φ(a ′ + b′i).
16. Let α : G1 → G2 and β : H1 → H2 be isomorphisms. Then γ : G1 ⊕ H1 → G2 ⊕ H2 given by γ((g1 , h1 )) = (α(g1 ), β(h1 )) is an isomorphism. A corresponding statement holds for the external direct product of any number of groups.
17. By Exercise 3 in this chapter, G is isomorphic to G ⊕ {eH} and H is isomorphic to {eG} ⊕ H Since subgroups of cyclic groups are cyclic, we know that G ⊕ {eH} and eG H are cyclic. In general, if the external direct product of any number of groups is cyclic, each of the factors is cyclic.
} ⊕
18. ⟨(10, 10⟩; ⟨20⟩ ⊕ ⟨5⟩.
19. ⟨m/r⟩ ⊕ ⟨n/s⟩.
20. Observe that Z9 ⊕ Z4 ≈ Z4 ⊕ Z9 ≈ ⟨3⟩ ⊕ ⟨2⟩.
21. Since ⟨(g, h)⟩ ⊆ ⟨g⟩ ⊕ ⟨h⟩, a necessary and sufficient condition for equality is that lcm(|g|, |h|) = |(g, h)| = |⟨g⟩ ⊕ ⟨h⟩| = |g||h|. This is equivalent to gcd(|g|, |h|) = 1.
22. 48; 6
23. In the general case there are (3n 1)/2.
24. In this case, observe that |(a, b)| = 2 if and only if |a| = 1 or 2 and |b| = 1 or 2 but not both |a| = 1 and |b| = 1. So, there are (m + 2)(n + 1) 1 = mn + m + 2n + 1 elements of order 2. For the second part, observe that |(a, b)| = 4 if and only if |a| = 4 and |b| = 1 or 2. So, there are 2(n + 1) elements of order 4.
25. Define a mapping φ from M to N by φ a b c d = (a, b, c, d). To verify that φ is one-to-one we note that φ a b c d = φ a ′ b′ implies (a, b, c, d) = (a ′ , b′ , c ′ , d′). Thus a = a ′ , b = b′ , c = c ′ , and d = d′ This proves that φ is one-to-one. The element (a, b, c, d) is the image of a b c d so φ is onto. Finally, φ a b c d a ′ b′ c ′ d′ a + a ′ b + b′ c + c ′ d + d′ = (a +a ′ , b +b′ , c +c ′ , d +d′) = (a, b, c, d) + (a ′ , b′ , c ′ , d′) = φ a b c d + φ a ′ b′
′ d′ = φ
26. D6. Since S3 ⊕ Z2 is non-Abelian, it must be isomorphic to A4 or D6. But S3 ⊕ Z2 contains an element of order 6 and A4 does not. +
d
Let Rk denote R R R (k factors). Then the group of m n matrices under addition is isomorphic to Rmn .
27. Since (g, g)(h, h) 1 = (gh 1 , gh 1), H is a subgroup. When G = R, G G is the plane and H is the line y = x.
32. In general, if m and n are even, then Zm Zn has exactly 3 elements of order 2. For if (a, b) = 2, then a = 1 or 2 and b = 1 or 2 but not both, a and b have order 1. Since any cycle group of even order has exactly 1 element of order 2 and 1 of order 1, there are only 3 choices for (a, b).
33. Noting that Z4 ⊕ Z3 ⊕ Z2 ≈ Z4 ⊕ Z6 we find ⟨15⟩ ⊕ ⟨10⟩ Noting that Z4 ⊕ Z3 ⊕ Z2 ≈ Z2 ⊕ Z12 we find ⟨50⟩ ⊕ ⟨5⟩.
34. ⟨25⟩ ⊕ ⟨R90⟩.
35. Let F be a reflection in D3 {R0, F } ⊕ {R0, R180, H, V }
36. ⟨4⟩ ⊕ ⟨0⟩ ⊕ ⟨5⟩
37. In R∗ ⊕ R∗ (1, 1), ( 1, 1) and ( 1, 1) have order 2, whereas in C∗ the only element of order 2 is 1. But isomorphisms preserve order.
38. Z3 ⊕ Z3
39. Define the mapping from G to Z Z by φ(3m 6n) = (m, n). To verify that φ is one-to-one note that φ(3m 6n) = φ(3s6t) implies that (m, n) = (s, t), which in turn implies that m = s and n = t So, 3m 6n = 3s6t The element (m, n) is the image of 3m 6n so φ is onto. Finally, φ((3m 6n )(3s 6t)) = φ(3m + s 6n + t) = (m + s, n + t) = (m, n) + (s, t) = φ(3m 6n )φ(3
6
) shows that φ is operation preserving. When G = {3m 9n | m, n ∈ Z} the correspondence from G to Z ⊕ Z given by φ(3m 9n) = (m, n) is not well-defined since φ(3290) /= φ(3091) and 3290 = 9 = 3091 .
40. |ai| = ∞ for some i.
41. Both D6 and D3 Z2 have 1 element of order 1, 7 of order 2, 2 of order 3, and 2 of order 6. (In fact, they are isomorphic as we see in Example 19 in Chapter 9.)
42. Observe that U (40) ⊕ Z6 ≈ U (8) ⊕ U (5) ⊕ Z6 ≈ Z2 ⊕ Z2 ⊕ Z4 ⊕ Z6 and U (72) ⊕ Z4 ≈ U (9) ⊕ U (8) ⊕ Z4 ≈ Z6 ⊕ Z2 ⊕ Z2 ⊕ Z4 so they are isomorphic.
43. C∗ has only one element of order 2 whereas Zm Zn has exactly one element of order 2 if and only if it is cyclic, which is true if and only if gcd(m, n) = 1.
44. If exactly one ni is even, then x is the unique element of order 2. Otherwise x is the identity.
To see this observe that for each a with |a| > 2, a and a 1 are distinct, so aa 1 contribute e to the product. Let H = {a ∈ G | a2 = e} and let |H| = 2n . Then H is isomorphic to a group of the form Z2 Z2 Z2. The problem is now reduced to H. If n = 0 or 1, we are done. If n 2, arrange the elements of H without the parentheses and commas in a vertical array with any element at the top, any unused
∈ { }
element below it, any unused element below the second one, and so on. Now let x = (a1, a2, . . . , an) be the product of all the elements of H and let i 1, 2, . . . , n . Because the set of elements of G that have 0 is the ith column form a subgroup of order 2n 1 and the set of elements of G that have 1 the ith column consist of the other 2n 1 elements of the column, the sum of all the elements in the ith column is 0. So, ai = 0.
45. Each cyclic subgroup of order 6 has two elements of order 6. So, the 24 elements of order 6 yield 12 cyclic subgroups of order 6. In general, if a group has 2n elements of order 6, it has n cyclic subgroups of order 6. (Recall from the Corollary of Theorem 4.4, if a group has a finite number of elements of order 6, the number is divisible by φ(6) = 2).
46. Z ⊕ D4 ⊕ A4.
47. Aut(U (25)) ≈ Aut(Z20) ≈ U (20) ≈ U (4) ⊕ U (5) ≈ Z2 ⊕ Z4
48. S3
49. In each position we must have an element of order 1 or 2 except for the case that every position has the identity. So, there are 2k 1 choices. For the second question, we must use the identity in every position for which the order of the group is odd. So, there are 2t 1 elements of order 2 where t is the number of n1 , n2 , . . . , nk that are even.
50. Z10 ⊕ Z12 ⊕ Z6 ≈ Z2 ⊕ Z5 ⊕ Z12 ⊕ Z6 ≈ Z2 ⊕ Z60 ⊕ Z6 ≈ Z60 ⊕ Z6 ⊕ Z2. Z10 ⊕ Z12 ⊕ Z6 has 7 elements of order 2 whereas Z15 ⊕ Z4 ⊕ Z12 has only 3.
51. Part a. φ(18) = 6; to find an isomorphism all we need do is take 1 to a generator of Z2 ⊕ Z9. So, φ(1) = (1, 1), which results in φ(x) = (x, x) Another is φ(1) = (1, 2), which results in φ(x) = (x, 2x). Part b 0, because Z2 ⊕ Z3 ⊕ Z3 is not cyclic.
52. Since φ((2, 3)) = 2 we have 8φ((2, 3)) = 16 = 1. So, 1 = φ((16, 24)) = φ((1, 4)).
53. Since (2, 0) has order 2, it must map to an element in Z12 of order 2. The only such element in Z12 is 6. The isomorphism defined by (1, 1)x → 5x with x = 6 takes (2, 0) to 6. Since (1, 0) has order 4, it must map to an element in Z12 of order 4. The only such elements in Z12 are 3 and 9. The first case occurs for the isomorphism defined by (1, 1)x → 7x with x = 9 (recall (1, 1) is a generator of Z4 ⊕ Z3); the second case occurs for the isomorphism defined by (1, 1)x → 5x with x = 9.
54. U4(140) ≈ U (35) ≈ U (5) ⊕ U (7) ≈ Z4 ⊕ Z6.
55. Since a ∈ Zm and b ∈ Zn , we know that |a| divides m and |b| divides n So, |(a, b)| = lcm(|a|, |b|) divides lcm(m, n).
56. Map ax2 + bx + c to (a, b, c). In general, {an 1xn 1 + · · · + a0| an 1, , a0 ∈ Zm} under addition modulo m is isomorphic to Zm ⊕ · · · ⊕ Zm (n copies).
57. Up to isomorphism, Z is the only infinite cyclic group and it has 1 and 1 as its only generators. The number of generators of Zm is |U (m)| so we must determine those m such that |U (m)| = 2. First consider the case where m = pn, where p is a prime. Then the number of generators is pn 1(p 1). So, if p > 3, we will have more than 2 generators. When p = 3 we must have n = 1. Finally, |U (2n)| = 2n 1 = 2 only when n = 2. This gives us Z3 and Z4. When m = p1p2 · · · pk, where the p’s are distinct primes, we have |U (m)| = |U (p1)||U (p2)| |U (pk)|. As before, no prime can be greater than 3. So, the only case is m = 2 · 3 = 6.
58. Identify A with (0,0), T with (1,1), G with (1,0) and C with (0,1). Then a string of length n of the four bases is represented by a string of 0s and 1s of length 2n and the complementary string of a1a2 . . . a2n is a1a2 . . . a2n + 11 . . . 1.
59. Each subgroup of order p consists of the identity and p 1 elements of order p So, we count the number of elements of order p and divide by p 1. In Zp Zp every nonidentity element has order p, so there are (p2 1)/(p 1) = p + 1 subgroups of order p
62. U (165) ≈ U (11) ⊕ U (3) ⊕ U (5) ≈ Z10 ⊕ Z2 ⊕ Z4.
63. U (165) ≈ U (15) ⊕ U (11) ≈ U (5) ⊕ U (33) ≈ U (3) ⊕ U (55) ≈ U (3) ⊕ U (5) ⊕ U (11).
64. Note that U9(72) ≈ U (8) ≈ Z2 ⊕ Z2 and U4(300)) ≈ U (75) ≈ U (3) ⊕ U (25) ≈ Z2 ⊕ Z20.
65. From Theorem 8.3 and Exercise 3 we have U (2n) ≈ U (2) ⊕ U (n) ≈ U (n).
66. Since U (2n) is isomorphic to Z2n 2 Z2, and Z2n 2 and Z2 each have exactly one element of order 2, U (2n) has exactly three elements of order 2.
67. We use the fact that Aut(Z105) U (105) U (3) U (5) U (7) Z2 Z4 Z6. In order for (a, b, c) to have order 6, we could have c = 6 and a and b have orders 1 or 2. So we have 2 choices for each of a, b, and c. This gives 8 elements. The only other possibility for (a, b, c) to have order 6 is for c = 3 and a and b have orders 1 or 2, but not both have order 1. So we have 3 choices for a and b together and 2 choices for c. This gives 6 more elements for a total of 14 in all. For the second part, use the fact that U (27) ≈ Z18 .
68. By Theorem 6.5 we have Aut(Aut(Z50 )) ≈ Aut(U (50)) ≈ Aut(U (50)) ≈ Aut(Z20 ) ≈ U (20) ≈ Z2 ⊕ Z4.
69. A4 ⊕ Z4 has 7 elements of order 2 whereas the subgroup D12 ⊕ {0} of D12 ⊕ Z2 has
Alternate solution 1. A4 Z4 has 16 elements of order 12 whereas D12 Z2 has 8. (Note that these are consistent with the corollary of Theorem 4.4.)
Alternate solution 2. By Exercise 77 in Chapter 5 |Z(A4 ⊕ Z4)| = 1, whereas (R180, 0) is in the center of D12 ⊕ Z2
70. D8 ⊕ D3 has elements of order 8 and D6 ⊕ D4 does not.
71. Z ⊕ D4
72. U (pm ) U (qn) = Zpm p m 1 Zqn q n 1 and both of these groups have even order. Now use Theorem 8.2.
73. Observe that U (55) ≈ U (5) ⊕ U (11) ≈ Z4 ⊕ Z10 and U (75) ≈ U (3) ⊕ U (25) ≈ Z2 ⊕ Z20
2 ⊕
⊕
≈
⊕
Z4 ⊕ Z10. U (144) ≈ U (16) ⊕ U (9) ≈ Z4 ⊕ Z2 ⊕ Z6; U (140) ≈ U (4) ⊕ U (5) ⊕ U (7) ≈ Z2 ⊕ Z4 ⊕ Z6
74. U (900) Z2 Z6 Z20, so the element of largest order is the lcm(2, 6, 20) = 60.
75. From Theorem 6.5 we know Aut(Zn) ≈ U (n). So, n = 8 and 12 are the two smallest.
76. Observe that Z2 ⊕ Z4 ⊕ Z9 ≈ Z4 ⊕ Z2 ⊕ Z9 ≈ Z4 ⊕ Z18 ≈ U (5) ⊕ U (27) ≈ U (135).
77. Since U (pq) ≈ U (p) ⊕ U (q) ≈ Zp 1 ⊕ Zq 1 if follows that k = lcm(p 1, q 1).
78. U50 (200) = 1, 51, 101, 151 has order 4 whereas U (4) has order 2. U (200) = 80; U (50) U (4) = 40. This is not a contradiction to Theorem 8.3 because 50 and 4 are not relatively prime.
79. Zpn 1 . To see this, first observe that because U (pn) is cyclic, so is Up(pn). Now list the elements of Up(pn) as follows: 1 + p, 1 + 2p, , 1 + pn 1p = 1 + pn = 1. This gives us pn 1 elements.
|
80. Since U (pk) is cyclic and U (pk) = pk pk 1 = pk 1(p 1), we know from Theorem 4.3 and Lagrange’s theorem that U (pk ) has an element of order if and only if 4 divides p 1.
81. U8 (40) ≈ U (5) ≈ Z4.
82. In the first case, there are 2k 1; in the second case, there are 2k+2 1.
83. U5(140) ≈ U (28); U4(140) ≈ U (35) ≈ Z4 ⊕ Z6.
84. 3, 6, 8, 12, 24.
85. If x ∈ Ust(n), then x ∈ U (n) and x 1 = stm for some m. So, x 1 = s(tm) and x 1 = t(sm). If x ∈ Us(n) ∩ Ut(n), then x ∈ U (n) and both s and t divide x 1. So, by Exercise 6 in Chapter 0, st divides x 1.
86. This follows directly from Cayley’s Theorem (6.5 in Chapter 6).
87. Z2 ⊕ Z2.
88. Z2 ⊕ Z2 ⊕ Z2
89. Let pn1 pn2 · · · pnk be the prime-power decomposition of n. From the discussion following the corollary of Theorem 9.7, we know that U (n) is the internal direct product of groups U (pni ) for i = 1, 2, . . . , k. For odd pi, U (pni ) is cyclic of even i i order so H contains the generator the subgroup isomorphic to U (pni ). For pi = 2, every nonidentity element of U (pni ) has even order so H contains U (2ni ) as well.
90. First observe that U (n)k = U (n) if and only if the mapping x xk is 1 1. Let d = gcd( U (n ), k). If d > 1, let p be a prime divisor of d By Theorem 9.5, U (n) has an element a of order p. Then ap = 1 and therefore ak = 1. Now suppose that d = 1. If a, b ∈ U (n) and ak = bk , we have (ab 1)k = 1. Thus, |ab 1| is a common divisor of |U (n)| and k. This forces ab 1 = 1 and a = b.
| | ⊕ | { } i i · | | → 1 2 k
91. Since 5 29 = 1 mod 36, we have that s = 29. So, we have 3429 mod 2701 = 1415, which converts to NO.
92. None. Because gcd(18,12) = 6, Step 3 of the Sender part of the algorithm fails.
93. Because the block 2505 exceeds the modulus 2263, sending 2505e mod 2263 is the same as sending 242e mod 2263, which decodes as 242 instead of 2505.
CHAPTER 9
Normal Subgroups and Factor Groups
1. No, (13)(12)(13) 1 = (23) is not in H
2. For every α in Sn, αAnα 1 is even.
3. HR90 = R270H; DR270 = R90D; R90V = V R270
4. Solving (12)(13)(14) = α(12) for α we have α = (12)(13)(14)(12). Solving (1234)(12)(23) = α(1234) for α we have α = (234).
5. Say i < j and h ∈ H i ∩ H j. Then h ∈ H1 H2 · · · H j 1 ∩ H j = {e}.
6. No. Let A = 1 0 0 1 and B = 1 0 Then A is in H and B is in GL(2, R) but BAB 1 is not in H
7. H contains the identity so H is not empty. Let A, B ∈ H. Then det (AB 1 ) = (det A)(det B) 1 ∈ K. This proves that H is a subgroup. Also, for A ∈ H and B ∈ G we have det (BAB 1 ) = (det B)(det A)(det B) 1 = det A ∈ K so BAB 1 ∈ H
8. If k divides n, then ⟨k⟩/⟨n⟩ is a cyclic group of order n/k. So, it is isomorphic to Zn/k.
1 1 ∈ ∈ /∈
9. Let x G. If x H, then xH = H = Hx. If x H, then xH is the set of elements in G, not in H and Hx is also the elements in G, not in H.
10. a. α5H /= Hα5. b. It proves that coset multiplication is not a binary operation.
11. Let G = ⟨a⟩. Then G/H = ⟨aH⟩.
12. Note that aHbH = abH = baH = bHaH
13. in H.
14. 4
15. |9H| = 2; |13H| = 4.
16. 3
17. 8 + ⟨3.5⟩ = 1 + 7 + ⟨3.5⟩ = 1 + ⟨3.5⟩.
18. Zk
19. Observe that in a group G, if |a| = 2 and {e, a} is a normal subgroup, then xax 1 = a for all x in G Thus a ∈ Z(G). So, the only normal subgroup of order 2 in Dn is {R0, R180} when n is even.
20. Since |U (40)| = 16 and U5(40) = {1, 11, 21, 31} and U8(40) = {1, 9, 17, 33} we know that both factor groups have order 4. In U5 (40), 3U8 (40) has order 4, whereas all three non-identity elements in U (40)/U8 (40) have order 2.
21. Observe that (57) is in H, whereas (56)(57)(56) sends 6 to 7. Observe that (578) is in H, whereas (456)(57)(654) sends 6 to 7.
22. Note that (x, y) + ⟨(2, 2)⟩ = (x y, 0) + (y, y) + ⟨(2, 2)⟩ = (x y)(1, 0) + y(1, 1) + ⟨(2, 2)⟩ If y is even, then (y, y) + ⟨(2, 2)⟩ = ⟨(2, 2)⟩. If y is odd, then (y, y) + ⟨(2, 2)⟩ = (1, 1) + ⟨(2, 2)⟩ (Z ⊕ Z)/⟨(2, 2)⟩ ≈ Z ⊕ Z2; (Z ⊕ Z)/⟨(n, n)⟩ ≈ Z ⊕ Zn
23. It follows from Theorem 7.2 that H is the unique subgroup of order p in G Then, since for every x in G, we have that xHx 1 is a subgroup of order p, xHx 1 = H.
24. Z4 ⊕ Z2 To see that there is no element of order 8 in the factor group, observe that for any element (a, b) in Z4 ⊕ Z12 , (a, b)4 = (4a, 4b) belongs to {(0, 0), (0, 4), (0, 8)} ∈ ⟨(2, 2)⟩ So, the order of every element in the factor group divides 4. This rules out Z8. By observation, (1, 0)⟨(2, 2)⟩ has order 4, which rules out Z2 ⊕ Z2 ⊕ Z2.
25. Since the element (3H)4 = 17H /= H, |3H| = 8. Thus G/H ≈ Z8.
26. Z4 ⊕ Z2; ⟨17⟩ × ⟨41⟩
27. Since H and K have order 2, they are both isomorphic to Z2 and therefore isomorphic to each other. Since G/H = 4 and 3H = 4 we know that G/H Z4. On the other hand, direct calculations show that each of the three nonidentity elements in G/K has order 2, so G/K ≈ Z2 ⊕ Z2
28. Z2 ⊕ Z2; Z4.
29. Observe that nontrivial proper subgroups of a group of order 8 have order 2 or 4 and therefore are Abelian. Then use Theorem 9.6 and Exercise 4 of Chapter 8
31. Certainly, every nonzero real number is of the form ±r, where r is a positive real number. Real numbers commute and R + ∩ {1, 1} = {1}.
32. By Corollary 4 of Theorem 7.1, xmN = (xN )m = N , so xm ∈ N .
33. In the general case that G = HK there is no relationship. If G = H × K, then |g| = lcm(|h|, |k|), provided the |h| and |k| are finite. If |h| or |k| is infinite, so is |g|.
34. Since 1 = 3 · 5 + ( 2)7, m = (3m)5 + ( 2m)7. No, since 35 ∈ H ∩ K. Suppose H is any subgroup of index 2. Then, since |R∗/H| = 2, we have a2 ∈ H for all a ∈ R ∗ . This implies that R + ⊆ H If there is some a ∈ H with a < 0, then since a ∈ H we have a 1( a) = 1 ∈ H also. But this implies that R ∗ ⊆ H.
35. For the first question, note that ⟨3⟩ ∩ ⟨6⟩ = {1} and ⟨3⟩⟨6⟩ ∩ ⟨10⟩ = {1}. For the second question, observe that 12 = 3 162 so ⟨3⟩⟨6⟩ ∩ ⟨12⟩ /= {1}.
36. Certainly, R + has index 2. Suppose that H has index 2 and is not R + Then R∗/H = 2. So, for every nonzero real number a we have (aH)2 = a2H = H. Thus the square of every real number is in H. This implies that H contains all positive real numbers. Since H is not R+ , it must contain some negative real number a. But then H contains aR+ , which is the set of all negative real numbers. This shows that H = R∗ .
37. Say |g| = n. Then (gH)n = gnH = eH = H. Now use Corollary 2 to Theorem 4.1.
38. Observe that (1/n + Z) has order n.
39. Let x belong to G. Then xHx 1H = xx 1H = H so xHx 1 H. Alternate solution: Let x belong to G and h belong to H Then xhx 1H = xhHx 1)H = xHx 1H = xx 1H = eH = H, so xhx 1 belongs to H.
41. Suppose that H is a proper subgroup of Q of index n. Then Q/H is a finite group of order n. By Corollary 4 of Theorem 7.1 we know that for every x in Q we have nx is in H. Now observe that the function f (x) = nx maps Q onto Q. So, Q ⊆ H.
42. Let g G. Then there is an element b in G so that gH = (bH)2 = b2H. Thus g = b2h for some h H. But there is a c in H such that h = c2 So, g = (bc)2 . The proof is valid for any integer.
45. Normality follows directly from Theorem 4.3 and Example 7.
46. By Example 14 of Chapter 3, Z(D13) is the identity. Then by Theorem 9.4, D13 is isomorphic to Inn(D13).
47. By Lagrange, |H ∩ K| divides both 63 and 45. If |H ∩ K| = 9, then H ∩ K is Abelian by Theorem 9.7. If |H ∩ K| = 3, then H ∩ K is cyclic by the Corollary of Theorem 7.1. |H ∩ K| = 1, then H ∩ K = {e} In general, if p is a prime and |H| = p2m and |K| = p2n where gcd(m, n) = 1, then H ∩ K| = 1, p, or p2 . So by Corollary 3 of Theorem 7.1 and Theorem 9.7, H ∩ K is Abelian.
48. Use Example 3 in Chapter 8 and the G/Z Theorem (Theorem 9.3).
49. By Lagrange’s Theorem, Z(G) = 1, p, p2 , or p3 By assumption, Z(G) = 1 or p3 (for then G would be Abelian). So, Z(G) = p or p2 . However, the “G/Z” Theorem (Theorem 9.3) rules out the latter case.
50. Use the G/Z Theorem.
51. Suppose that K is a normal subgroup of G and let gH ∈ G/H and kH ∈ K/H . Then gHkH(gH) 1 = gHkHg 1H = gkg 1H ∈ K/H . Now suppose that K/H is a normal subgroup of G/H and let g ∈ G and k ∈ K Then gkg 1H = gHkHg 1H = gHkH(gH) 1 ∈ K/H so gkg 1 ∈ K
52. Say |aH| has finite order n. Then H = (aH)n = anH so that an is in H. But this implies that |an| and therefore |a| is finite. Thus aH = H
53. Say H has an index n Then (R∗)n = xn x R∗ H If n is odd, then (R∗)n = R∗; if n is even, then (R∗)n = R+ . So, H = R∗ or H = R+ .
54. a. Since 1 and 1 commutes with every element of the group, H is normal in G.
b. 1 i j k
1 1 i j k i i 1 k j
j j k 1 i
k k j i 1
Z2 ⊕ Z2
55. By Exercise 9, we know that K is normal in L and L is normal in D4. But V K = {V, R270}, whereas KV = {V, R90} So, K is not normal in D4
56. x(H N )x 1 = xHx 1 xNx 1 = H N . The same argument works for the intersection of any family of normal subgroups.
Alternate proof. Let a ∈ H ∩ K and x ∈ G Then, because H and K are normal in G, we have xax 1 ∈ H and xax 1 ∈ K. So xax 1 ∈ H ∩ K.
57. G has elements of orders 1, 2, 3, and 6.
58. By Example 5, NM is a subgroup. Also
xNM = (xN )M = (Nx)M = N (xM ) = N (Mx) = NMx.
59. Let H = ak be any subgroup of N = a Let x G and let (ak )m H We must show that x(ak )m x 1 H. Note that x(ak )m x 1 = x(akm )x 1 = (xax 1 )km = (ar)km = (ak )rm ak . (Here we used the normality of N to replace xax 1 by ar.)
60. Let H be the unique maximal subgroup of G and suppose that G is not normal in G. Let a be any element in G such that aHa 1 = H. Then aHa 1 is not a maximal subgroup of G Let K be a proper subgroup of G that properly contains aHa 1 Then, |aHa 1| < |K| < |G|. So, applying the inner automorophism φ(a 1 ) induced by a 1 to aHa 1 ⊂ K ⊂ G, we have H ⊂ a 1Ka ⊂ G, which is a contradiction.
61. gcd(|x|, |G|/|H|) = 1 implies gcd(|xH|, |G/H|) = 1. But |xH| divides G/H Thus |xH| = 1 and therefore xH = H
62. It follows from Theorem 7.2 that H is the only subgroup of G of order p and from Example 7 that H is normal.
63. If H and K are subgroups of order 3 and one of them is normal then HK is a subgroup of order 9. This contradicts Lagrange.
64. a. Observe that for any g in G,
g (x 1 y 1 xy )g 1 = (gx 1 g 1 )(gy 1 g 1 )(gxg 1 )(gyg 1 ) ∈ S.
b. Observe that xG′yG′ = yG′xG′ if and only if x 1 y 1 xy ∈ G′ .
c. Observe that xNyN = yNxN implies x 1 y 1 xyN = N Thus x 1 y 1 xy ∈ N
d. Let h ∈ H and g ∈ G Then ghg 1h 1 ∈ G′ ≤ H so that ghg 1 ∈ Hh = H
65. Observe that for every positive integer n, (1 + i)n is not a real number. So, (1 + i)R∗ has infinite order.
66. Let C be the collection of all subgroups of G of order n. Then, since |H| = n implies |xHx 1| = n, we have x(∩H ∈ C H)x 1 = ∩H ∈ C xHx 1 = ∩H ∈ C H.
67. Suppose that Aut(G) is cyclic. Then Inn(G) is also cyclic. So, by Theorem 9.4, G/Z is cyclic and from Theorem 9.3 it follows that G is Abelian. This is a contradiction.
68. Let x ∈ K. Then, |x| = pi for some i. So (xH)p i = xp i H = H and, because |G/H| = m, also we have (xH)m = H. Thus |xH| divides both m and pi . From gcd(p, m) = 1, we have |xH| = 1 and therefore xH = H This proves that K ⊆ H
69. Because g = 16 implies that gH divides 16, it suffices to show that (gH)4 is not H. Suppose that (gH)4 = g4H = H Then g4 is in H But then 2, g4 , and g8 are in H, which is a contradiction. In the general case, say |gH| = k. Then (gH)k = gkH = H. So, gk is in H and therefore |gk| = 1 or 2. It follows that k = 2n or n.
70. Suppose that H is a subgroup of S4 of order 12 distinct from A4. Then, Example 5 in this chapter and Theorem 7.2 HA4 = S4 and |HA4| = 12 · 12/|H ∩ A4| It follows that |H ∩ A4| = 6. But this contradicts Example 5 of Chapter 7.
71. If |Z(G)| = 5, then |G/Z(G)| = 6 and by Theorem 7.3 G is isomorphic to Z6 or D6. Theorem 9.3 rules out Z6. If |Z(G)| = 3, then |G/Z(G)| = 10 and by Theorem 7.3 G is isomorphic to Z10 or D5 Theorem 9.3 rules out Z10 If |G| = 2pq where p and q are distinct odd primes and |Z(G)| = p or q, then G/Z(G) is isomorphic to Dq or Dp, respectively.
72. By the classification of groups of order p2-theorem (9.3), G would be cyclic or isomorphic to Zp ⊕ Zp. But the G/Z-Theorem (9.3) rules out the cyclic case.
73. If A5 had a normal subgroup of order 2, then the subgroup has a nonidentity element that commutes with every element of A5 . An element of A5 of order 2 has the form (ab)(cd). But (ab)(cd) does not commute with (abc), which also belongs to A5 .
74. Since H has index 2 in G is it a normal subgroup of G and G/H = 2. It follows that for every a in G we have (aH)2 = H If a is an element of G of order 2n + 1, then H = a2n+1H = ((aH)2)naH = aH. Thus, a is in H.
75. By the classification of groups of order 2p-theorem (7.3), G/Z(G) would be cyclic or isomorphic to Dp. But the G/Z-Theorem (9.3) rules out the cyclic case for otherwise G would be Abelian and therefor G would be Z(G).
76. |9H| = 2; |17H| = 4.
77. Suppose that H is a normal subgroup of A5 of order 12. Since A5/H = 5 we know that for any of the 20 3-cycles α in A5 we have H = (αH) = α5H = α2H. So, α 1 = α2 ∈ H. Then α is also in H. But H only has 12 elements.
78. Because U (60) ≈ U (4) ⊕ U (3) ⊕ U (5) ≈ Z2 ⊕ Z2 ⊕ Z4 every element of U (60) satisfies the equations x4 = 1. So for all x ∈ U (60) we have would have (xH)4 = x4H = H.
79. Because 51H = H we have 153H = (3 51)H = 3(51H) = 3H.
80. By the 2p-theorem, G/Z(G) is isomorphic to Z14 or D7. The G/Z(G)-theorem rules of Z14 Since D7 has 7 elements of order 2, so does G/Z(G).
81. Let |gH| = d and |g| = m. We know by Exercise 37 that |gH| divides |g| and because gd ∈ H we also know that |gd| = m/d divides |H|. This means that m/d = 1.
82. Let H = U16(80) = {1, 17, 33, 49} Since none of 3, 32 , 33 is in H and 34 is in H the element 3H in the factor group U (80)/H has order 4. Thus ⟨3H⟩ is a subgroup of U (80)/H of order 4 and H ∪ 3H ∪ 9H ∪ 27H is a subgroup of U (80) order 16.
83. By definition, every element of G can be written in the form aj1 aj2 . . . ajk where aj1 , aj2 , , ajk ∈ ⟨a1 , a2 , , an ⟩ Then gH = aj1 Haj2 H, , ajk H
84. If n is odd then the n rotations of Dn form a normal subgroup of order n (see Exercise 9) that does not contain R180. So, D2n = Dn × {R0, R180}. If n is even, then the elements of Dn contains R180 and {R0, R180} is the only normal subgroup of order 2 in D2n (see Exercise 19.)
85. Since G is Abelian, the subgroups H1, H2, . . . , Hk are normal in G. By assumption, G = H1H2 · · · Hk, So, all that remains to prove is that for all i = 2, 3, , k 1 we have H1H2 · · · Hi ∩ Hi+1 = {e}. But if x ∈ H1H2 · · · Hi ∩ Hi+1 and x /= e, then x can be written in the two distinct forms h1h2 · · · hie · · · e (k i e terms) and e · · · ehi+1e · · · e (i e terms) on the left and k (i + 1) e terms on the right and each hj ∈ Hj. This contradicts our assumption about G.
86. eHe 1 = H. Let x, y N (H). Then (xy 1)H(xy 1) 1 = x(y 1Hy)x 1 = xHx 1 = H (Note that yHy 1 = H implies y 1 Hy = H.)
87. We know from Theorem 9.7 that G/Z(G) ≈ Zp2 or Zp ⊕ Zp and from the G/Z Theorem (Theorem 9.3) Zp2 is ruled out.
88. Theorem 9.3 would imply that G = Z(G).
89. By Theorem 7.2 and Example 5 in Chapter 9, if H and K were distinct subgroups of order p2, then HK would be a subgroup of order p3 or p4, which contradicts Lagrange.
91. If G is cyclic, then Theorem 4.4 says that G has exactly one element of order 2. If G is not cyclic, let a be any non-identity element of G and b be any element of G not in a Then a b is isomorphic to a group of the form Z2s Z2t where s and t are positive. But then G has at least three elements of order 2. The appropriate generalization is: “An Abelian group of order pn for a prime p and some positive integer n is cyclic if and only if it has exactly p 1 elements of order p.”
92. First note that the argument in Example 4 of Chapter 3 shows that Gm and Gn are subgroups of G and since G is Abelian they are normal in G. Because gcd(m, n) = 1, there are integers a and b such that am + bn = 1. By closure GmGn ⊆ G To show that G ⊆ GmGn let x ∈ G. Then we have that x = xam +bn = xam xbn , (xam )n = (xa )mn = (xa )|G | = e, and (xbn )m = (xb )mn = (xb )|G | = e So, x ∈ Gm Gn Finally, if x ∈ Gm ∩ Gn Then xm = e = xn and therefore |x| divides both m and n. So, |x| = 1.
93. Observe that for every two distinct primes p and q we have pH = qH. (For if there are integers a, b, c, d such that pa2 /b2 = qc2 d2 , then p occurs an odd number of times on the left side of pa2 d2 = qb2 d2 but an even number of times on the right side). Every nonidentity element in Q/H has order 2.
CHAPTER 10
Group Homomorphisms
1. Note that det (AB) = (det A)(det B).
2. Observe that |ab| = |a| |b|.
3. Note that (f + g)′ = f ′ + g ′
4. Let E denote any even permutation and O any odd. Observe that φ(EE) = φ(E) = 0 = 0+0 = φ(E)+φ(E). φ(EO) = φ(O) = 1 = 0+1 = φ(E)+φ(O).
The other cases are similar.
5. By property 3 of Theorem 10.1, k divides both m and n.
Now suppose that k divides both m and n and let d = gcd(m, n). Since k divides both m and n, it divides d, and because the subgroup n/d has order d, every element whose order divides d is contained in n/d So, k = tn/d for some integer t
To show that φ is well-defined we must show that if a = b mod m, then ka = kb mod n Since a b = mq for some integer q, we have ka kb = kmt = (tn/d)mt = tn(m/d)t = 0 mod n
To see why the number of homomorphisms from Zm to Zn is gcd(m, n) observe that the above argument shows that k yields a homomorphism if and only if k is in n/d and the order of n/d is d.
6. Recall, ∫ (f + g) = ∫ f + ∫ g. Kernel = {0}. No.
7. (σφ)(g1g2) = σ(φ(g1g2)) = σ(φ(g1)φ(g2)) =
(φ(g1))σ
=
)(
)(g2). It follows from Theorem 10.3 that |G/Ker φ| = |H| and |G/Ker σφ| = |K|. Thus, [Ker σφ : Ker ] = |Ker σφ/Ker φ| = |H|/|K|.
8. See Exercise 16 of Chapter 5 The kernel is the set of even permutations in G When G is Sn the kernel is An and from Theorem 10.3 we have that Sn/An is isomorphic to +1, 1 . So, An has index 2 in Sn and is normal in Sn The kernel is the subgroup of even permutations in G. If the members of G are not all even then the coset other than the kernel is the set of odd permutations in G. All cosets have the same size.
9. φ((g, h)(g ′ , h′)) = φ((gg ′ , hh′)) = gg ′ = φ((g, h))φ((g ′ , h′)). The kernel is {(e, h) | h ∈ H}.
10. See Exercise 9 of Chapter 1 The kernel is the subgroup of rotations in G If the members of G are not all rotations then the coset other than the kernel is the set of reflections in G. All cosets have the same size.
11. The mapping φ : Z ⊕ Z → Za ⊕ Zb given by φ((x, y)) = (x mod a, y mod b) is operation preserving by Exercise 9 in Chapter 0 If (x, y) ∈ Ker φ, then x ∈ ⟨a⟩ and y ∈ ⟨b⟩. So, (x, y) ∈ ⟨(a, 0)⟩ × ⟨(0, b)⟩. Conversely, every element in ⟨(a, 0)⟩ × ⟨(0, b)⟩ is in Ker φ So, by Theorem 10.3, Z ⊕ Z → Za ⊕ Zb is isomorphic to ⟨(a, 0)⟩ × ⟨(0, b)⟩
12. x x mod k is a homomorphism with kernel k So, Zn/ k is a cyclic group of order n/k.
13. (a, b) → b is a homomorphism from A ⊕ B onto B with kernel A ⊕ {e}. So, by Theorem 10.3, (A ⊕ B)/(A ⊕ {e}) ≈ B
14. Observe that since 1 has order 12, φ(1) = 3 must divide 12. But in Z10, 3 = 10. Alternate proof. Observe that φ(6 + 7) = φ(1) = 3 while φ(6) + φ(7) = 8 + 1 = 9. Second alternate proof. Observe that {0, 6} is a subgroup of Z12 but φ({0, 6}) = {0, 8} is not a subgroup of Z10.
15. By property 6 of Theorem 10.1, we know φ 1(9) = 23 + Ker φ = {23, 3, 13}.
16. Observe that such a mapping would be an isomorphism and isomorphisms preserve orders of elements.
17. Suppose φ is such a homomorphism. By Theorem 10.3, Ker φ = 2. Let φ(1, 0) = (a, b). Then φ(4, 0) = 4φ(1, 0) = 4(a, b) = (4a, 4b) = (0, 0). But then Ker φ contains an element of order 4.
Alternate proof. Suppose φ is such a homomorphism and let H = Ker φ. By Theorem 10.3, (Z16 ⊕ Z2)/H ≈ Z4 ⊕ Z4 Thus every element of (Z16 ⊕ Z2)/H has order 1, 2 or 4 and |H| = 2. Then ((1, 0)H)4 = (4, 0)H = H implies that (4, 0) ∈ H. But (4,0) has order 4 whereas |H| = 2.
18. No, because of part 3 of Theorem 10.1. No, because the homomorphic image of a cycle group must be cyclic. Yes, φ(x) = (x mod 3, x mod 2) is a homomorphism.
19. Since |Ker φ| is not 1 and divides 17, φ is the trivial map.
20. 0 onto Z8; 4 to Z8.
21. By Theorem 10.3 we know that |Z30/Ker φ| = 5. So, |Ker φ| = 6. The only subgroup of Z30 of order 6 is ⟨5⟩.
22. Let φ(g) = 8. By Theorem 10.1 part 3, g = 8k. Then gk = 8. To generalize replace 8 by n
23. |φ 1(H)| = |H||Ker φ|.
24.
a. Let φ(1) = k. Then φ(7) = 7k mod 15 = 6 so that k = 3 and φ(x) = 3x.
b. ⟨3⟩.
c. ⟨5⟩
d. 4 + ⟨5⟩.
25. To define a homomorphism from Z20 onto Z10 we must map 1 to a generator of Z10 . Since there are four generators of Z10 we have four homomorphisms. (Once we specify that 1 maps to an element a, the homomorphism is x → xa.) To define a homomorphism from Z20 to Z10 we can map 1 to any element of Z10 . (Be careful here, these mappings are well defined only because 10 divides 20.)
26. There are four: x → (x mod 2, 0); x → (0, x mod 2); x → (x mod 2, x mod 2); x → (0, 0).
27. If φ is a homomorphism from Zn to Zn with φ(1) = k, then by property 2 of Theorem 10.1, φ(x) = kx Moreover, for each k with 0 k n 1, the mapping φ(x) = kx is a homomorphism.
28. Ker φ = A4. The trivial homomorphism and the one given in Example 11 are the only homomorphisms. To verify this observe that by Theorem 10.3, |Ker φ| = 12 or 24. If |Ker φ| = 12 one possibility for Ker φ is A4. If H is another one not A4, then since A4 is normal in S4, HA4 is a subgroup of S4 of order greater than 12. So, by Theorem 7.2, |HA4| = 12 · 12/|H ∩ A4| = 24, which implies that H ∩ K is a subgroup of A4 of order 6. But Example 5 in Chapter 7 rules that out.
29. Say the kernel of the homomorphism is K By Theorem 10.3, G/K ≈ Z10 So, |G| = 10|K| In Z10, let H = ⟨2⟩ By properties 5, 7, and 9 of Theorem 10.2, φ 1(H) is a normal subgroup of G of order 2|K|. So, φ 1(H) has index 2. To show that there is a subgroup of G of index 5, use the same argument with H = ⟨5⟩ If there is a homomorphism from a finite group G onto Zn, then the same argument shows that G has a normal subgroup of index d for any divisor D of n.
30. Z6 Z2 has normal subgroups of orders 1, 2, 3, 4,6, and 12. So by parts 5 and 9 of Theorem 10.2, G has normal subgroups of orders 5, 10, 15, 20, 30, and 60.
31. Suppose that the correspondence from Zm to Zn given by x → kx is a homomorphism. Then, in Zn, we have 0 = φ(0) = φ(m) = mk So, n divides mk Now suppose n divides mk We first show that φ is well-defined. That is, if a = b mod m, we must show that φ(a) = φ(b) mod n. From a = b mod m, we have a b = mt for some integer t Then, φ(a) φ(b) = ka kb = k(a b) = kmt = 0 mod n. Finally, observe that for a and b in Zm , we have that in Zn, φ(a + b) = k(a + b) = ka + kb = φ(a) + φ(b).
33. By property 6 of Theorem 10.1, φ 1(11) = 11Ker φ = {11, 19, 27, 3}.
34. First observe from Table 5.1 that A4 has three elements of order 2 and eight of order 3. Thus, by property 3 of Theorem 10.1 the orders of the elements of G must be 1, 2, or 3. So, since G is cyclic, |G| = 1, 2, or 3. If |G| = 2, then |Ker φ| = 6. This means that for every element α of order 3, we have that |φ(α)| must divide both 2 and 3 and therefore φ(α) = ǫ. But then Ker φ would contain all 8 elements of A4 of order 3, which is a contradiction.
35. φ((a, b) + (c, d)) = φ((a + c, b + d)) = (a + c) (b + d) = (a b) + (c d) = φ((a, b)) + φ((c, d)). Ker φ = {(a, a) | a ∈ Z} φ 1(3) = {(a + 3, a) | a ∈ Z}
36. 4 a 4b.
37. Consider the mapping φ f r√o m C∗ onto R+ , given by φ(x) = |x|. (Recall from Chapter 0 that |a + bi| = a2 + b2 .) By straightforward algebra we have xy = x y . Thus φ is a homomorphism with Ker φ = H. So, by Theorem 10.3, C∗/H is isomorphic to R+
| | | || |
38. Ker γ = Ker α ⊕ Ker β.
39. φ(xy) = (xy)6 = x6 y6 = φ(x)φ(y). Ker φ = ⟨cos 60◦ + i sin 60◦⟩
40. ⟨12⟩; ⟨12⟩; in general, the kernel is ⟨lcm(m, n)⟩.
41. Since φ(e) = e = e 2 , e ∈ H. If a ∈ H, then φ(ab) = φ(a)φ(b) = a 2 b 2 = (ab) 2 ∈ H. Also, φ(a 1) = φ(a) 1 = (a 2) 1 = (a 1) 2 ∈ H If φ(x) = x3 and a ∈ H, then φ(a) = a3 = a 2 implies that a5 = e. Thus, |a| = 5 or 1.
42. Because |D6/Ker φ| = |D3| = 6, we know |Ker φ| = 2. Since {R0, R180} is the only normal subgroup of D6 of order 2, it is Ker φ
43. Property 2 of Theorem 10.2 handles the 2m case. Suppose that there is a homomorphism from G = Z2m Z2n onto Z2 Z2 Z2 where m and n are at least 1 and let H be the kernel. We may assume that m + n 3. Then H = 2m +n 3 . Because every nonidentity element of G/H has order 2, we know that ((1, 0)H)2 = (2, 0)H = H and ((0, 1)H)2 = (0, 2)H = H. This means that H1 = (2, 0) and H2 = (0, 2) are subgroups of H Then H1H2 is also a subgroup of H. But H1H2 = 2m 1 2n 1 = 2m +n 2 exceeds H = 2m +n 3 . This argument works for any prime p.
44. If there were a homomorphism φ from S3 onto Z3 , then φ would be a 2-1 mapping. But, because Z3 has 2 elements of order 3, S3 would have to have four elements of order 3, which it does not. For S4 the mapping would be 8-1. Then S4 would have to have 16 elements of order 3 to map only the 2 elements in Z3 of order 3. But it does not.
45. Let H be the normal subgroup of order 4 defined in Example 9. Then S4/H is a group of order 6 but has no element of order 6 (because S4 does not have one). So, it follows from Theorem 7.3, S4/H is isomorphic S3.
46. Let φ be a homomorphism of Zm ⊕ Zn Then φ(Zm Zn) = φ( (1, 0) (0, 1) ) = φ( (1, 0) ) φ( (0, 1) ). Now use property 3 of Theorem 10.1.
47. It follows from Exercise 11 in Chapter 0 that the mapping φ from U (st) to U (s) given by φ(x) = x mod s is a homomorphism. Since Ker φ = Us(st) we have by Theorem 10.3 that U (st)/Us(st) is isomorphic to a subgroup of U (s). To see that φ is onto, note that it follows from Theorem 8.3 that |U (st)/Us(st)| = |U (st)|/|Us(st)| = |U (s) ⊕ U (t)|/|U (t)| = |U (s)||U (t)|/|U (t)| = |U (s)|.
48. Observe that every element can be written in the form sn 1 sn 2 . . . sn m where 1 2 m s1s2, . . . , sm ∈ S and the exponents are integers. So, φ(g) = φ(s1)n1 φ(s2)n2 · · · φ(sm)nm ∈ φ(S).
49. Consider the mapping φ from K to KN/N given by φ(k) = kN . Since φ(kk′) = kk′N = kNk′N = φ(k)φ(k′) and kN ∈ KN/N , φ is a homomorphism. Moreover, Ker φ = K ∩ H. So, by Theorem 10.3, K/(K ∩ N ) ≈ KN/N .
50. Show that the mapping from G/N to G/M given by gN gM is an onto homomorphism with kernel M/N
51. Since the eight elements of A4 of order 3 must map to an element of order that divides 3, by Lagrange’s Theorem, each of them must map to the identity. But then the kernel has at least 8 elements and its order and divides 12. So, the kernel has order 12.
52. Uk(n) is the kernel.
53. Let N be a normal subgroup of D4 By Lagrange’s Theorem, the only possibilities for |N | are 1, 2, 4, and 8. By Theorem 10.4, the homomorphic images of D4 are the same as the factor groups D4/N of D4 When |N | = 1, we know N = {e} and D4/N ≈ D4. When |N | = 2, then N = {R0, R180}, since this is the only normal subgroup of D4 of order 2, and D4/N ≈ Z2 ⊕ Z2 because D4/N is a group of order 4
with three elements of order 2. When |N | = 4, |D4/N | = 2 so D4/N ≈ Z2. When |N | = 8, we have D4/N ≈ {e}.
54. Use Theorem 10.4 and part 3 of Theorem 10.1.
55. It is divisible by 10. In general, if Zn is the homomorphic image of G, then G is divisible by n
56. It is divisible by 30. In general, the order of G is divisible by the least common multiple of the orders of all its homomorphic images.
57. It is infinite. Z is an example.
58. Let A be the coefficient matrix of the system. If A is an n m matrix, then matrix multiplication by A is a homomorphism from Rm into Rn whose kernel is S.
59. Let γ be a natural homomorphism from G onto G/N . Let H be a subgroup of G/N and let γ 1(H) = H. Then H is a subgroup of G and H/N = γ(H) = γ(γ 1 (H)) = H.
60. Use Theorem 10.1, part 2.
61. The mapping g → φg is a homomorphism with kernel Z(G).
62. a. Since 4 = |Z2 ⊕ Z2| does not divide |D5|, there are none. b. There are four. In addition to the trivial homomorphism, we can map all rotations to the identity and all reflections to any one of the three elements of order 2.
63. Since (f + g)(3) = f (3) + g(3), the mapping is a homomorphism. The kernel is the set of elements in Z[x] whose graphs pass through the point (3, 0). 3 can be replaced by any integer.
64. For the first part, use trig identities. The kernel is ⟨2π⟩.
65. Let g belong to G Since φ(g) belongs to Z2 Z2 = (1, 0) (0, 1) (1, 1) , it follows that G = φ 1( (1, 0) ) φ 1( (0, 1) ) φ 1( (1, 1) ). Moreover, each of these three subgroups is proper since φ is onto and each is normal by property 8 of Theorem 10.2.
66. Try g → (gH, gK).
67. Map (a, b) to (a mod 4, b).
68. Since φ(Z(D12)) ⊆ Z(D3) = {R0}, we know φ(R180) = R0.
69. It fails because 5 does divide |Aut(Z11)| = 10.
70. The argument in Example 18 shows that G has a unique subgroup H of order 11 and a subgroup K of order 5. If K was a normal subgroup, then both H and K would be normal (see Example 7 in Chapter 9) and we would have G = H K H K Z11 Z5 Z55 , which is a contradiction. Let a H and a = e Since K N (K) and N (K) = G, we know a is not in N (K). We claim that the subgroups aiKa i for i = 0, 1, 2, . . . , 10 are distinct. To verify this, suppose that i, j ∈ {0, 1, 2, . . . , 10} with i > j and aiKa i = ajKa j . Then ai jKa i j = K. This means that ai j ∈ N (K). But then a ∈ N (K) because ai j /= e and |a| = 11 implies that ⟨ai j ⟩ = ⟨a⟩.
71. Mimic Example 18.
72. There are no homomorphisms from Z onto S3 since the image of a cyclic group must be cyclic. For each element x in S3 the mapping from Z to S3 given by φ(n) = xn is a homomorphism. It follows from part 2 of Theorem 10.2 that there are no others.
73. Let φ be a homomorphism from S3 to Zn Since |φ(S3)| must divide 6, we have that φ(S3) = 1, 2, 3, or 6. In the first case, φ maps every element to 0. If φ(S3) = 2, then n is even and φ maps the even permutations to 0, and the odd permutations to n/2. The case that |φ(S3)| = 3 cannot occur because it implies that Ker φ is a normal subgroup of order 2 whereas S3 has no normal subgroup of order 2. The case that φ(S3) = 6 cannot occur because it implies that φ is an isomorphism from a non-Abelian group to an Abelian group.
74. Suppose that φ is a homomorphism and m = nq + r where 1 r < n. Then 0 = φ(0) = φ(nq + r) = φ(nq) + φ(r) = r = 0. This contradiction shows that for φ to be a homomorphism it is necessary that n divides m. If n divides m, then φ is well-defined and is operation preserving from Exercise 11 of Chapter 0.
75. φ(zw) = z2w2 = φ(z)φ(w). Ker φ = {1, 1} and, because φ is onto C∗ , we have by Theorem 10.3, that C∗/{1, 1} is isomorphic to C∗ . If C∗ is replaced by R∗ , we have that φ is onto R + and by Theorem 10.3, R∗/{1, 1} is isomorphic to R+
76. p2 To verify this, note that for any homomorphism φ from Zp Zp into Zp we have φ(a, b) = aφ(1, 0) + bφ(0, 1). Thus we need only count the number of choices for φ(1, 0) and φ(0, 1). Since p is prime, we may let φ(1, 0) be any element of Zp The same is true for φ(0, 1).
77. If a homomorphism φ maps 1 to k, then φ(x) = kx. Thus, by property 3 of Theorem 10.1, a must be a divisor of 5. This gives us that the possibilities for k are 0, 3, 6, 9, and 12. These five possibilities for k can be checked with one argument by noting that they all have the form 3s and proceed as in Example 11.
78. Say φ is such a homomorphism. Then, in all three cases, for any x we have φ(x) = φ(x/2 + x/2) = φ(x/2) + φ(x/2) = 2φ(x/2) = 0 mod 2. This proof works for Zn where n is even, since φ(x) = n(x/n) = 0 mod n
79. First note that by by the Classification of Group of order 2p, where p is an odd prime (Theorem 7.3), we have that S3 is isomorphic to D3 By properties of cosets, |S4/H| = 6 and by Theorem 7.3, we have that S4/H is isomorphic to Z6 or D3. Because the largest order of any element in S4 is 4, S4/H is isomorphic to D3
80. First note that the orders of the elements in S4 are 1, 2, 3, and 4 and that the number of elements of order 3 is 4 3 2/3 = 8. This means that the number of elements of orders 1, 2 or 4 is 16. Suppose that H is a normal subgroup of S4 of order 6. Then there is a homomorphism φfrom S4/H onto a group of order 4. But, for any element α in S4 of order 3 we know that φ(α) must divide both 3 and 4. So, H would contain all 8 elements of order 3. Now suppose that S4 has a normal subgroup H of order 8. Then there is a homomorphism φ from S4/H onto a group of order 3. But, for any element α in S4 of order 1, 2 or 4 we know that φ(α) must divide both 3 and 4. So, φ(α) = 1 and H would contain all 16 elements of orders 1, 2, or 4.
81. We know that φ(1) must divide 24 and 9. So, φ(1) = 1 or 3. This tells us that φ(1) = 0, 3, or 6.
5. The only Abelian groups of order 45 are Z45 and Z3 ⊕ Z3 ⊕ Z5. In the first group, 3 = 15; in the second one, (1, 1, 1) = 15. Z3 Z3 Z5 does not have an element of order 9.
| | | | ⊕ ⊕
6. Z27 ⊕ Z4; Z27 ⊕ Z2 ⊕ Z2.
7. In order to have exactly four subgroups of order 3, the group must have exactly 8 elements of order 3. When counting elements of order 3 we may ignore the components of the direct product that represent the subgroup of order 4 since their contribution is only the identity. Thus, we examine Abelian groups of order 27 to see which have exactly 8 elements of order 3. By Theorem 4.4, Z27 has exactly 2 elements of order 3; Z9 ⊕ Z3 has exactly 8 elements of order 3 since for |(a, b)| = 3 we can choose |a| = 1 or 3 and |b| = 1 or 3, but not both |a| and |b| of order 1; in Z3 ⊕ Z3 ⊕ Z3, every element except the identity has order 3. So, the Abelian groups of order 108 that have exactly four subgroups of order 3 are Z9 ⊕ Z3 ⊕ Z4 and Z9 ⊕ Z3 ⊕ Z2 ⊕ Z2. The subgroups of Z9 ⊕ Z3 ⊕ Z4 of order 3 are ⟨(3, 0, 0)⟩, ⟨(0, 1, 0)⟩, ⟨(3, 1, 0)⟩ and ⟨(3, 2, 0)⟩ The subgroups of Z9 ⊕ Z3 ⊕ Z2 ⊕ Z2 of order 3 are ⟨(3, 0, 0, 0)⟩, ⟨(0, 1, 0, 0)⟩, ⟨(3, 1, 0, 0)⟩ and ⟨(3, 2, 0, 0)⟩.
8. Z3 ⊕ Z3 ⊕ Z3 ⊕ Z4; Z3 ⊕ Z3 ⊕ Z3 ⊕ Z2 ⊕ Z2
9. Elements of order 2 are determined by the factors in the direct product that have order a power of 2. So, we need only look at Z8, Z4 ⊕ Z2 and Z2 ⊕ Z2 ⊕ Z2. By Theorem 4.4, Z8 has exactly one element of order 2; Z4 ⊕ Z2 has exactly three elements of order 2; Z2 ⊕ Z2 ⊕ Z2 has exactly 7 elements of order 2. So, G ≈ Z4 ⊕ Z2 ⊕ Z3 ⊕ Z5
11. By the Fundamental Theorem, any finite Abelian group G is isomorphic to some direct product of cyclic groups of prime-power order. Now go across the direct product and, for each distinct prime you have, pick off the largest factor of that prime-power. Next, combine all of these into one factor (you can do this, since their orders are relatively prime). Let us call the order of this new factor n1 Now repeat this process with the remaining original factors and call the order of the resulting
factor n2 . Then n2 divides n1 , since each prime-power divisor of n2 is also a prime-power divisor of n1 Continue in this fashion. Example: If
then
Now note that 2 divides 3 · 25 · 2 and 3 · 25 · 2 divides 27 · 125 · 4.
12. By the corollary to the Fundamental Theorem of Finite Abelian Groups, the given group has a subgroup of order 10. But this group must be isomorphic to Z2 ⊕ Z5 ≈ Z10.
13. Z2 ⊕ Z2
14. If G is an Abelian group of order n and m is a divisor of n, then G has a cyclic subgroup of order m if m is square-free (i.e., each prime factor of m occurs to the 1st power only).
15. a. 1 b. 1 c. 1 d. 1 e. 1 f. There is a unique Abelian group of order n if and only if n is not divisible by the square of any prime.
16. a. same b. same c. same d. same e. twice as many of order m compared with the number of order n
17. This is equivalent to asking how many Abelian groups of order 16 have no element of order 8. From the Fundamental Theorem of Finite Abelian Groups, the only choices are Z4 ⊕ Z4, Z4 ⊕ Z2 ⊕ Z2, and Z2 ⊕ Z2 ⊕ Z2 ⊕ Z2.
18. 5n
19. The symmetry group is {R0, R180, H, V }. Since this group is Abelian and has no element of order 4, it is isomorphic to Z2 ⊕ Z2
20. Consider every possible isomorphism class one by one and show each has the desired subgroup. For instance, in Z2 ⊕ Z2 ⊕ Z2 ⊕ Z27 ⊕ Z5 the subgroup is {(a, b, 0, c, d) | a, b ∈ Z2, c ∈ ⟨3⟩, d ∈ Z5}
21. Because the group is Abelian and has order 9, the only possibilities are Z9 and Z3 ⊕ Z3. Since Z9 has exactly 2 elements of order 3 and 9, 16, and 22 have order 3, the group must be isomorphic to Z3 ⊕ Z3.
22. Because of the Fundamental Theorem and Corollary 1 of Theorem 8.2, we may assume |G| is a prime-power. Let x be an element of G of maximum order. Then for any y in G we have |y| divides |⟨x⟩|. Since ⟨x⟩ has a unique subgroup for each divisor of |⟨x⟩|, it follows that ⟨y⟩ ⊆ ⟨x⟩
23. By the Corollary of Theorem 8.2, n must be square-free (no prime factor of n occurs more than once).
24. n = p2p2 or p2p2p3p4 pk where k ≥ 3 and p1, p2, . . . , pk are distinct primes.
25. Among the first 11 elements in the table, there are 9 elements of order 4. None of the other isomorphism classes has this many.
26. Z4 ⊕ Z2; one internal direct product is ⟨7⟩ × ⟨17⟩. Some others are ⟨7⟩ × ⟨65⟩ and ⟨23⟩ × ⟨65⟩.
27. First observe that G is Abelian and has order 16. Now we check the orders of the elements. Since the group has 8 elements of order 4 and 7 of order 2, it is isomorphic to Z4 ⊕ Z2 ⊕ Z2. One internal direct product is ⟨7⟩ × ⟨101⟩ × ⟨199⟩.
28. Z2 ⊕ Z2 ⊕ Z3; One internal direct product is ⟨19⟩ × ⟨26⟩ × ⟨31⟩
29. Since |⟨(2, 2)⟩| = 8, we know |(Z16 ⊕ Z16)/⟨(2, 2)⟩| = 32. Because the largest order of any element in Z16 ⊕ Z16 is 16, the same is true about the factor group. Then observing that |(1, 0) + ⟨(2, 2)⟩| = 16, we know that the isomorphism class is Z16 ⊕ Z2.
30. Z4 ⊕ Z2 ⊕ Z4
31. Since Z9 has exactly 2 elements of order 3, once we choose 3 nonidentity elements we will either have at least one element of order 9 or 3 elements of order 3. In either case we have determined the group. The Abelian groups of order 18 are Z9 Z2 Z18 and Z3 Z3 Z2 By Theorem 4.4, Z18 group has 6 elements of order 18, 6 elements of order 9, 2 of order 6, 2 of order 3, 1 of order 2, and 1 of order 1. Z3 Z3 Z2 has 8 elements of order 3, 8 of order 6, 1 of order 2, and 1 of order 1. The worst-case scenario is that at the end of 5 choices we have selected 2 of order 6, 2 of order 3, and 1 of order 2. In this case we still have not determined which group we have. But the sixth element we select will give us either an element of order 18 or 9, in which case we know the group Z18 or a third element of order 6 or 3, in which case we know the group is Z3 ⊕ Z3 ⊕ Z2
32. The element of order 8 rules out all but Z16 and Z8 ⊕ Z2 and two elements of order 2 precludes Z16
33. If a2 /= b2 , then a /= b and a /= b3 . It follows that ⟨a⟩ ∩ ⟨b⟩ = {e}. Then G = ⟨a⟩ × ⟨b⟩ ≈ Z4 ⊕ Z4.
{ ∈ | }
34. Observe that the subgroup H = x G x2 = e consist exactly of the elements of G of order 2 and e. So, it follows for the corollary of Theorem 11.1 that H has the form 2n for some positive integer n and that the number of elements of order 2 is 2n 1. This argument shows that for any finite Abelian group the number of elements of order p, where is a prime, has the form pn 1.
35. By Theorem 11.1, we can write the group in the form Zp1 n1 ⊕ Zp2 n2 ⊕ · · · ⊕ Zpknk where each pi is an odd prime. By Theorem 8.1 the order of any element (a1, a2, . . . , ak) = lcm(|a1|, |a2|, . . . , |ak|). And from Theorem 4.3 we know that |ai| i
divides pn i , which is odd.
36. Z2 ⊕ Z2 ⊕ ⊕ Z2 (n terms).
37. By Theorem 7.2 we have, |⟨a⟩K| = |a||K|/|⟨a⟩ ∩ K| = |a||K| = |a||K|p = |G|p = |G|
38. If G = pn , use the Fundamental Theorem and Theorem 9.6. If every element has order a power of p, use the corollary to the Fundamental Theorem.
39. By the Fundamental Theorem of Finite Abelian Groups, it suffices to show that every group of the form Z n1 ⊕ Z n2 ⊕ · · · ⊕ Z nk is a subgroup of a U -group. Consider first a group of the form Z n1 ⊕ Z n 2 (p1 and p2 need not be distinct). By Dirichlet’s Theorem, for some s and t there are distinct primes q and r such that q = tpn1 + 1 and r = spn2 + 1. Then U (qr) = U (q) ⊕ U (r) ≈ Z n1 ⊕ Z n2 , and this
latter group contains a subgroup isomorphic to Z n1 ⊕ Z n2 The general case follows in the same way.
40. Observe that p1 p2
41. It follows from Exercise 4 of Chapter 8 and Theorem 9.6 that if D4 could be written in the form ⟨a⟩ × K where |a| = 4, it would be Abelian.
42. If is an Abelian group G of order 2n is not cyclic, then by Theorem 11.1, G is isomorphic to a group of the form Z2 m 1 ⊕ Z2 m 2 ⊕ · · · ⊕ Z2 m k where k ≥ 2 and each mi is at least 1. But the subgroup Z2 m 1 Z2m2 0 0 has three elements of order 2. The other half follows from Theorem 4.4. In general, an Abelian group of order pn is cyclic if and only if it has exactly p 1 elements of order p
43. If G has an element of order greater than 2, then φ(x) = x 1 is a non-trivial automorphism of G (see Exercise 12 of Chapter 6). If not, then |G| = 2n and is G isomorphic to Z2 ⊕ Z2 ⊕ · · · ⊕ Z2 (n terms). Then φ(x1, x2, x3, . . . , xn) = φ(x2, x1, x3, . . . , xn) is an automorphism of G.
44. This follows directly from Exercise 11. An example for the second part is D3
45. By Theorem 11.1 and Corollary 1 of Theorem 8.2, it suffices to do the case where |G| = pm and p is prime. By Theorem 11.1, if G is not cyclic, then G is isomorphic to a group of the form Zpm1 ⊕ Zpm2 ⊕ ⊕ Zpmk where k ≥ 2 and each mi is at least 1. But then, by Theorem 4.3, G has a subgroup of the form Zp Zp Zp of order pk and every element of this subgroup is a solution to xp = e
46. By the Fundamental Theorem of Finite Abelian groups, G can be written as an internal direct product of the form A B where A is a power of 2. Then, for any nonidentity elements a A and any element b B we know that ab has even order. This gives us H ( A 1)B ( A /2) B = A B /2 = G /2 elements of even order. Since H also contains the identity, H = G
47. First, observe by direct calculations we have that |8| = |12| = |18| = |21| = |27| = 4. Since for all x in G we also have |x| = | x| = |65 x|, we know that G has at least 10 elements of order 4. By Theorem 4.4, Z18 has only 2 elements of order 4 and by Theorem 8.1 Z2 ⊕ Z2 ⊕ Z2 ⊕ Z2 has none, so these two groups are eliminated. Finally, arguing as in Examples 5 and 6 in Chapter 8 we know that Z8 ⊕ Z2 has only 4 elements of order 4 and Z4 ⊕ Z2 ⊕ Z2 has only 8. So, G ≈ Z4 ⊕ Z4.
48. Zp1n1 ⊕ Zp2n2 ⊕ · · · ⊕ Zpknk where the pi terms are distinct primes.
CHAPTER 12
Introduction to Rings
1. For any n > 1, the ring M2(Zn) of 2 × 2 matrices with entries from Zn is a finite noncommutative ring. The set M2(2Z) of 2 2 matrices with even integer entries is an infinite noncommutative ring that does not have a unity.
×
2. 6 is the unity; 2 8 = 6, 4 4 = 6.
3. In R, {n √ 2| n ∈ Z} is a subgroup but not a subring. Another example is the ring M2 (Z) and the subgroup of all elements with the entry 0 in the upper left corner.
4. In Z4, 2x = 2 has solutions 1 and 3. In a group, x = a 1b.
5. a and b
6. Consider Z6 or Z10
7. First observe that every nonzero element a in Zp has a multiplicative inverse a 1 For part a, if a = 0, then a2 = a implies that a 1a2 = a 1a and therefore a = 1. For part b, if a = 0, then ab = 0 implies that b = a 1(ab) = a 10 = 0. For part c, ab = ac implies that a 1(ab) = a 1(ac). So b = c.
8. Consider aba = aba.
9. If a and b belong to the intersection, then they belong to each member of the intersection. Thus a b and ab belong to each member of the intersection. So, a b and ab belong to the intersection.
10. Observe that all the sets in the examples are closed under subtraction and multiplication.
11. Rule 3: 0 = 0( b) = (a + ( a))( b) = a( b) + ( a)( b) = (ab) + ( a)( b). So, ab = ( a)( b).
Rule 4: a(b c) = a(b + ( c)) = ab + a( c) = ab + ( (ac)) = ab ac
Rule 5: By Rule 2, ( 1)a = 1( a) = a. Rule 6: By Rule 3, ( 1)( 1) = 1 1 = 1.
12. If c = db, then c = d(a 1 )ab = (da 1 )ab. If c = (ab)d, then c = (ad)b.
13. Let S be any subring of Z By definition of a ring, S is a subgroup under addition. By Theorem 4.3, S = ⟨k⟩ for some integer k.
14. Use induction.
15. If m or n is 0, the statement follows from part 1 of Theorem 12.1. For simplicity, for any integer k and any ring element x we will use kx instead of k x Then for positive m and n, observe that (ma)(nb) = (a + a + · · · + a) + (b + b + · · · + b) = (ab + ab + · · · + ab), where the terms a + · · · + a, b + b + · · · + b, and the last term have mn summands.
For the case that m is positive and n is negative, we first observe that nb means ( b) + ( b) + + ( b) = ( n)( b). So, nb + ( n)b = (( b) + ( b) + + ( b)) + (b + b + + b) = 0. Thus, 0 = (ma)(nb + ( n)b) = (ma)(nb) + (ma)( n)b = (ma)(nb) + m( n)ab = (ma)(nb) + ( (mn))ab So, adding (mn)ab to both ends of this string of equalities gives (mn)ab = (ma)(nb).
For the case when m is negative and n is positive, just reverse the roles of m and n is the preceding argument. If both m and n are negative, note that (ma)(nb) = (( a) + ( a) + · · · + ( a))(( b) + ( b) + · · · + ( b)) = (( m)( a))(( n)( b)) = ( m)( n)(( a)( b)) = (mn)(ab).
16. Observe that n · ( a) + n · a = 0.
17. From Exercise 15, we have (n · a)(m · a) = (nm) · a2 = (mn) · a2 = (m · a)(n · a).
18. r0 = 0, so S is not empty. Let a, b S. Then r(a b) = ra rb = 0 0 = 0. Also r(ab) = (ra)b = 0b = 0.
19. Let a, b belong to the center. Then (a b)x = ax bx = xa xb = x(a b). Also, (ab)x = a(bx) = a(xb) = (ax)b = (xa)b = x(ab).
20. ⟨1⟩ ⊃ ⟨2⟩ ⊃ ⟨4⟩ ⊃ ⟨8⟩ ⊃ · · · .
21. (x1 , . . . , xn )(a1 , . . . , an ) = (x1 , . . . , xn ) for all xi in Ri if and only if xiai = xi for all xi in Ri and i = 1, . . . , n, and xiai = xi for all xi in Ri if and only if xi is a unity of Ri
22. By the One-Step Test, we must show ab 1 is a unit wherever a and b are. But ab 1 ba 1 = 1.
23. By observatio n ±1 and ± i are units. To see that there are no others, note that (a + bi) 1 = 1 = 1 a bi = a bi . But a is an integer only when a2 + b2 = 1 a+bi a+bi a bi a2 +b2 a2 +b2 and this holds only when a = ±1 and b = 0 or a = 0 and b = ±1.
24. Say ei is the unity of Ri Then (a1, , an)(b1, , bn) = (e1, , en) if and only if aibi = ei for i = 1, , n That is, if and only if ai ∈ U (Ri).
25. Note that the only f (x) ∈ Z[x] for which 1/f (x) is a polynomial with integer coefficients are f (x) = 1 and f (x) = 1.
26. {f (x) = c | c ∈ R, c /= 0}.
27. If a is a unit, then b = a(a 1b).
28. In Z6, 4 · 2 = 2; in Z8, 3 · 5 = 7, in Z15, 9 · 3 = 12.
29. Note that (a + b)(a 1 a 2b) = 1 a 1b + ba 1 a 2b2 = 1.
30. (1 a)(1+a+a2+ +an 1) = 1+a+a2+ +an 1 a a2 an = 1 an = 1 0 = 1.
31. 01 = 0 so the set is nonempty. Let am = 0 and bn = 0. We may assume that m n. Then in the expansion of (a b)2m each term has an expression of the form a2m ibi So when i = 0, 1, . . . , m we have a2m i = 0 and when i = m + 1, m + 2, m + m we have bi = 0. So, all terms in the expansion are 0. (This argument also works when the exponent of (a b) is m + n 1.) Finally, (ab)m = amrm = 0.
32. We may assume that m is the smallest positive integer such that am = 0. If m < n, then a = an = aman m = 0an m = 0. If m = n, then a = an = am = 0. If m > n, observe that 0 = am = am nan = am na = am n+1 and m n + 1 < m, which is a contradiction to our assumption.
First alternate solution. Beginning with a and successively multiplying each prior term by a we get a, a2 , a3 , . . . an = a, a2 , . . . , an = a, . . ., returning to a after every n 1 iterations. Thus, there exist arbitrarily large k such that ak = a. So, for such k m we have a = ak = am am k = 0.
Second alternate solution. If m < n, then a = an = am an m = 0an m = 0. If m = n, then a = an = am = 0.
If m > n, observe that a = an means that we can replace each of the n terms in an 2 by an to obtain a = (an)n = an Then we can replace each of the n2 terms in an 2 2 4 a == (a ) = a . Continuing this process we eventually reach Case 1.
If m > n, observe that a = an = aan 1 = an an 1 =
= 23 n 3 . Continuing this for k iterations we have a = akn (k 1) Thus we eventually reach Case 1.
33. In M2(Z), let a = 0 1 and b = 1 0 0 0 0 0
34. ba = (ba)n = baba · · · = b(ab)a · · · = 0.
35. By inspection, R is closed under addition and multiplication. The elements 0 1 0 0 and 0 1 do not commute, 0 1
For the general case, use m × m matrices with the first m 1 columns all 0 and that last columns elements from Zn.
36. For part a observe that 2x = (2x)3 = 8x3 = 8x So, 0 = 8x 2x = 6x By part a we must have 1 divides 6 so we need only check n = 1, 2, 3 and 6 and these have the desired property.
37. Observe that x = ( x)4 = x4 = x
38. Assume that a2n = a2 . Then a2(n+1) = a2
39. For Z6 use n = 3. For Z10 use n = 5. Say m = p2t where p is a prime. Then (pt)n = 0 in Zm since m divides (pt)n .
40. Say k = ms and k = nt. Then ka = (ms)a = m(sa) and ka = (nt)a = n(ta) and therefore ka ∈ mZ ∩ nZ Now suppose b ∈ mZ ∩ nZ Then b is a common multiple of m and n. So, by Exercise 10 of Chapter 0, b ∈ kZ.
41. Every subgroup of Zn is closed under multiplication.
42. No. The operations are different.
43. Since ara asa = a(r s)a and (ara)(asa) = ara2sa = arsa, S is a subring. Also, a1a = a2 = 1, so 1 ∈ S.
44. The set is not closed under multiplication. 2
45. Let a a b and a ′ a ′ b ′ ∈ R. Then a b b a ′ b′ b′ a a b a ′ a ′ b′ = ′ ′ ′ a b b a b b a a ′ (a a ′) (b b′) (a a ′) (b b′) b b′ ∈ R. Also, a a b a ′ a ′ b′ = ′ ′ ′ a b b a b b aa ′ + aa ′ ab′ ba′ + bb′ aa ′ bb′ aa ′ bb′ aa ′ ab′ ba′ + bb′ + bb′ belongs to R.
46. The subring test is satisfied.
47. They satisfy the subring test but the multiplication is trivial. That is, the product of any two elements is zero.
48. The subring test is satisfied.
49. S is not a subring because (1, 0, 1) and (0, 1, 1) belong to S but (1, 0, 1)(0, 1, 1) = (0, 0, 1) does not belong to S.
50. Say n = 2m. Then a = ( a)n = ( a)2 m = (( a)2 )m = ((a)2 )m = a2 m = an = a.
51. Observe that n 1 m 1 = (n m ) 1. Also, (n 1)(m 1) = (nm) ((1)(1)) = (nm) 1.
52. 2Z ∪ 3Z contains 2 and 3, but not 2 + 3.
53. an (2/3)n + an 1 (2/3)n 1 + + a1 (2/3) a1 , a2 , . . . , an Z, n a positive integer . This set is a ring that contains 2/3 and is contained in every ring that contains 2/3. Alternate solution. 2n/3m n Z, m is a positive integer This set is a ring that contains 2/3 and is contained in every ring that contains 2/3.
54. Observe that 1 = (a 1b)c and 1 = b(ca 1).
55. (a + b)(a b) = a2 + ba ab b2 = a2 b2 if and only if ba ab = 0.
56. First note that a + b = (a + b)2 = a2 + ab + ba + b2 = a + ab + ba + b so that 0 = ab + ba or ab = ba. Then observe that ab = ( ab)2 = (ab)2 = ab.
58. 2x = 1 has no solution in Z4; 2x = 0 has two solutions in Z4; x = a 1(c b) is the unique solution when a 1 exists.
59. If (a, b) is a zero-divisor in R S, then there is a (c, d) = (0, 0) such that (a, b)(c, d) = (0, 0). Thus ac = 0 and bd = 0. So, a or b is a zero-divisor or exactly one of a or b is 0. Conversely, if a is a zero-divisor in R, then there is a c = 0 in R such that ac = 0. In this case, (a, b)(c, 0) = (0, 0). A similar argument applies if b is a zero-divisor. If a = 0 and b /= 0, then (a, b)(x, 0) = (0, 0) where x is any nonzero element in A. A similar argument applies if a /= 0 and b = 0.
60. The inverse is 2x + 3.
61. Fix some a in R, a = 0. Then there is a b in R such that ab = a. Now if x R and x = 0, then there is an element c in R such that ac = x. Then xb = acb = c(ab) = ca = x. Thus b is the unity. To show that every nonzero element r of R has an inverse, note that since rR = R, there is an element s in R such that rs = b.
62. In Z8, 22 = 4 = 62 and 23 = 0 = 63 .
63. One solution is R0 = ⟨20⟩ = Z, R1 = ⟨21⟩, R2 = ⟨22⟩,
64 Note that is r any element in a ring then r = a(a 1r).
65. In words, S is the set of all polynomials in π with the constant term 0. In symbols, S = cnπn + cn 1πn 1 + + c1π c1, c2, . . . , cn Z To see why S is the intersection of all subrings of R that contain π, let T be the intersection of all subrings of R that contain π Then S is one subring in the intersection, so T ⊆ S, and, by definition, S ⊆ T .
66. Both parts can be done by solving 1 = (a 1 + b 1)x = 1 for x. Doing so, we obtain ab = ab(a 1 + b 1 )x = (b + a)x = (a + b)x, which gives x = ab(a + b) 1 .
67. Experimenting with p = 3 and p = 5 reveals that in both case p + 1 is the unity. So, we want to show that for all x R we have that (p + 1)x = x mod 2p. Observe that because every x in R has 2 as a factor, (p + 1)x = px + x = x mod 2p
68. R is closed under subtraction and multiplication.
69. R is not closed under multiplication.
70. Note that every element of R has the form 2A where A is in M2(Z). So, if 2A and 2B are in R, then 2A 2B = 2(A B) and (2A)(2B) = 2(2AB) are in R. The argument works for any positive integer k
CHAPTER 13
Integral Domains
1. For Example 1, observe that Z is a commutative ring with unity 1 and has no zero-divisors. For Example 2, note that Z[i] is a commutative ring with unity 1 and no zero-divisors since it is a subset of C, which has no zero-divisors. For Example 3, note that Z[x] is a commutative ring with unity h(x) = 1 and if f (x) = anxn + + a0 and g(x) = bmxm + + b0 with an /= 0 and bm /= 0, then√ f (x)g(x) = an bm xn + m + · · · + a0 b0 and an bm /= 0. For Example 4, elements of Z[ 2] c o m m √ u t e since t h e √ y are real numbers; 1 is√the unity; (a + b√2) (c +√d 2) = (a c) + (b d) 2√a nd √ √ (a + b 2)(c + d 2) = (ac + 2bd) + (bc + ad) 2 so Z[ 2] is a ring; Z[ 2] has no zero-divisors because it is a subring of R, which has no zero-divisors. For Example 5, note that Zp is closed under addition and multiplication and multiplication is commutative; 1 is the unity; in Zp, ab = 0 implies that p divides ab. So, by Euclid’s Lemma (see Chapter 0), we know that p divides a or p divides b. Thus, in Zp, a = 0 or b = 0. For Example 6, if n is not prime, then n = ab where 1 < a < n and 1 < b < n But then a /= 0 and b /= 0 while ab = 0. For Example 7, note that 1 0 0 0 = 0 0 . For Example 8, note that (1, 0)(0, 1) = (0, 0).
2. Example 5
3. Let ab = 0 and a /= 0. Then ab = a · 0, so b = 0.
4. 2, 4, 5, 6, 8, 10, 12, 14, 15, 16, 18. The zero-divisors and the units constitute a partition of Z20 .
5. Let k Zn . If gcd(k, n) = 1, then k is a unit. If gcd(k, n) = d > 1, write k = sd. Then k(n/d) = sd(n/d) = sn = 0.
6. 2 in Z or x in Z[x].
7. Let s ∈ R, s /= 0. Consider the set S = {sr | r ∈ R}. If S = R, then sr = 1 (the unity) for some r. If S /= R, then there are distinct r1 and r2 such that sr1 = sr2. In this case, s(r1 r2) = 0.
Alternate solution. Let a R, a = 0, and a = 1. If there is some positive integer m such am = 0, let n be the least such integer. Then aan 1 = 0 and a is a zero-divisor. If there is no such m consider, the infinite list a, a2 , a3 , Since R is finite, we must have some distinct positive integers m and n (m > n) with am = an . Then am an = an(am n 1) = 0. If am n 1 = 0, a is a unit. If am n 1 = 0, a is a zero-divisor. To see what happens when the “finite” condition is dropped, note that in the ring of integers, 2 is neither a zero-divisor nor a unit.
8. Suppose that a is a zero-divisor and let ab = 0 for some b /= 0. Then a2b = a(ab) = 0 and b /= 0. Next assume that a2b = 0 for some b /= 0. If ab = 0, then a is a zero-divisor. If ab /= 0, then a2b = a(ab) = 0 and we are done.
9. Take a = (1, 1, 0), b = (1, 0, 1) and c = (0, 1, 1).
10. {(2, b, c) | b ∈ Z5, c ∈ Z6} ∪ {(a, b, c) | a ∈ Z4, b ∈ Z5, c = 2, 3, 4}. √ √ √ (a1 + b1 √ d)(a2 + b2 √ d) = (a1a2 + b1b2d) + (a1b2 + a2b1) √ d Thus the set is a ring. Since Z[ √ d] is a subring of the ring of complex numbers, it has no zero-divisors.
12. Let 1 = x Then 2x = 1. So, x = 4. Let 2 = x Then 2 = 3x which means 2 √ √ 3 1 1 6 5 = 3x. So x = 4. Note that 3 = 4 = 2 or 5. 6 = 6 = 6 = 1.
13. The ring of even integers does not have a unity.
14. Look in Z6.
15. Suppose that a2 = a for some a in Zpn Then a2 a = 0 and therefore, pn divides a2 a = a(a 1). So, by Euclid’s Lemma, p divides a or p divides a 1. But p cannot divide both a and a 1, for otherwise, p would divide a (a 1) = 1. Thus, pn divides a or pn divides a 1 This means that a = 0 or a 1 = 0.
16. First note that a2 = a implies that a3 = a2 = a. Then (a 1)3 = a3 3a2 + 3a 1 = a 1.
17. a2 = a implies a(a 1) = 0.
18. Suppose a = 0 and an = 0, where we take n to be as small as possible. Then a 0 = 0 = an = a an 1, so by cancellation, an 1 = 0. This contradicts the assumption that n was as small as possible.
19. If a2 = a and b2 = b, then (ab)2 = a2b2 = ab. The other cases are similar.
20. Note that if n can be written is the form p2 m where p is a prime, then (pm)2 = p2m2 = nm = 0 in Zn and pm is not 0. On the other hand, if n is of the form p1p2 pt where the pi are distinct primes and ak = 0 mod n it follows from Euclid’s Lemma that each pi divides a Thus a = 0 in Zn
21. Let f (x) = x on [ 1, 0] and f (x) = 0 on (0,1] and g(x) = 0 on [ 1, 0] and g(x) = x on (0,1]. Then f (x) and g(x) are in R and f (x)g(x) = 0 on [ 1, 1].
22. We proceed by induction. The n = 1 case is trivial. Assume that an = a. Then an +1 = aan = aa = a.
23. Suppose that a is an idempotent and an = 0. By the previous exercise, a = 0.
24. (3 + 4i)2 = 3 + 4i; (3 + i)2 = 3 + i
25. There are four in all. Since i = 2i = 4, all we need do is use the table to find an element whose square is i or 2i. These are 1 + i, 1 + 2i, 2 + i, and 2 + 2i.
26. Units: (1, 1), (1, 5), (2, 1), (2, 5); zero-divisors: (a, b) a 0, 1, 2 , b 2, 3, 4 ; idempotents: (a, b) a = 0, 1, b = 1, 3, 4 ; nilpotents: (0, 0).
27. a2 = a implies a(a 1) = 0. So if a is a unit, a 1 = 0 and a = 1.
28. a. If f (a) = 0 for some a, let g(x) = 0 for all x = a and g(a) = 1. Then f (x)(g(x) = 0.
b. f (x) = 0
c. If f (x) is never 0, then 1/f (x) is defined for all x.
29. Since F is commutative so is K. The assumptions about K satisfy the conditions for the One-Step Subgroup Test for addition and for multiplication (excluding the 0 element). So, K is a subgroup under addition and a subgroup under multiplication (excluding 0). Thus K is a subring in which every nonzero element is a unit.
30. The proof that Q[ √ d] is a s√u b r i n g is the same as in Exercise 11. √M o r e o v e r , if a and b are not both 0, then (a + b d) 1 = a/(a2 db2 ) (b/(a2 db2 ) d. T h i√ s also shows that there are n√ o zero divisors. To check that a2 db2 /= 0, n o t√e that if d is an integer then Q[ is false. d] = Q. Otherwise, a2 db2 = 0 implies that d is rational, which
31. Note that ab = 1 implies aba = a. Thus 0 = aba a = a(ba 1). So, ba 1 = 0.
32. 6 is the unity; 4 and 6 are their own inverses and 2 and 8 are inverses of each other. {0, 4, 8}
33. A subdomain of an integral domain D is a subset of D that is an integral domain under the operations of D. To show that P is a subdomain, note that n 1 m 1 = (n m) 1 and (n 1)(m 1) = (mn) 1 so that P is a subring of D. Moreover, 1 P, P has no zero-divisors since D has none, and P is commutative because D is. Also, since every subdomain contains 1 and is closed under addition and subtraction, every subdomain contains P Finally, we note that |P | = char D when char D is prime and |P | is infinite when char D is 0.
34. An integral domain of order 6 would be an Abelian group of order 6 under addition and, by Theorem 13.4, the characteristic is prime. But the only Abelian group of order 6 is Z6 , which by Theorem 13.3 has characteristic 6. This argument cannot be adapted to a ring with exactly four elements because the ring under addition could be isomorphic Z2 Z2 , which has characteristic 2. The argument can be adapted for 15 elements. In general, there is no integral domain R with n elements when n is divisible by two distinct primes p and q because by the Fundamental Theorem of Finite Abelian Groups (or Theorem 9.5) R would have elements with additive orders p and q. So, |1| would be a prime divisible by both p and q.
35. By Theorem 13.3, the characteristic is 1 under addition. By Corollary 2 of Theorem 7.1, 1 divides 2n . By Theorem 13.4, the characteristic is prime. Thus, the characteristic is 2.
36. Solve the equation x2 = 1.
37. By Exercise 36, 1 is the only element of an integral domain that is its own inverse if and only if 1 = 1. This is true only for fields of characteristic 2.
38. If n is a prime, then Zn is a field and therefore has no zero-divisors. If n is not a prime, we may write n = ab where both a and b are less than n If a = b, then (n 1)! includes both a and b among its factors so (n 1)! = 0. If a = b and a > 2, then (n 1)! = (a2 1)(a2 2) (a2 a) (a2 2a) 2 1. Since this product includes a2 a = a(a 1) and a2 2a = a(a 2), it contains a2 = n = 0. The only remaining case is n = 4 and in this case 3! = 2 is a zero-divisor.
39. a. First note that a3 = b3 implies that a6 = b6 . Then a = b because we can cancel a5 from both sides (since a5 = b5).
b. Since m and n are relatively prime, by the corollary of Theorem 0.2, there are integers s and t such that 1 = sn + tm Since one of s and t, is negative we may assume that s is negative. Then a(an) s = a1 sn = (am )t = (bm )t = b1 sn = b(bn) s = b(an) s Now cancel (an) s .
Alternate proof for part b. Suppose there are relative prime positive integers m and n for which the statement is false. Among all such integers let n < m and n be the minimum. Write m = nq + r where 0 r < n. Since n does not divide m, r = 0. Then we can cancel (an)q on the right side of am = bm and (bn)q on the left side to get ar = br . Moreover, n and r are relative prime because any common divisor of r and n would be a common divisor of m and n. This contradicts the minimality of n.
40. In Z, take a = 1, b = 1, m = 4, n = 2.
41. If K is a subfield of F , then K∗ is a subfield of F ∗ , which has order 31. So, |K∗| must divide 31. This means that |K∗| = 1 or 31.
42. No. No.
0 0 0 0 0
43. In Z [k], note that (a + b √ k) 1 = 1√ (a b √ k) = a b √
if and only if p
a2 b2k /= 0 where a /= 0 and b /= 0.
44. Observe that (1 + i)4 = 1, so |1 + i| = 8 and therefore the group is isomorphic to Z8
45. Let a be a non-zero element. If the ring has n elements then the sequence a, a2 , . . . , an +1 has two equal elements. Say ai = ak + i = ak ai . Let x be a non-zero element in the ring. Then xakai = xai implies that 0 = xakai xai = ai(xak x). So, 0 = xak x and therefore xak = x.
Alternative solution: Let S = {a1, a2, . . . , an} be the nonzero elements of the ring. Then a1a1, a1a2, , a1an are distinct elements for if a1ai = a1aj, then a1(ai aj) = 0 and therefore ai = aj. If follows that S = {a1a1, a1a2, . . . , a1an}. Thus, a1 = a1ai for some i Then ai is the unity, for if ak is any element of S, we have a1ak = a1aiak, so that a1(ak aiak) = 0. Thus, ak = aiak for all k.
46. Say (abb)c = 0 where c = 0. If ac = 0, then a is a zero-divisor. If ac = 0, then (ac)b = 0 so that b is a zero-divisor.
47. Suppose that x and y are nonzero and |x| = n and |y| = m with n < m Then 0 = (nx)y = x(ny). Since x /= 0, we have ny = 0. This is a contradiction to the fact that |y| = m
48. Use Exercise 47 and the observation that if |a| = mn, then |ma| = n
49. a. For n = 2 the Binomial Theorem gives us (x1 + x2 )p = xp + pxp 1 x2 + · · · + px1 xp 1 + xp , where the coefficient p!/k!(p k)! of every term between x1 and x2 is divisible by p. Thus, (x1 + x2)p = xp + xp . The general case follows by induction on n. 1 2
b. This case follows from Part a and induction on m
c. Note Z4 is a ring of characteristic 4 and (1 + 1)4 = 24 = 0, but 14 + 14 = 1 + 1 = 2.
50. Let the characteristic be p and an = 0. Then pn > n implies that ap n = ap n nan = 0 and, by Exercise 49, (1 + a)p n = 1 + ap n = 1.
51. By Theorem 13.4, 1 has prime order, say p. Then, by Exercise 47, every nonzero element has order p If the order of the field were divisible by a prime q other than p, Cauchy’s Theorem (9.5) implies that the field also has an element of order q. Thus, the order of the field is pn for some prime p and some positive integer n
52. Z3[x]
53. n a b = 0 0 for all members of M2(R) if and only if na = 0 for all a in R
54. Observe char R = least common multiple x x R (additive order). Now use Corollary 2 to Theorem 7.1.
55. This follows directly from Exercise 54.
56. 2 + i and 2 + 2i
57. a. 2 b. 2, 3 c. 2, 3, 6, 11 d. 2, 3, 9, 10
c d 0 0 1 2 k
58. char S is a divisor of m. To verify, this let char S = n and write m = nq + r where 0 r < n Then for all x in S we have rx = (m nq)x = mx nqx = 0 0 = 0. Since n is the least positive integer such that nx = 0 for all x in S we have r = 0. Alternate proof. Let char S = n By Theorem 0.2 there are integers s and t such that d = gcd(m, n) = ms = nt. Then for all x in S we have dx = mxs + nxt = 0. So, d n. Since d is a divisor of n we have n = d. Alternate proof. First observe that for a ring with positive characteristic, the characteristic is the least common multiple of the orders of the elements. Let char S = n = pn 1 pn2 · · · pnk where the pi are distinct primes. Then for each pi there is an element si in S such that pni divides |si| Thus ⟨si⟩ has an element s ′ of order pn i i i i Since ms ′ = 0, we have that pni divides m i i
59. By Theorem 13.3, char R is prime. From 20 1 = 0 and 12 1 = 0 and Corollary 2 of Theorem 4.1, we know that char R divides both 12 and 20. Since the only prime that divides both 20 and 12 is 2, the characteristic is 2.
60. If a2 = a and b2 = b, then (a b)2 = a2 + b2 = a b and (ab)2 = a2 b2 = ab
61. Suppose a ∈ Zp and a2 + 1 = 0. Then (a + i)(a i) = a2 + 1 = 0.
62. Let a be an idempotent other than 0 or 1. Then a2 = a implies that a(a 1) = 0.
63. Suppose that F is a field of order 16 and K is a subfield of F of order 8. Then K∗ is a subgroup of F ∗ and |K∗| = 7 and |F ∗| = 15, which contradicts Lagrange’s theorem.
64. Use Corollary 4 of Theorem 7.1.
65. By Exercise 49, x, y K implies that x y K Also, if x, y K and y = 0, then (xy 1 )p = xp (y 1 )p = xp (yp ) 1 = xy 1 . So, by Exercise 29, K is a subfield.
66. Note that (a + bi) (c + di) = a c + (b d) ∈ Q(i). Also, in the complex numbers, 1 = 1 a bi = a bi = (a2 + b2) 1(a bi), which is in Q(i). a+bi a+bi a bi a2+b2
67. The unity is (1, 1, . . .. Let Sn = 0 0 Zn 0 0 . Then Sn is a subring with characteristic n This shows that R cannot have a positive characteristic.
68. (1 a)2 = 1 2a + a2 = 1 2a + a = 1 a.
69. Observe that because uSuR = uS = uSuS we have uS(uR uS) = 0.
70. If ⟨a⟩ = ⟨b⟩, then there are elements u and v such that a = bu and b = va Then a = bu = vau and a(1 vu) = 0 and vu = 1. If a = bu, then a ∈ ⟨b⟩ and if u is a unit then b = au 1 ∈ ⟨a⟩.
71. Since a field of order 27 has characteristic 3, we have 3a = 0 for all a. Thus, 6a = 0 and 5a = a
72. Since the characteristic of a field of order 2n is 2, it suffices to show that a = b for then 0 = a2 + ab + b2 = 3a2 = 2a2 + a2 = a2 . Note that a3 b3 = (a b)(a2 + ab + b2) = 0 so that a3 = b3 . In a field of order 2n , x2n +1 = x2 for all x (see Exercise 54). Also, for all odd n, 2n + 1 is divisible by 3. Thus a2 = a2n +1 = (a3)(2n +1)/ 3 = (b3)(2n +1)/ 3 = b2n+1 = b2 . Finally, a3 = b3 and a2 = b2 imply a = b.
73. Let a F , where a = 0 and a = 1. Then (1 + a)3 = 13 + 3(12 a) + 3(1a2 ) + a3 = 1 + a + a2 + a3 If (1 + a)3 = 13 + a3 , then a + a2 = 0. But then a(1 + a) = 0 so that a = 0 or a = 1 = 1. This contradicts our choice of a
74. 0, 1, p, p + 1. To verify that this list is complete say that a ∈ Z2p and a2 = a, where 1 < a < 2p. Then we have a(a 1) = 0 in Z2p. So, p divides a or p divides a 1. Thus, a = p or a 1 = p.
75. 0, 1, p 1, p, p + 1. To verify that this list is complete say that a ∈ Z2p and a3 = a, where 1 < a < 2p Then we have a(a 1)(a + 1) = 0 in Z2p So, p divides a, or p divides a 1, or p divides a + 1. Thus, a = p or a 1 = p or a + 1 = p.
76. Z3 ⊕ Z3 ⊕ · · · ⊕ Z3
77. Z2 ⊕ Z2 ⊕ · · · ⊕ Z2
78. f (x) = x, f (x) = |x|, f (x) = ⌊x⌋, f (x) = ⌈x⌉, and the function defined by f (x) = 1/x for x /= 0 and f (0) = 0.
79. Suppose that Zp [i] is a field and p = 1 mod 4 and a and b are integers such that a2 + b2 = p. Then (a + bi)(a bi) = a2 + b2 = p = 0. So, Zp[i] has a zero-divisor. Now suppose that p = 3 mod 4. By Theorem 13.2, it suffices to show that Zp[i] has no zero-divisors. If a + bi is a zero-divisor, then so is (a + bi)(a bi) = a2 + b2 in Zp[x]. But (a2 + b2)(x + yi) = 0 implies that a2 + b2 = 0 mod p. So, there is an integer t such that a2 + b2 = pt for some integer t So, mod 4, we have a2 + b2 = 3t′ for some t′ in 0, 1, 2, 3 . Trying all four cases for t′ gives that mod 4, a = b = 0 or a = b = 2. So, a2 + b2 is 0 or 4. If a = b = 0, then a + bi is not a zero-divisor. If a = b = 2, then a2 + b2 = 4 mod p, which is not 0 mod p
80. Say that ab = 0 and b = 0. Then, pn divides ab. If pn divides b, then b = 0. So, p divides a Say a = pc Then, ak 1 = pk 1 pc = pk c = 0.
CHAPTER 14
Ideals
and Factor Rings
1. Let r1a and r2a belong to ⟨a⟩. Then r1a r2a = (r1 r2)a ∈ ⟨a⟩. If r ∈ R and r1a ∈ ⟨a⟩, then r(r1a) = (rr1)a ∈ ⟨a⟩
2. To prove that A is an ideal, note that f (x) A if and only if f (0) = 0. If f (x), g(x) A, then f (0) = 0 and g(0) = 0. So, f (0) g(0) = 0 0 = 0 and h(0)f (0) = h(0)0 = 0 for all h(x) in Z[x]. Finally, note that f (x) x if and only if f (0) = 0.
3. Clearly, I is not empty. Now observe that (r1 a1 + + rn an ) (s1 a1 + + sn an ) = (r1 s1 )a1 + + (rn sn )an ∈ I. Also, if r R, then r(r1 a1 + + rn an ) = (rr1 )a1 + + (rrn )an I That I J follows from closure under addition and multiplication by elements from R.
4. {(a, a) | a ∈ Z}; {(a, a) | a ∈ Z}
5. Let a + bi, c + di ∈ S. Then (a + bi) (c + di) = a c + (b d)i and b d is even. Also, (a + bi)(c + di) = ac bd + (ad + cb)i and ad + cb is even. Finally, (1 + 2i)(1 + i) = 1 + 3i ∈ / S.
6. a. ⟨2⟩ b. ⟨2⟩ and ⟨5⟩ c. ⟨2⟩ and ⟨3⟩ d. ⟨p⟩ where p is a prime divisor of n.
7. Suppose that s is not prime. Then we can write s = pm where p is prime and m > 1. Then s is properly contained in p and p is properly contained in Zn. So s is not maximal. Now suppose that s is prime and there is a divisor t > 1 of n such that t properly contains s (recall every subgroup of Zn has the form k where k is a divisor of n). Then s = rt for some r. So we have t = s.
8. If |a| = n, then n(ma) = m(na) = 0.
9. If aR is an non-zero ideal of R, we know that aR = R. So, a belongs to R.
10. Because 2x ∈ ⟨x, 3⟩ we have that ⟨2x, 3⟩ ⊆ ⟨x, 3⟩. Because 4x, 3x ∈ ⟨2x, 3⟩ we have x ∈ ⟨2x, 3⟩. So, ⟨x, 3⟩ ⊆ ⟨2x, 3⟩ ⊆.
11. Since ar1 ar2 = a(r1 r2) and (ar1)r = a(r1r), aR is an ideal. 4R = {. . . , 16, 8, 0, 8, 16, . . .}.
12. Mimic Exercise 9 of Chapter 12.
13. If n is a prime and ab Z, then by Euclid’s Lemma (Chapter 0), n divides a or n divides b. Thus, a nZ or b nZ. If n is not a prime, say n = st where s < n and t < n, then st belongs to nZ but s and t do not.
14. (a1 + b1 ) (a2 + b2 ) = (a1 a2 ) + (b1 b2 ) ∈ A + B; r(a + b) = ra + rb ∈ A + B; (a + b)r = ar + br ∈ A + B
15. a. a = 1 b. a = 2 c. a = gcd(m, n)
16. Let a1b1 + + anbn and a ′ 1b′ 1 + a ′ mb′ m then (a1b1 + · · · + anbb) (a ′ 1b′ 1 + · · · + a ′ mb′ m) = a1b1 + · · · + anbn + ( a ′ 1)b′ 1 + · · · + ( a ′ m)b′ m ∈ AB.
Also and
17. a. a = 12 r(a1 b1 + + an bn ) = (ra1 )b1 + + (ran )bn ∈ AB (a1 b1 + · · · + an bn )r = a1 (b1 r) + · · · + an (bn r) ∈ AB.
b. a = 48. To see this, note that every element of ⟨6⟩⟨8⟩ has the form 6t18k1 + 6t28k2 + · · · + 6
. So, ⟨6⟩⟨8⟩ ⊆ ⟨48⟩. Also, since 48 ∈ ⟨6⟩⟨8⟩, we have ⟨48⟩ ⊆ ⟨6⟩⟨8⟩.
c. a = mn
18. Since A and B are ideals, ab A and ab B when a A and b B. Now AB is just the sum of such terms.
19. Let r ∈ R. Then r = 1r ∈ A.
20. By Exercise 18, we have AB ⊆ A ∩ B. So, let x ∈ A ∩ B. To show that x ∈ AB, start by writing 1 = a + b where a ∈ A, b ∈ B.
21. Let u ∈ I be a unit and let r ∈ R Then r = r(u 1 u) = (ru 1 )u ∈ I
22. Let A be a prime ideal of R. By Theorem 14.3, R is an integral domain. Then, by Theorem 13.2, R/A is a field and, by Theorem 14.4, A is maximal.
23. Observe that ⟨2⟩ and ⟨3⟩ are the only nontrivial ideals of Z6 , so both are maximal. More generally, Zpq , where p and q are distinct primes, has exactly two maximal ideals.
24. Observe that as groups R : I = 3. So there is not a proper subgroup of R that strictly contains I.
25. I is closed under subtraction since the even integers are closed under subtraction.
Also, if b1 , b2 , b3 , and b4 are even, then every entry of a1 a2 b1 b2 b3 b4 is even.
26. 0 I. Let f (x), g(x) I and h(x) Z[x]. Then f (1) g(1) even and f (1)h(1) is even.
27. The proof that I is an ideal is the same as the case that n = 2 in Exercise 25. The number of elements in I is n4
28. Since A is non-empty and closed under subtraction and multiplication, it is a subring. It is not an ideal because 0 0 1 0 = 0 0 is not in A
29. That I satisfies the ideal test follows directly from the definitions of matrix addition and multiplication. To see that R/I is a field, first observe that a b + I = a 0 + 0 b + I = a 0 + I.
Thus we need only show that a 0 + I has an inverse in R/I when a /= 0. To this end, note that a 0 a 1 0 1 0 + I = 0
I
30. Since ( i)i = i2 = 1 ∈ ⟨i⟩ every element of Z[i] is in ⟨i⟩ So, Z[x]/⟨i⟩ = {0 + ⟨i⟩}
31. R = {0 + ⟨2i⟩, 1 + ⟨2i⟩, i + ⟨2i⟩, 1 + i + ⟨2i⟩}. R is not an integral domain because (1 + i + ⟨2i⟩)2 = (1 + i)2 + ⟨2i⟩ = 1 + 1 + ⟨2i⟩ = 0 + ⟨2i⟩.
32. 1 = |1| < |2 i||a + bi| = √ 5 √ a2 + b2 .
33. Let I = ⟨x2 2x 3, x⟩ Observe that x ∈ I implies that x2 and 2x belong to I Then, so does (x2 2x 3) + (x2 + 2x) = 3. But then ( 3)( 2) = 1 belongs to I. So, I = Z5[x].
34. First note that I[x] is the set of all polynomials in Z[x] for which every coefficient is even. Since the element 3 + I[x] does not have a multiplicative inverse in Z[x]/I[x], it not a field. So, by Theorem 14.4, I[x] is not maximal.
35. Use the observation that every member of R can be written in the form
2q1 + r1 2q2 + r2
2q3 + r3 2q4 + r4 where each ri is 0 or 1. Then note that 2q1 + r1 2q2 + r2 2q3 + r3 2q4 + r4 + I = r1 r2 r3 r4 + I
36. Let R be the ring 0, 2, 4, 6 under addition and multiplication mod 8. Then 0, 4 is maximal but not prime.
37. (br1 + a1 ) (br2 + a2 ) = b(r1 r2 ) + (a1 a2 ) ∈ B; r ′(br + a) = b(r ′ r) + r ′ a ∈ B since r ′ a ∈ A.
38. A = ⟨2⟩
39. Suppose that I is an ideal of F and I = 0 . Let a be a nonzero element of I. Then by Exercise 21, I = F .
/ { } { } { } { } { }
40. Since for every nonzero element a in R, aR is a nonzero ideal of R, we have R = aR. Thus, there is an element b in R such 1 = ab This shows that every non-zero element in R has an inverse in R. So, R is a field. Alternate proof. Since the only ideal that properly contains 0 is R, 0 is a maximal ideal of R. So, by Theorem 14.4 R/0 is a field. So, for any non-zero element a in R there is an element b in R such that 1 + {0} = (a + {0})(b + {0}) = ab + {0}. Thus {1} = {ab} and 1 = ab
41. Let a be an idempotent other that 0 or 1. Then a2 = a implies that a(a 1) = 0.
42. Use Example 15 and Theorem 14.4. 0 0 0 0 + I =
43. Since every element of x has the form xg(x), we have x I. If f (x) I, then f (x) = anxn + + a1x = x(anxn 1 + + a1) x We know that I is not maximal because J = f (x) Z[x] f (0) is even is a proper ideal of Z[x] that properly contains I.
44. Since A = ⟨(3, 0), (0, 1)⟩, it is an ideal. Then observe that Z ⊕ Z/A = {(0, 0) + A, (1, 0) + A, (2, 0) + A} ≈ Z3 and use Theorem 14.4. In general, for A = {(nx, y) | x, y ∈ Z}, Z ⊕ Z/A = {(0, 0) + A, (1, 0) + A, (2, 0) + A, , (n 1, 0) + A} ≈ Zn Thus, A is a maximal ideal of Z ⊕ Z if and only if n is prime.
45. Suppose J is an ideal that properly contains I and let f (x) J but not in I. Then, since g(x) = f (0) f (x) belongs to I and f (x) belongs to J , we know that f (0) = g(x) + f (x) is a non-zero constant contained in J So, by Exercise 17, J = R
47. Since (3 + i)(3 i) = 10 we know 10 + 3 + i = 0 + 3 + i Also, i + 3 + i = 3 + 3 + i = 7 + 3 + i . Thus, every element a + bi + 3 + i can be written in the form k + 3 + i where k = 0, 1, , 9. Finally, Z[i]/ 3 + i = k + 3 + i k = 0, 1, . . . , 9 , since 1 + 3 + i has additive that divides order 10. To see that the additive order is not 5, we note that if so then there are integers a and b such that 5 = (3 + i)(a + bi) = 3a b + (a + 3b)i. This gives us the two equations 5 = 3a b and 0 = a + 3b. But these have no solution in Z. The same argument rules out the orders 1 and 2.
48. Since I = 2m + 2n)i m, n Z = 2 , it is an ideal. Elements are 0 + i, 1 + I, i + I, 1 + i + I . Note that (1 + i + I)2 = 0 + I, so Z[i]/I is not a field or an integral domain.
49. Because I = (1, 0) , I is an ideal. To prove that I is prime suppose that (a, 0)(b, 0) = (ab, 0) = (0, 0). Then ab = 0 and therefore a = 0 or b = 0. So, I is prime. Finally, because (2, 0) has no multiplicative inverse in Z Z, A Z )/I is not a field and I is not maximal. Or note that I is a proper subring of the ideal J = {(a, b) | a, b ∈ Z, and b is even}
50. rs sr I if and only if rs sr + I = I or rs + I = sr + I. This is equivalent to (r + I)(s + I) = (s + I)(r + I).
51. Since every element in x, 2 has the form f (x) = xg(x) + 2h(x), we have f (0) = 2h(0), so that f (x) I If f (x) I, then f (x) = anxn + + a1x + 2k = x(anxn 1 + + a1) + 2k x, 2 . By Theorems 14.3 and 14.4, to prove that I is prime and maximal, it suffices to show that Z[x]/I is a field. To this end, note that every element of Z[x]/I can be written in the form anxn + · · · + a1x + 2k + I = 0 + I or anxn + · · · + a1x + (2k + 1) + I = 1 + I So, Z[x]/I ≈ Z2
52. I = ⟨2x, 4⟩.
53. One example is J = x2 + 1, 2 To see that 1 is not in J , note that if there were f (x), g(x) Z[x] such that (x2 + 1)f (x) + 2g(x) = 1, then evaluating the left side at 1 yields an even integer.
54. We first determine the characteristic of Z[i]/I. Note that 2(1 + i)(1 i) = 2(1 + 1) = 4 is in ⟨2 + 2i⟩ and therefore
4(1 + 2 + 2i ) = 0 + 2 + 2i . So, the characteristic is 4 or 2. But 2(1 + 2 + 2i ) = 0 + 2 + 2i So, by Theorem 13.3, the characteristic of Z[i]/I is 4. Because 0 + I = 2 + 2i + I, we have that 2i + I = 2 + I = 2 + I. So, every element of Z[i]/I can be written in the form a + I or a + i + I where a = 0, 1, 2, or 3. Finally, as in Example 11, we can show that i is not in I so these 8 elements are distinct. Finally, because 2 · 2 ∈ I and 2 is not, I is not prime.
55. 3x + 1 + I
56. Let a, b ∈ Ip. Say |a| = pn and |b| = pm . Then pn+m (a b) = 0 so |a b| divides pn + m Also, pn (ra) = r(pn a) = 0 so |ra| divides pn
57. Every ideal is a subgroup. Every subgroup of a cyclic group is cyclic.
58. Since a b r s = ar as + bt and r s a b 0 t 0 d = ra rb + sd for all a, b, and d, we must have that r and t 0 td are even.
59. Let I be any ideal of R ⊕ S and let IR = {r ∈ R| (r, s) ∈ I for some s ∈ S} and IS = {s ∈ S| (r, s) ∈ I for some r ∈ R}. Then IR is an ideal of R and IS is an ideal of S Let IR = ⟨r⟩ and IS = ⟨s⟩ Since, for any (a, b) ∈ I there are elements a ′ ∈ R and b′ ∈ S such that (a, b) = (a ′ r, b′ s) = (a ′ , b′)(r, s), we have that I = ⟨(r, s)⟩.
60. Suppose A is prime and B is an ideal which properly contains A. Say A = ⟨a⟩ and B = ⟨b⟩ It suffices to show b is a unit. Write a = br Then, since A is prime, b ∈ A or r ∈ A. If b ∈ A, then B ⊆ A so r ∈ A. Say r = ar ′ . Then a = br = bar′ so that a(1 br′) = 0. Thus 1 = br′ and b is a unit.
61. Say b, c ∈ Ann(A). Then (b c)a = ba ca = 0 0 = 0. Also, (rb)a = r(ba) = r · 0 = 0.
62. Suppose b, c ∈ N (A). Say, bn ∈ A and cm ∈ A Then the binomial theorem shows that (b c)n+m ∈ A. Also, (rb)n = rnbn ∈ A.
63. a. ⟨3⟩ b. ⟨3⟩ c. ⟨3⟩
64. a. ⟨6⟩ b. ⟨2⟩ c. ⟨6⟩
65. Suppose (x + N (⟨0⟩))n = 0 + N (⟨0⟩). We must show that x ∈ N (⟨0⟩). We know that xn + N (⟨0⟩) = 0 + N (⟨0⟩), so that xn ∈ N (⟨0⟩). Then, for some m, (xn)m = 0, and therefore x ∈ N (⟨0⟩).
66. Clearly N (A) ⊆ N (N (A)). Suppose x ∈ N (N (A)). Then xn ∈ N (A) for some n Thus (xn)m ∈ A for some m, so x ∈ N (A)).
⟩ / ⟨ ⟩ ⟨ ⟩ ⟨ ⟩ ⟨ ⟩ { } 0 d 0 t 0 dt
67. Let I = x2 + x + 1 . Then Z2[x]/I = 0 + I, 1 + I, x + I, x + 1 + I . 1 + I is its own multiplicative inverse and (x + I)(x + 1 + I) = x2 + x + I = x2 + x + 1 + 1 + I = 1 + I So, every nonzero element of Z2[x]/I has a multiplicative inverse.
68. R = {0 + ⟨2i⟩, 1 + ⟨2i⟩, i + ⟨2i⟩, 1 + i + ⟨2i⟩} R is not an integral domain because (1 + i + ⟨2i⟩)2 = (1 + i)2 + ⟨2i⟩ = 1 + 1 + ⟨2i⟩ = 0 + ⟨2i⟩.
69. x + 2 + ⟨x2 + x + 1⟩ is not zero, but its square is.
70. {na + ba | n ∈ Z, b ∈ R}
71. If f and g ∈ A, then (f g)(0) = f (0) g(0) is even and (f g)(0) = f (0) g(0) is even. f (x) = 1/2 ∈ R and g(x) = 2 ∈ A, but f (x)g(x) ∈ / A
72. Observe that 1 + ⟨1 i⟩ = i + ⟨1 i⟩ so any coset can be written in the form a + ⟨1 i⟩ where a ∈ Z. But 1 + ⟨1 i⟩ = (1 + ⟨1 i⟩)2 = (i + ⟨1 i⟩)2 = 1 + ⟨1 i⟩
so 2 + ⟨1 i⟩ = 0 + ⟨1 i⟩ This means that there are only two cosets: 0 + ⟨1 i⟩ and 1 + ⟨1 i⟩.
73. Any ideal of R/I has the form A/I where A is an ideal of R So, if A = ⟨a⟩, then A/I = ⟨a + I⟩/I.
74. There is 1 element. To see this, let I = ⟨1 + i⟩. Then (1 + i)(1 i) = 2 belongs to I and therefore 3 2 = 1 is in I. It follows that I = Z5[i] and the only element in the factor ring is 0 + I
75. In Z, ⟨2⟩ ∩ ⟨3⟩ = ⟨6⟩ is not prime.
76. Suppose that (a, b) is a nonzero element of an ideal I in R ⊕ R. If a /= 0, then (r, 0) = (ra 1 , 0)(a, b) ∈ I Thus, R ⊕ {0} ⊆ I Similarly, if b /= 0, then {0} ⊕ R ⊆ I So, the ideals of R ⊕ R are {0} ⊕ {0}, R ⊕ R, R ⊕ {0}, {0} ⊕ R. The ideals of F ⊕ F are {0} ⊕ {0}, F ⊕ F, F ⊕ {0}, {0} ⊕ F .
77. According to Theorem 13.3, we need only determine the additive order of 1 + ⟨2 + i⟩. Since 5(1 + ⟨2 + i⟩) = 5 + ⟨2 + i⟩ = (2 + i)(2 i) + ⟨2 + i⟩ = 0 + ⟨2 + i⟩, we know that 1 + ⟨2 + i⟩ has order 5.
78. Use the fact that a2 + b2 = (a + bi)(a bi).
79. The set K of all polynomials whose coefficients are even is closed under subtraction and multiplication by elements from Z[x] and therefore K is an ideal. By Theorem 14.3, to show that K is prime, it suffices to show that Z[x]/K has no zero-divisors. Suppose that f (x) + K and g(x) + K are nonzero elements of Z[x]/K. Since K absorbs all terms that have even coefficients, we may assume that f (x) = amxm + + a0 and g(x) = bnxn + + b0 are in Z[x] and am and bn are odd integers. Then (f (x) + K)(g(x) + K) = ambnxm+n + + a0 b0 + K and ambn is odd. So, f (x)g(x) + K is nonzero.
80. That I is closed under s ub t r a c√ t i o n follows from the d√efinition of addition in the c o m p l √ e x number. √S o , let a + b 5 ∈ I and let c√+ d 5 ∈ R. Then (a + b 5)(c + d 5) = (ac 5bd) + (bc + ad) 5. So we must show that (ac 5bd) (bc + ad) = (a b)c + ( 5b a)d is even when a b is even. This reduces to showing that 5b a is even. Note that a b is even means that both a and b are odd or both a and b are even. Either way, 5b a is even.
81. By Theorem 14.3, R/I is an integral domain. Since every element in R/I is an idempotent and Exercise 16 in Chapter 13 says that the only idempotents in an integral domain are 0 and 1, we have that R/I = {0 + I, 1 + I}.
82. Say J /= I is also a maximal ideal. Let x be an element of J that is not in I Then 1 = x 1x ∈ J so that J = R.
83. ⟨x⟩ ⊂ ⟨x, 2n⟩ ⊂ ⟨x, 2n 1⟩ ⊂ · · · ⊂ ⟨x, 2⟩
84. Use the ideal test to show that I is an ideal of R. To show that I is not generated by a single element observe that every element of I has the form (b1 , b2 , b3 , , bk , 0, 0, 0, ) where all terms
85. Suppose that J is an ideal that properly contains I. Then J/I is a subring of R/I with more than one element and |J/I| divides |R/I|. So J = R.
86. Let R = Z4 and I = ⟨2⟩.
87. Look at coset representatives and use the fact that (a + bx) 1 = 1 = 1 ax b = ax b . Then check to see if this works. ax+b ax+b ax b a2 b2
CHAPTER 15
Ring Homomorphisms
1. Property 1: φ(nr) = nφ(r) holds because a ring is a group under addition. To prove that φ(rn) = (φ(r))n, we note that by induction
Property 2: If φ(a) and φ(b) belong to φ(A), then φ(a) φ(b) = φ(a b) and φ(a)φ(b) = φ(ab) belong to φ(A).
Property 3: φ(A) is a subgroup because φ is a group homomorphism. Let s S and φ(r) = s. Then sφ(a) = φ(r)φ(a) = φ(ra) and φ(a)s = φ(a)φ(r) = φ(ar).
Property 4: Let a and b belong to φ 1(B) and r belong to R. Then φ(a) and φ(b) are in B So, φ(a) φ(b) = φ(a) + φ( b) = φ(a b) B Thus, a b φ 1(B).
Also, φ(ra) = φ(r)φ(a) B and φ(ar) = φ(a)φ(r) B So, ra and ar φ 1(B).
Property 5: φ(a)φ(b) = φ(ab) = φ(ba) = φ(b)φ(a).
Property 6: Because φ is onto, every element of S has the form φ(a) for some a in R Then φ(1)φ(a) = φ(1a) = φ(a) and φ(a)φ(1) = φ(a1) = φ(a).
Property 7: If φ is an isomorphism, by property 1 of Theorem 10.1 and the fact that φ is one-to-one, we have Ker φ = 0 If Ker φ = 0 , by property 5 of Theorem 10.2, φ is one-to-one.
Property 8: That φ 1 is one-to-one and preserves addition comes from property 3 of Theorem 6.3. To see that φ 1 preserves multiplication, note that φ 1(ab) = φ 1(a)φ 1(b) if and only if φ(φ 1(ab)) = φ(φ 1(a)φ 1(b)) = φ(φ 1(a))φ(φ 1(b)). But this reduces to ab = ab.
2. Since φ is a group homomorphism, Ker φ is a subgroup. Let a Ker φ and r R. The φ(ar) = φ(a)φ(r) = 0φ(r) = 0. Similarly, φ(ra) = 0.
3. We already know the mapping is an isomorphism of groups. Let Φ(x + Ker φ) = φ(x). Note that Φ((r + Ker φ)(s + Ker φ)) = Φ(rs + Ker φ) = φ(rs) = φ(r)φ(s) = Φ(r + Ker φ)Φ(s + Ker φ).
7. Observe that (x + y)/1 = (x/1) + (y/1) and (xy)/1 = (x/1)(y/1).
8. If φ is a ring-homomorphism from Zn to itself then φ(x) = φ(1x) = φ(1)x. Moreover, if φ(1) = a, then a2 = (φ(1))2 = φ(12) = φ(1) = a.
9. The condition that the order of k divides both m and n is necessary and sufficient for the mapping to be group homomorphism is Exercise 5 in Chapter 10 So, we need to show that k is an idempotent in Zn is necessary and sufficient for φ(ab) = φ(a)φ(b). If φ(ab) = φ(a)φ(b), then, in Zn , k = φ(1) = φ(1 1) = φ(1)φ(1) = k2 . On the other hand, if k2 = k in Zn , then φ(ab) = k(ab) = k2(ab) = (ka)(kb) = φ(a)φ(b).
10. Let φ be the mapping that takes a1 a2 a3 a4 ai ∈ Z to a1 mod 2 a2 mod 2 and use Theorem 15.3. a3 mod 2 a4 mod 2
11. For groups, φ(x) = ax for a = 2, 4, 6, 8 since each of these has additive order 5. For rings, only φ(x) = 6x since 6 is the only non-zero idempotent in R
12. Parts a and b. No. Suppose 2 → a. Then 4 = 2 + 2 → a + a = 2a and 4 = 2 · 2 → aa = 2a.
13. If a and b (b = 0) belong to every member of the collection, then so do a b and ab 1 . Thus, by Exercise 29 of Chapter 13, the intersection is a subfield.
14. Try a + bi → a + bx + ⟨x2 + 1⟩.
15. By observation, φ is one-to-one and onto. Since
φ((a + bi) + (c + di)) = φ((a + c) + (b + d)i) = a + c b + d = a
(
+ di) (b + d) a + c addition is preserved. Also,
φ((a + bi)(c + di)) = φ((ac bd) + (ad + bc)i) = ac bd ad + bc = a b b a c d d c = φ(a + bi)φ(c + di) (ad + bc) ac bd so multiplication is preserved.
16. Try a + b √ 2 → a 2b .
17. Since φ a b c d a ′ b′ c ′ d′ aa ′ + bc′ ab′ + bd′ ca ′ + dc′ cb′ + dd′ = aa ′ + bc′ /= aa ′ = φ a b c d φ a ′ b′ , multiplication is not preserved.
18. It is a ring homomorphism.
19. Yes. φ(x) = 6x is well defined because a = b in Z5 implies that 5 divides a b. So, 30 divides 6a 6b. Moreover, φ(a + b) = 6(a + b) = 6a + 6b = φ(a) + φ(b) and φ(ab) = 6ab = 6 · 6ab = 6a6b = φ(a)φ(b).
20. No. For φ(x) = 2x we have 2 = φ(1) = φ(1 · 1) = φ(1)φ(1) = 4.
21. The set of all polynomials passing through the point (1, 0).
22. Z
/ b a c ′ d′ | |
24. Say a ring homomorphism φ maps 1 to a. Then the additive order of a must divide 25 and 20. So a = 1 or 5 and therefore a = 0, 4, 8, 12 or 16. But 1 = 11 means that a = a2 . Checking each possibility we obtain that a = 0 or 16. Both of those give ring homomorphisms. = φ
23. If φ is a ring homomorphism from Zn to Zp2 , then |φ(1)| is 1 or p, since it must divide both n and p2 So, φ(1) is in ⟨p⟩ Because the only idempotent in ⟨p⟩ is 0, φ(x) = 0.
25. For Z6 to Z6, 1 → 0, 1 → 1, 1 → 3, and 1 → 4 each define a homomorphism. For Z20 to Z30, 1 → 0, 1 → 6, 1 → 15, and 1 → 21 each define a homomorphism.
26. By property 6 of Theorem 15.1, 1 must map to 1. Thus, the only ring-isomorphism of Zn to itself is the identity.
27. Suppose that φ is a ring homomorphism from Z to Z. Let φ(1) = a. By property 1 of Theorem 15.1, φ(m) = φ(m 1) = mφ(1) = ma. Moreover, a = φ(1) = φ(1 1) = (φ(1))2 = a2 . This gives us that a = 0 or 1. So, φ is the zero map or φ(m) = m for all integers m
28. Since (1, 0) is an idempotent and idempotents must map to idempotents; the possibilities are (0, 0), (1, 0), (0, 1), (1, 1).
29. If ab = 1, then φ(a)φ(b) = φ(1), which is a unity in S (see property 7 of Theorem 15.1). If a is a zero-divisor in R, then φ(a) is a zero-divisor in S or is 0.
30. The group A/B is cyclic of order 4. The ring A/B has no unity.
31. First, note that every element of R[x]/⟨x2 ⟩ can be written u niquely in the form a b 0 a
ax + a + ⟨x2⟩. Then mapping that takes ax + b + ⟨x2⟩ to is a ring isomorphism.
Alternate proof: The mapping an xn + an 1xn 1 + · · · + a1 x + a0 to a0 a1 is a ring homomorphism with kernel ⟨x2⟩.
32. The mapping from Z3[x] onto Z3[i] given by φ(f (x)) = f (i) is a ring homomorphism with Ker φ = ⟨x2 + 1⟩. So, by Theorem 15.3, the ring Z3[x]/⟨x2 + 1⟩ is isomorphic to Z3[i].
34. Consider the mapping given by (x, y) → (x mod a, y mod b) and use Theorem 15.3.
35. Suppose that φ is a ring homomorphism from Z Z to Z. Let φ((1, 0)) = a and φ((0, 1)) = b. Since φ((x, y)) = φ(x(1, 0) + y(0, 1)) = φ(x(1, 0)) + φ(y(1, 0)) = xφ((1, 0)) + yφ((0, 1)) = ax + by it suffices to determine a and b Because (1, 0)2 = (1, 0) and (0, 1)2 = (0, 1), we know that a2 = a and b2 = b. This means the a = 0 or 1 and b = 0 or 1. Thus there are four cases for (a, b):
(0,0) corresponds to (x, y) 0; (1,0) corresponds to (x, y) x; (0,1) corresponds to (x, y) y; (1,1) corresponds to (x, y) x + y
Each of the first is obviously a ring homomorphism. The last case is not because 2 = φ((1, 1)) = φ((1, 1)(1, 1)) = φ((1, 1))φ((1, 1)) = 4.
36. (n2 + (n + 1)2 + (n + 2)2) mod 3 = 2 while k2 = 2 mod 3 has no solution.
37. Say m = akak 1 · · · a1a0 and n = bkbk 1 · · · b1b0 Then m n = (ak bk)10k + (ak 1 bk 1)10k 1 + + (a1 b1)10 + (a0 b0). By the test for divisibility by 9 given in Example 8, m n is divisible by 9 provided that ak bk + ak 1 bk 1 + · · · + a1 b1 + a0 b0 = (ak + ak 1 + + a1 + a0) (bk + bk 1 + + b1 + b0) is divisible by 9. But this difference is 0 since the second expression has the same terms as the first expression in some other order.
38. (ak ak 1 . . . a1 a0 ) mod 11 = (a0 + 10a1 + 102 a2 + · · · + 10k ak ) mod 11 = (a0 a1 + a2 ( 1)kak) mod 11.
39. Since the sum of the digits of the number is divisible by 9, so is the number (see Example 8); the test for divisibility by 11 given in Exercise 38 is not satisfied.
40. This follows directly from Exercise 74 of Chapter 10 and Exercise 11 of Chapter 0.
41. Let α be the homomorphism from Z to Z3 given by α(n) = n mod 3. Then, noting that α(10i) = α(10)i = 1i = 1, we have that n = ak ak 1 a1 a0 = a
10
+
k 1 10k 1
+ a0 is divisible by 3 if and only if, modulo 3, 0 =
). But α(ak + ak 1 +
+ a0 ) = 0 mod 3 is equivalent a
being divisible by 3.
42. akak 1 · · · a1a0 mod 4 = (akak 1 · · · a
43. Observe that the mapping φ from Zn[x] is isomorphic to Zn given by, φ(f (x)) = f (0), is a ring-homomorphism onto Zn with kernel ⟨x⟩ and use Theorem 15.3.
44. By Exercise 43 Z[x]/⟨x⟩ ≈ Zn The statement now follows from Theorem 14.4 and the fact that Zn is a field if and only if n is prime.
45. The ring homomorphism from Z ⊕ Z to Z given by φ(a, b) = a takes (1,0) to 1. Or define φ from Z6 to Z6 by φ(x) = 3x and let R = Z6 and S = φ(Z6 ). Then 3 is a zero-divisor in R and φ(3) = 3 is the unity of S
46. Since F ∗ is a group under multiplication any automorphism φ of F takes 1 to 1. By Corollary 3 of Theorem 15.3 we may assume the prime subfield is Zp or Q. Then, by properties of isomorphisms, for any element n in Zp or integer n in Q we have
φ(n) = n and for any element m/n in Q we have φ(m/n) = φ(m)/(φ(n)) 1 = mn 1 = m/n.
47. Observe that 10 mod 3 = 1. So, (2 · 1075 + 2) mod 3 = (2 + 2) mod 3 = 1 and (10100 + 1) mod 3 = (1 + 1) mod 3 = 2 = 1 mod 3. Thus, (2 · 1075 + 2)100 mod 3 = 1100 mod 3 = 1 and (10100 + 1)99 mod 3 = 299 mod 3 = ( 1)99 mod 3 = 1 mod 3 = 2.
48. By Exercise 27 we know that φ is the zero map or φ(m) = m for all integers m. Then, by property of Theorem 15.1, for all rationals m/n, we have φ(m/n) = φ(mn 1) = φ(m)φ(n 1) = φ(m)(φ(n)) 1 = mn 1 = m/n.
49. By Theorem 13.3, the characteristic of R is the additive order of 1 and by property 6 of Theorem 15.1, the characteristic of S is the additive order of φ(1). Thus, by property 3 of Theorem 10.1, the characteristic of S divides the characteristic of R.
50. Use Exercise 49(a) of Chapter 13
51. No. The kernel must be an ideal.
52. Use Exercise 37 of Chapter 14.
53. a. Suppose ab ∈ φ 1(A). Then φ(ab) = φ(a)φ(b) ∈ A, so that a ∈ φ 1(A) or b ∈ φ 1(A).
b. Let Φ be the homomorphism from R to S/A given by Φ(r) = φ(r) + A. Then φ 1(A) = Ker Φ and, by Theorem 15.3, R/Ker Φ S/A So, φ 1(A) is maximal.
54. If φ 1 (A) = ⟨a⟩, then A = ⟨φ(a)⟩.
55. a. Since φ((a, b) + (a ′ , b′)) = φ((a + a ′ , b + b′)) = a + a ′ = φ((a, b)) + φ((a ′ , b′)), φ preserves addition. Also, φ((a, b)(a ′ , b′)) = φ((aa ′ , bb′)) = aa ′ = φ((a, b))φ((a ′ , b′)) so φ preserves multiplication.
b. φ(a) = φ(b) implies that (a, 0) = (b, 0), which implies that a = b φ(a + b) = (a + b, 0) = (a, 0) + (b, 0) = φ(a) + φ(b). Also, φ(ab) = (ab, 0) = (a, 0)(b, 0) = φ(a)φ(b).
c. Define φ by φ(r, s) = (s, r). By Exercise 7 in Chapter 8, φ is one-to-one and preserves addition. Since φ((r, s)(r ′ , s ′)) = φ((rr ′ , ss ′)) = (ss ′ , rr ′) = (s, r)(s ′ , r ′) = φ((r, s))φ((r ′ , s ′)) multiplication is also preserved.
56. Note that I ⊕ S is the pull back of the homomorphism from R ⊕ S onto R given by (a, b) → a (see Exercises 55 and 53).
57. The mapping φ(x) = (x mod m, x mod n) from Zmn to Zm Zn is a ring isomorphism.
58. Suppose that φ is a ring isomorphism from mZ onto nZ. Observe that by property 1 of Theorem 15.1 we have φ(mm) = mφ(m) and φ(mm) = φ(m)2 . This gives us that φ(m) = m.
59. First, note that φ(1) = 1 implies that φ(m) = φ(m 1) = mφ(1) = m. Now let φ ( √ 3 2) = a. Then 2 = φ(2) = φ ( √ 3 2 ) = (φ( 3 2))3 and therefore φ ( √ 3 2) = √ 3 2.
60. First show that any automorphism φ of R acts as the identity map on the rationals (compare with Exercise 68 of Chapter 68) Then show that if a < b then φ(a) < φ(b) (see Exercise 71 of Chapter 6). Next suppose that there is some a such that φ(a) = a. Say, a < φ(a). Pick a rational number r such that a < r < φ(a). Then φ(a) < φ(r) = r, a contradiction. A similar argument applies if φ(a) < a.
61. By Exercise 52, every non-trivial ring homomorphism from R to R is an automorphism of R. And by Exercise 60 the only automorphism of R is the identity.
62. To check that multiplication is operation preserving, observe that xy → a(xy) = a2xy = axay For the second part take m = 4, n = 6 and a = 4. Then 0 = φ(0) = φ(2 · 2) but φ(2)φ(2) = 2 · 2 = 4.
63. If a/b = a ′/b′ and c/d = c ′/d′ , then ab′ = ba′ and cd′ = dc′ . So, acb′d′ = (ab′)(cd′) = (ba′)(dc′) = bda′ c ′ . Thus, ac/bd = a ′ c ′/b′d′ and therefore (a/b)(c/d) = (a ′/b′)(c ′/d′).
64. Suppose that √ 2 maps to a + b √ 5. Then 2 maps to (a + b √ 5)2 = a2 + 2ab √ 5 + b2 √ Solving for 5 we get a rational number, which is a contradiction.√
A l t e √ r n a √ t e proof. First observe that 1 → 1 √ s o that 2 → 2. Suppose √ 2 → x. Then 2 = 2 2 → x2 This implies that x = ± 2, which are not in Q( 5).
65. Let F be the field of quotients of Z[i]. By definition F = {(a + bi)/(c + di)| a, b, c, d ∈ Z} Since F is a field that contains Z and i, we know that Q[i] ⊆ F . But for any (a + bi)/(c + di) in F we have a + bi = a + bi c di = (ac+ bd )+(bc ad )i = ac+ bd + (bc ad )i ∈ Q[i]. c+di c + di c di c2+d2 c2+d2 c2+d2
66. Map [a/b] to ab 1 .
67. The subfield of E is ab 1 a, b D, b = 0 . Define φ by φ(ab 1) = a/b. Then φ(ab 1 + cd 1) = φ((ad + bc)(bd) 1)) = (ad + bc)/bd = ad/bd + bc/bd = a/b + c/d = φ(ab 1) + φ(cd 1). Also, φ((ab 1)(cd 1)) = φ(acb 1d 1) = φ((ac)(bd) 1) = ac/bd = (a/b)(c/d) = φ(ab 1)φ(cd 1).
68. Zero-divisors do not have multiplicative inverses.
69. Reflexive and symmetric properties follow from the commutativity of D. For transitivity, assume a/b c/d and c/d e/f . Then adf = (bc)f = b(cf ) = bde, and cancellation yields af = be.
70. The set of even integers is a subring of the rationals.
71. Let φ be the mapping from T to Q given by φ(ab 1) = a/b Now see Exercise 67.
72. From Example 2 we have that the mapping that sends x + yi + ⟨a + bi⟩ to x + yi + ⟨a bi⟩ is an isomorphism.
{ | ∈ / } ≡ ≡ · · ·
73. Let anxn + an 1xn 1 + · · · + a0 ∈ R[x] and suppose that f (a + bi) = 0. Then an(a + bi)n + an 1(a + bi)n 1 + + a0 = 0. By Example 2, the mapping φ from C to itself given by φ(a + bi) = a bi is a ring isomorphism. So, by property 1 of Theorem 10.1, 0 = φ(0) = φ(an(a + bi)n + an 1(a + bi)n 1 + + a0) = φ(an)φ((a + bi))n
φ(an 1)φ((
74. a. Apply the definition.
b. Ker φ = a b c d a ∈ Z .
c. Use Theorem 15.3.
d. Yes, by Theorem 14.3.
e. No, by Theorem 14.4.
75. Certainly, the unity 1 is contained in every subfield. So, if a field has characteristic p, the subfield {0, 1, . . . , p 1} is contained in every subfield. If a field has characteristic 0, then {(m · 1)(n · 1) 1 | m, n ∈ Z, n /= 0} is a subfield contained in every subfield. This subfield is isomorphic to Q [map (m · 1)(n · 1) 1 to m/n].
).
76. By part 5 of Theorem 6.1, the only possible isomorphism is given by 1 → n If this mapping is an isomorphism, then 1 = 12 → n2 . So n2 = n mod 2n and it follows that n is odd. Now suppose n is odd. Then n(n 1) is divisible by 2n and n2 = n mod 2n. This guarantees that 1 → n is an isomorphism.
77. x → 6x is the only possibility.
78. By Exercise 5 it suffices to find an element of order 3 in Z6 that is an idempotent. The only choice is 4.
79. Say φ is a ring homomorphism from 2Z to 2Z and φ(2) = a. Then, φ(4) = φ(2) + φ(2) = 2a and φ(4) = φ(2 2) = φ(2)2 = a2 So, 2a = a2 , which implies that a = 0 or a = 2.
· ·
For the general case, let φ(n) = a. Then φ(n n) = nφ(n) = na and φ(n n) = φ(n)2 = a2 . So, a = n2a, which implies that a = 0 or n2 = 1, which implies that n = 1. In words, the only ring homomorphisms from nZ to nZ are the identity and the zero map.
CHAPTER 16
Polynomial Rings
1. f + g = 3x4 + 2x3 + 2x + 2 f · g = 2x7 + 3x6 + x5 + 2x4 + 3x2 + 2x + 2
2. Let f (x) = x4 + x and g(x) = x2 + x. Then f (0) = 0 = g(0); f (1) = 2 = g(1); f (2) = 0 = g(2)
3. The zeros are 1, 2, 4, 5.
4. Since R is isomorphic to the subring of constant polynomials, charR ≤ char R[x]. On the other hand, char R = c implies c(anxn + · · · + a0) = (can)xn + · · · + (ca0) = 0.
5. The only place in the proof of Theorem 16.2 and its corollaries that uses the fact the coefficients were from a field is where we used the multiplicative inverse of lead coefficient bm of g(x).
6. x2 , x2 + 1, x2 + x, x2 + x + 1. No two define the same function from Z2 to Z2.
7. Note that the functions defined by f (x) = x3 , x5 , x7 , . . . , are the same as the one defined by f (x) = x and the ones defined by f (x) = x4 , x6 , x8 , . . . , are the same as the one defined by f (x) = x2 So all such terms may be replaced by x and x2 In the general case note that by Fermat’s Little Theorem (Corollary 5 to Theorem 7.1) the function from Zp to Zp defined by g(x) = xp is the same as the function f (x) = x from Zp to Zp. So, every polynomial function with coefficients from Zp can be written in the form ap 1xp 1 + + a0 where ap 1, . . . , a0 ∈ Zp.
8. In the ring functions from Z3 to Z3 the functions f (x) = x3 and g(x) = x are equal. So, 0 = f (x) g(x) = x3 x = x(x2 x) but h(x) = x and k(x) = x2 x are not zero functions. This does not contradict Theorem 16.1 because the ring of functions from Z3 to Z3 is not the same as the ring of polynomials from Z3 to Z3.
9. (x 1)2(x 2).
10. There are 2n polynomials over Z2 There are 4 polynomial functions from Z2 to Z2
11. 4x2 + 3x + 6 is the quotient and 6x + 2 is the remainder.
12. (x i)(x + i)(x (2 + i))(x (2 i))
13. Let f (x), g(x) ∈ R[x]. By inserting terms with the coefficient 0 we may write f (x) = a
Multiplication is done similarly.
14. Use Exercise 13 and observe that φ(an)xn + · + φ(a0) = 0 if and only if φ(an) = 0, , φ(a0) = 0. Since φ is an isomorphism, this holds if and only if an = 0, . . . , a0 = 0.
15. Note that (2xn + 1)2 = 1 and (2xn )2 = 0 for all n
/
16. If f (x) = g(x), then f (x) g(x) would be a polynomial with degree at most 3 with 4 zeros, contradicting Theorem 16.3. In general, if f (x) and g(x) are polynomials over a field of degree at most n and there are more than n values of a such that f (a) = g(a), then f (x) = g(x).
17. Observe that (2x + 1)(2x + 1) = 4x2 + 4x + 1 = 1. So, 2x + 1 is its own inverse.
18. Let a be a zero-divisor. Then f (x) = ax has degree 1 with zeros 0 and a
19. If f (x) = an xn + · · · + a0 and g(x) = bm xm + · · · + b0 , then f (x) g(x) = anbmxm+n + + a0b0 and anbm /= 0 when an /= 0 and bm /= 0.
20. Suppose that f (x) = anxn + + a0 from Z4[x] is an idempotent where an = 0 and n 1. Because the lead term of f (x)2 is a2 x2n and the lead term of f (x) is anxn we have a2 = 0. Thus an = 2. Comparing the coefficients of xn in both f (x)2 and f (x) we obtain 2ana0 = an and therefore an = 0. This is a contradiction.
/ /
21. Let m be the multiplicity of b in q(x). Then we may write f (x) = (x a)n(x b)mq ′(x) where q ′(x) is in F [x] and q ′(b) = 0. This means that b is a zero of f (x) of multiplicity at least m If b is a zero of f (x) of multiplicity greater than m, then b is a zero of g(x) = f (x)/(x b)m = (x a)nq ′(x). But then 0 = g(b) = (b a) q (b) and therefore q (b) = 0, which is a contradiction.
22. 1 is a zero when the field has characteristic 2. Over all other fields 2 is a zero.
23. Let f (x), g(x) ∈ R[x]. By adding coefficients with coefficient 0 in the front we can write f (x) = anxn + an 1xn 1 + · · · + a1x + a0 and g(x) = bnxn + bn 1xn 1 + + b1x + b0. Then
φ ( f ( x )+ g ( x )) = φ ( a n
) = φ((an + bn)xn + (an 1 +
+
) = (an + bn)rn + (an 1 + bn 1)rn 1 + · · · +
The analogous argument works for multiplication.
24. By Example 2 of Chapter 15 f (i) = 0 implies f ( i) = 0. Now observe that f (i3 ) = f (i2 i) = f ( i).
25. Since f (2) = 16 4 2 = 10, p = 2 or 5.
26. Multiple an bn + an 1 bn 1 + ·
1 b + a0 = 0 by b n .
27. By observation, U (2) and U (3) are cyclic. If U (p) is not cyclic and p > 3, then by the Fundamental Theorem of Finite Abelian groups there is some prime q such that U (p) has a subgroup isomorphic to Zq ⊕ Zq . But the polynomial xq 1 in Zq [x] has q2 1 zeros, which contradicts Theorem 16.3.
28. For every positive integer m the element 2x2 m 1 has multiplicative order 2.
29. In Z10, let f (x) = 5x Then 0, 2, 4, 6, 8 are zeros.
30. Let n = ab where a > 1, b > 1). If a = b denote the remaining n 2 non-zero elements of Zn by a3 , , an Then every element of Zn is a zero of f (x) = ax(x a3) · · · (x an) and f (x) has degree n 1. If a /= b, denote the remaining n 3 non-zero elements of Zn by a4, . . . , an. Then every element of Zn is a zero of f (x) = ax(x a)(x a4) · · · (x an) and f (x) has degree n 1.
31. If (f (x)/g(x))2 = x, then x2(k(x))2 = x(g(x))2 . But the right side has even degree whereas the left side has odd degree. Alternate solution. Say (f (x)/g(x))2 = x We may assume that f (x) and g(x) have no common factor for, if so, we can cancel them. Since (f (x))2 = x(g(x))2 we see that f (0) = 0. Thus, f (x) has the form xk(x). Then x2(k(x))2 = x(g(x))2 and therefore x(k(x))2 = (g(x))2 . This implies that g(0) = 0. But then f (x) and g(x) have x as a common factor.
32. f (x) = x(x 1)(x 2)(x 3)(x 4) + 1.
33. Suppose that f (x) ∈ D[x] has degree at least 1 and there is a g(x) ∈ D[x] such that f (x)g(x) = 1. Then by Exercise 19 0 = deg f (x)g(x) = deg (f (x) + deg g(x) ≥ 1.
34. (x 1)2 (x + 1).
35. 2x(x 1) has zeros 0, 1, 2, and 4.
36. This follows directly from Theorem 16.5 and Exercise 19.
37. By Theorem 16.5, g(x) = (x 1)(x 2).
38. Suppose that f (x) = amxm + + a0 is an idempotent where am = 0 and m 1. Because the lead term of (f (x))2 is a2 x2m and the lead term of f (x) is amxm we have a2 = 0 and therefore am = 0.
39. First, note that 1 = 16 is zero. Since x9 + 1 = 0 implies x18 = 1 in the group U (17), for any solution a of x9 + 1 = 0 in the group U (17), we know that |a| must divide 18 and |a| must divide |U (17)| = 16. This gives us |a| = 2 and a9 + 1 = a + 1 so that a = 1. Because U (17) is cyclic, 16 is the unique element of order 2.
40. The ring homomorphism from φ from Q[x] to Q(i) = a + bi a, b Q given by φ(f (x)) = f (i) has kernel x2 + 1 . So, by Theorem 15.3 Q[x]/ x2 + 1 is isomorphic to Q(i).
41. Since 1 is a zero of x25 + 1, x + 1 is a factor. Suppose that x25 + 1 = (x + 1)2g(x) for some g(x) Z37[x]. Then the derivative f ′(x) = 25x24 = (x + 1)2 g ′(x) + g(x)2(x + 1). This gives f ′( 1) = 25 = 0, which is false.
42. Let f (x) be a non-constant polynomial of minimum degree with the stated property. Then g(x) = f (x) 2 has five zeros and since Z5 is a field, g(x) has degree 5 and has the same degree as f (x).
43. Since F [x] is a PID, f (x), g(x) = a(x) for some a(x) F [x]. Thus a(x) divides both f (x) and g(x). This means that a(x) is a constant. So, by Exercise 17 in Chapter 14, ⟨f (x), g(x)⟩ = F [x]. Thus, 1 ∈ ⟨f (x), g(x)⟩.
44. Suppose that d1(x) and d2(x) are both greatest common divisors of f (x) and g(x) and monic. Then, By Theorem 16.2, d1(x) = d2(x) . Thus, d1(x) = cd2(x) for some c in F . But then, c = 1.
45. Suppose that I = ⟨f (x)⟩. Then there is some g(x) ∈ Z[x] such that 2 = f (x)g(x). This implies that f (x) = ± 2. But x + 2 ∈ I and is not in ⟨2⟩
46. If f (x) = g(x) for infinitely many elements of F then h(x) = f (x) g(x) has infinitely many zeros. So, by Exercise 16, h(x) = 0.
47. If f (x) /= g(x), then deg[f (x) g(x)] < deg p(x). But the minimum degree of any member of ⟨p(x)⟩ is deg p(x). So, f (x) g(x) does not have a degree. This means that f (x) g(x) = 0.
48. We start with (x 1/2)(x + 1/3) and clear fractions to obtain (6x 3)(6x + 2) as one possible solution.
49. For any positive integer k that are at most k zeros of xk 1. So, there are at most k elements in the field that are solutions to xk = 1.
50. We know f (x) can be written in the form (x a)g(x). Then f ′(x) = (x a)g ′(x) + g(x) and g(a) = f ′(a) /= 0. So, x a is not a factor of g(x).
51. The proof given for Theorem 16.2 with g(x) = x a is valid over any commutative ring with unity. Moreover, the proofs for Corollaries 1 and 2 of Theorem 16.2 are also valid over any commutative ring with unity.
52. Notice that the proof of the division algorithm holds for integral domains when g(x) has the form x a Likewise the proofs of the Factor Theorem and Corollary 3 of Theorem 16.2 hold.
53. Observe that f (x) ∈ I if and only if f (1) = 0. Then if f and g belong to I and h belongs to F [x], we have (f g)(1) = f (1) g(1) = 0 0 and (hf )(1) = h(1)f (1) = h(1) · 0 = 0. So, I is an ideal. By Theorem 16.5, I = ⟨x 1⟩
54. Use the Factor Theorem.
55. For each positive integer k observe that (pxk + 1)( pxk + 1) = 1.
56. Every element in the ideal ⟨x3 x⟩ satisfies the condition.
57. For any a in U (p), ap 1 = 1, so every member of U (p) is a zero of xp 1 1. From the Factor Theorem (Corollary 2 of Theorem 16.2) we obtain that g(x) = (x 1)(x 2) (x (p 1)) is a factor of xp 1 1. Since both g(x) and xp 1 1 have lead coefficient 1, the same degree, and their difference has p 1 zeros, their difference must be 0 (for otherwise their difference would be a polynomial of degree less than p 1 that had p 1 zeros).
58. When n is prime, use Exercise 57. The case n = 4 is done by observation. When n is composite and greater than 4, (n 1)! mod n = 0.
59. By Exercise 58, (p 1)! mod p = p 1. Since p 1 = 1, mod p we have (p 2)! = 1 and the statement follows.
60. The problem is to solve 98! = x mod 101 for x By Exercise 59, modulo 101, we have 1 = 99! = ( 2)98! = 2x. Then, by observation, x = 51 = 50.
61. Let x48 + x21 + a. Since x + 4 = x 1 ∈ Z5[x] from the Factor Theorem we need only find an a in Z5 such that f (1) = 1 + 1 + a = 0. So a = 3.
62. First, note that x 1 ∈ Ker φ Let f (x) ∈ Ker φ Then by Theorem 16.5, x 1 is a factor of f (x). So Ker φ = ⟨x 1⟩. By Theorem 15.3, Q[x]/Ker φ is isomorphic Q.
63. C(x) (field of quotients of C[x]).
64. This follows directly from the definitions.
65. Note that I = ⟨2⟩ is maximal in Z but I[x] is not maximal in Z[x] since I[x] is properly contained in the ideal {f (x) ∈ Z[x]| f (0) is even}.
66. That I[x] is an ideal is straightforward. To prove that I[x] is prime let f (x) = am xm + + a0 and g(x) = bnxn + + b0 and suppose f (x)g(x) I[x]. By filling in with coefficients of 0 we may assume that m = n. We must show that all ai ∈ I or all bi ∈ I. Suppose some bi /∈ I and let k be the least integer such that bk /∈ I. The coefficient xk in f (x)g(x) is akb0 + ak 1b1 + + a0bk and belongs to I. Thus a0bk ∈ I and therefore a0 ∈ I. The coefficient of xk+1 in f (x)g(x) is ak+1b0 + akb1 +
+1 ∈ I Thus, a1bk ∈ I and therefore a1 ∈ I Continuing in this fashion we obtain all ai ∈ I.
67. A solution to x25 1 = 0 in Z37 is a solution to x25 = 1 in U (37). So, by Corollary 2 of Theorem 4.1, |x| divides 25. Moreover, we must also have that |x| divides |U (37)| = 36. So, |x| = 1 and therefore x = 1.
68. Mimic Example 3.
69. By the Factor Theorem (Corollary 2 of Theorem 16.2) we may write f (x) = (x a)g(x). Then f ′(x) = (x a)g ′(x) + g(x). Thus, g(a) = 0 and by the Factor Theorem, x a is a factor of g(x).
70. Observe that the zero polynomial is in I. If f and g I, then (f g)(a) = f (a) g(a) = 0 for all a. If f I and g F [x], then (gf )(a) = g(a)f (a) = 0 for all a. If F = a1, a2, , an then for every non-negative integer k the polynomial xk(x a1)(x a2) (x an ) is in I. If F is infinite and f (x) I, then f (x) has infinitely many zeros. So, by Theorem 16.3, f (x) = 0
71. Say deg g(x) = m, deg h(x) = n, and g(x) has leading coefficient a Let k(x) = g(x) axm n h(x). Then deg k(x) < deg g(x) and h(x) divides k(x) in Z[x] by induction. So, h(x) divides k(x) + axm n h(x) = g(x) in Z[x].
72. The mapping φ(f (x)) = f (x2) is a ring-isomorphism from R[x] onto R[x2].
73. If f (x) takes on only finitely many values, then there is at least one a in Z with the property that f (x) = a for infinitely many x in Z. But then g(x) = f (x) a has infinitely many zeros. This contradicts Corollary 3 of Theorem 16.2.
74. By Theorem 16.5, I = ⟨x(x 1)⟩. In general, if A = {a1, a2, . . . , an} is any finite subset of a field F and I = {f (x) ∈ F [x] | f (a i) = 0 for all a i ∈ A}, then I = ⟨(x a1)(x a2) · · · (x an)⟩
75. Let φ be a ring homomorphism from Z onto a field and let Ker φ = nZ. Then by Theorem 15.3 we have Z/nZ Zn is a field. From Theorem 14.3 we have that nZ is a prime ideal of Z, and from Example 14 in Chapter 14, we know that n is a prime.
76. Say f (x) = anxn + an 1xn 1 + + a1x + a0. Then using the fact that ap = ai for all i and Exercise 49 of Chapter 13, we have f (bp ) = ap (bp )n + ap (bp )n 1 + · · · + ap bp + ap = n p n p p n n 1 p p 1 0 an (b ) +an 1(b 1 )p + +
77. Let f (x) = anxn + an 1xn 1 + + a1x + a0 where a0, a1, . . . an are odd integers and assume that p/q is a zero of f (x) where p and q are integers and n is even. We may assume that p and q are not both even. Substituting p/q for x and clearing fractions we have anpn + an 1pn 1q + + a1pqn 1 = a0qn . If both p and q are odd, then the left side is even since it has an even number of odd terms. If p is even and q is odd, then the left side is even and the left side odd. If p is odd and q is even, then the first term on the left is odd and all the other terms on the left are even. So, the left side is odd and the right side is even. Thus, in each case we have a contradiction.
78. Since x + 4 = x 3 in Z7[x], we have by the Remainder Theorem that the remainder is 351 mod 7. Since 3 is in U (7) we also know that 36 = 1 mod 7. Thus, 351 mod 7 = 34833 mod 7 = 6.
79. By the Division Algorithm (Theorem 16.2) we may write x43 = (x2 + x + 1)q(x) + r(x) where r(x) = 0 or deg r(x) < 2. Thus, r(x) has the form cx + d Then x43 cx d is divisible by x2 + x + 1. Finally, let a = c and b = d.
80. 2x(x 4)
CHAPTER 17
Factorization of Polynomials
1. By Theorem 17.1, f (x) is irre√d ucible o√ver R. Over C we have 2x2 + 4 = 2(x2 + 2) = 2(x + 2i)(x 2i).
2. f (x) factors over D as ah(x) where a is not a unit.
3. If f (x) is not primitive, then f (x) = ag(x), where a is an integer greater than 1. Then a is not a unit in Z[x] and f (x) is reducible.
4. Say r = p/q where p and q are relatively prime. Viewing f (x) as an element of Q[x] we have from the Factor Theorem (Corollary 2 of Theorem 16.2) that f (p/q) = 0. Clearing fractions and collecting all terms and isolating the pn term on one side we see that q divides pn . Using the fact that p and q are relatively prime, we conclude that q = 1.
5. a. If f (x) = g(x)h(x), then af (x) = ag(x)h(x).
b. If f (x) = g(x)h(x), then f (ax) = g(ax)h(ax).
c. If f (x) = g(x)h(x), then f (x + a) = g(x + a)h(x + a).
d. Let f (x) = 8x3 6x + 1. Then f (x + 1) = 8(x + 1)3 6(x + 1) + 1 = 8x3 + 24x2 + 24x + 8 6x 6 + 1 = 8x3 + 24x2 18x + 3. By Eisenstein’s Criterion (Theorem 17.4), f (x + 1) is irreducible over Q and by part c, f (x) is irreducible over Q
6. Use Exercise 5(a).
7. Suppose that r + 1/r = 2k + 1 where k is an integer. Then r2 2kr r + 1 = 0. It follows from Exercise 4 of this chapter that r is an integer. But the mod 2 irreducibility test shows that the polynomial x2 (2k + 1)x + 1 is irreducible over Q and an irreducible quadratic polynomial cannot have a zero in Q
8. Use the Mod p Test with 3, 2, 5, and 7 respectively.
9. Use Exercise 5a and clear fractions.
10. By Corollary 1 of Theorem 17.5 we know the set is a field. To see that it has pn elements, note that any element g(x) + ⟨f (x)⟩ = f (x)q(x) + r(x) + ⟨f (x)⟩ = r(x) + ⟨f (x)⟩ where r(x) = 0 or deg r(x) < n So, all cosets have the form an 1xn 1 + + a0 + ⟨f (x)⟩ and they are all distinct.
11. It follows from Theorem 17.1 that p(x) = x2 + x + 1 is irreducible over Z5. Then, from Corollary 1 of Theorem 17.5, we know that Z5[x]/⟨p(x)⟩ is a field. To see that this field has order 25, note that if f (x) + ⟨p(x)⟩ is any element of Z5[x]/⟨p(x)⟩, then by the Division Algorithm (Theorem 16.2) we may write f (x) + ⟨p(x)⟩ in the form
p(x)q(x) + ax + b + ⟨p(x)⟩ = ax + b + ⟨p(x)⟩. Moreover, ax + b + ⟨p(x)⟩ = cx + d + ⟨p(x)⟩ only if a = c and b = d, since (a c)x + b d is divisible by ⟨p(x)⟩ only when it is 0. So, Z5[x]/⟨p(x)⟩ has order 25.
12. Find an irreducible cubic over Z3 and mimic Example 10. One such cubic is x3 + x2 + 2 (by the Mod 3 Test).
13. a. Irreducible by Eisenstein
b. Irreducible by the Mod 2 Test (but be sure to check for quadratic factors as well as linear)
c. Irreducible by Eisenstein
d. Irreducible by the Mod 2 Test
e. Irreducible by Eisenstein (after clearing fractions)
14. Note that 1 is a zero. No, since 4 is not a prime.
15. x; x + 1
| | | | | | | |
16. Note that F ∗ = 7 and a divides 7. Yes, for F = 32, but no for F = 16. (Because 31 is prime but 15 is not.)
17. f (x) is irreducible over Q. Nothing.
18. If f (x) is reducible over Z2 and does not have 0 or 1 as a zero then it must factor as an irreducible quadratic and an irreducible cubic. But the only irreducible quadratic over Z2 is x2 + x + 1.
19. We first observe that the multiplicative group of the field has order 48. So, the nonzero elements have orders that divide 48. To find |x| we use the fact that mod I, x2 = 2 = 5 mod 7. Since x2 = 5, we have that x3 = 5x and x6 = 4x2 = 8 = 1. So, x12 = 1. To find |x + 1|, we note that (x + 1)2 = x2 + 2x + 1 = 2x 1; (x + 1)4 = (2x 1)2 = 4x2 4x + 1 = 4x; (x + 1)8 = 2x2 = 3. Since observe that |3| = 6. So, |x + 1| = 48. Because |x| = 12, we know that x 1 = x11 = x6 x3 x3 = 1(5x)(5x) = 4x = 3x. To check this we note that x(3x) = 3x2 = 6 = 1.
20. Let f (x) = g(x)h(x) where deg g(x) < deg f (x) and deg h(x) < deg f (x).
21. Let f (x) = x4 + 1 and g(x) = f (x + 1) = x4 + 4x3 + 6x2 + 4x + 2. Then f (x) is irreducible over Q if g(x) is. Eisenstein’s Criterion shows that g(x) is irreducible over Q. Alternate proof. Since x4 + 1 has no real zeros, the only possible factorizations over Z are x4 + 1 = (x2 + ax + 1)(x2 + bx + 1) or x4 + 1 = (x2 + ax 1)(x2 + bx 1). Evaluating x4 + 1 = (x2 + ax + 1)(x2 + bx + 1) at 1 gives us 2 = (a + 2)(b + 2). So, one of a or b is 0. But long division shows x2 + 1 is not a factor of x4 + 1. Evaluating x4 + 1 = (x2 + ax 1)(x2 + bx 1) at 1 gives us 2 = ab. So, one of a or b is 1. But long division shows x2 + x 1 is not a factor of x4 + 1.
22. Observe that if f (x) = p(x)q(x), then f (x) = f ( x) = p( x)q( x) and p( x), q( x) ∈ Z[x].
23. (x + 3)(x + 5)(x + 6).
24. (x + 1)3 .
25. By the Mod 2 Irreducibility Test (Theorem 17.3 with p = 2) it is enough to show that x4 + x3 + 1 is irreducible over Z2 By inspection, x4 + x3 + 1 has no zeros in Z2 and so it has no linear factors over Z2. The only quadratic irreducible in Z2[x] is x2 + x + 1 and it is ruled out as a factor by long division.
26. For f (x), both methods yield 4 and 5. (Notice that √ 47 = √ 2 = ±3). Neither m√ e t h o d y√ i e lds a solution for g(x). The quadratic formula applied to g(x) involves 23 = 2 and there is no element of Z5 whose square is 2. ax2 + bx + c (a /= 0) has a zero in Zp [x] if and only if b2 4ac = d2 for some d in Zp.
27.
a. Since every reducible polynomial of the form x2 + ax + b can be written in the form (x c)(x d), we need only count the number of distinct such expressions over Zp Note that there are p(p 1) expressions of the form (x c)(x d) where c /= d. However, since (x c)(x d) = (x d)(x c) there are only p(p 1)/2 distinct such expressions. To these we must add the p cases of the form (x c)(x c). This gives us p(p 1)/2 + p = p(p + 1)/2.
b. First, note that for every reducible polynomial of the form f (x) = x2 + ax + b over Zp the polynomial cf (x) (c /= 0) is also reducible over Zp. By part a, this gives us at least (p 1)p(p + 1)/2 = p(p2 1)/2 reducible polynomials over Zp Conversely, every quadratic polynomial over Zp can be written in the form cf (x) where f (x) has lead coefficient 1. So, the p(p2 1)/2 reducibles we have already counted include all cases.
28. Use Exercise 27.
29. By Exercise 28, for each prime p there is an irreducible polynomial p(x) of degree 2 over Zp. By Corollary 1 of Theorem 17.5, Zp[x]/⟨p(x)⟩ is a field. By the Division Algorithm (Theorem 16.2) every element in Zp[x]/⟨p(x)⟩ can be written in the form ax + b + ⟨p(x)⟩ Moreover, ax + b + ⟨p(x)⟩ = cx + d + ⟨(p(x)⟩ only when a = c and c = d since (ax + b) (cx + d) is divisible by p(x) only when it is 0. Thus, Zp[x]/⟨p(x)⟩ has order p2 .
30. By Eisenstein, xn + p where p is a prime is irreducible over Q
31. Consider the mapping from Z3[x] onto Z3[i] given by φ(f (x)) = f (i). Since φ(f (x) + g(x)) = φ((f + g)(x)) = (f + g)(i) = f (i) + g(i) = φ(f (x)) + φ(g(x)) and φ(f (x)g(x)) = φ((fg)(x) = (fg)(i) = f (i)g(i) = φ(f (x))φ(g(x)), φ is a ring homomorphism. Because φ(x2 + 1) = i2 + 1 = 1 + 1 = 0 we know that x2 + 1 ∈ Ker φ From Theorem 16.4 we have that Ker φ = ⟨x2 + 1⟩ Finally, Theorem 15.3 gives us that Z3[x]/⟨x2 + 1⟩ ≈ Z3[i].
32. If a2 x2 + a1 x + a0 Zp [x] is a factor of f (x) with a2 = 0 then a2 1(a2 x2 + a1 x + a0 ) is a factor of f (x).
33. x2 + 1, x2 + x + 2, x2 + 2x + 2.
34. If so, then π is a zero of x2 ax b.
35. We know that an(r/s)n + an 1(r/s)n 1 + + a0 = 0. So, clearing fractions we obtain anrn + san 1rn 1 + + sna0 = 0. This shows that s anrn and r sna0 By Euclid’s Lemma (Chapter 0), s divides an or s divides rn . Since s and r are relatively prime, s must divide an Similarly, r must divide a0
36. Since a1(x)a2(x) ak(x) = a1(x)(a2(x) · ak(x)), we have by Corollary 2 of Theorem 17.5 that p(x) divides a1(x) or p(x) divides a2(x) ak(x). In the latter case, the Second Principle of Mathematical Induction implies that p(x) divides some ai(x) for i = 2, 3, . . . , k.
37. Suppose that p(x) can be written in the form g(x)h(x) where deg g(x) < deg p(x) and deg h(x) <deg p(x) with g(x), h(x) F [x]. By Theorem 14.4, F [x]/ p(x) is a field with 0 + ⟨p(x)⟩ = p(x) + ⟨p(x)⟩ = g(x)h(x) + ⟨p(x)⟩ = (g(x) + ⟨p(x)⟩)(h(x) + ⟨p(x)⟩).
Thus g(x) + ⟨p(x)⟩ = 0 + ⟨p(x)⟩ or h(x) + ⟨p(x)⟩ = 0 + ⟨p(x)⟩ This implies that g(x) ∈ ⟨p(x)⟩ or h(x) ∈ ⟨p(x)⟩. In either case we have contradicted Theorem 16.4.
38. Use the corollary to Theorem 17.4 and Exercise 5b with a = 1.
39. Since (f + g)(a) = f (a) + g(a) and (f g)(a) = f (a)g(a), the mapping is a homomorphism. Clearly, p(x) belongs to the kernel. By Theorem 17.5, ⟨p(x)⟩ is a maximal ideal, so the kernel is ⟨p(x)⟩.
40. It follows from Theorem 17.1 that x2 + 1 is irreducible over Z and from Theorem 17.6 that if f (x)g(x) x2 + 1 where f (x), g(x) Z[x] then x2 + 1 is a factor of f (x) or g(x). So, x2 + 1 is a prime ideal. For the second part, let I = x2 + 1 and observe that if I is a maximal ideal of Z[x] then, by Theorem 14.4, Z[x]/I = ax + b + I a, b Z is field. Thus, the element 2 + I has a multiplicative inverse in Z[x]/I, say (2 + I)(ax + b) + I = 2(ax + b) + I = 1 + I. Then, 2(ax + b) 1 is in I. Since every non-zero element in I has degree at least 2 and the left side is 0 or has degree at 1, we have that 2(ax + b) 1 = 0. But then the left side is odd.
41. Consider the mapping φ from F to F [x]/ p(x) given by φ(a) = a + p(x) . By observation, φ is one-to-one and onto. Moreover, φ(a + b) = a + b + ⟨p(x)⟩ = a + ⟨p(x)⟩ + b + ⟨p(x)⟩ = φ(a) + φ(b) and φ(ab) = ab + ⟨p(x)⟩ = (a + ⟨p(x)⟩)(b + ⟨p(x)⟩) = φ(a)φ(b) so φ is a ring isomorphism.
42. Let f (x) = g(x)h(x) where deg g(x) < deg f (x) and deg h(x) < deg f (x). Then g(x)h(x) belongs to ⟨f (x)⟩ but neither g(x) nor h(x) belongs to ⟨f (x)⟩.
43. f (x) is primitive.
44. 11
45. Since x2 + x + 1 is the only quadratic irreducible over Z2, the possibilities are x2(x2 + x + 1), (x + 1)2(x2 + x + 1), and x(x + 1)(x2 + x + 1).
46. Note that by Exercise 49 in Chapter 13, (x + 2)3 = (x 1)3 .
47. Let f (x) = (5/7)x4 + (9/2)x3 + (3/4)x2 + 6x + 1/2. Then, clearing fractions, we have 28f (x) = 20x4 + 14 9x3 + 21x2 + 28 6x + 14. This polynomial meets the conditions for Eisenstein’s Criterion with the prime 7. So, 28f (x) is irreducible over Q. But then by Exercise 5a, f (x) is also irreducible over Q
48. By Eisenstein, p(x) = x5 + 15x3 20x 10 is irreducible over Z. Suppose that g(x)h(x) is in p(x) Then g(x)h(x) = p(x)k(x) for some k(x) in Z[x]. By the unique factorization over Z property, we have that p(x) must be one of the irreducible factors of g(x)h(x). So, one of g(x) or h(x) has p(x) as a factor and therefore, one of them is in ⟨p(x)⟩.
49. If a ∈ Zp is a zero of x2 + x + 1, then a3 = 1. So, |a| = 3, and therefore 3 divides |U (p)| = p 1. This gives us that 3 divides both p 1 and p 2, But then 3 must divide (p 1) (p 2) = 1.
50. By Theorem 17.3, we can say that f (x) is irreducible over Q, provided that the lead coefficient is not divisible by p.
51. The general result is that x, n is prime ideal in Z[x] if and only if n = 0 or n is a prime.
⟩ { ∈ | }
To see this note that I = x, n = f (x) Z[x] f (0) = n If n = ab where 1 < a < n, g(x) = a and h(x) = b have the property that g(x)h(x) belongs to I but g(x) and h(x) do not.
52. The degree is at least 3 or f (x) g(x) = 0, and so f (x) g(x) has no degree.
CHAPTER 18
Divisibility in Integral Domains
1. 1. a2 db2 = 0 implies a2 = db2 . Thus a = 0 = b, since otherwise d = 1 or d is divisible by the square of a prime.
2. N ((a + b √ d)(a ′ + b′ √ d)) = N (aa ′ + dbb′ + (ab′ + a ′b) √ d) = |(a2 db2)(a ′2 db′2)| = |(aa ′ + dbb′)2 d(ab′ + a ′b)2| = |a2a ′2 +d2b2b′2 da2b′2 da′2b2| = |a2 db2||a ′2 db′2| = N (a+b √ d)N (a ′ +b′ √ d).
3. If xy = 1√ , then 1 = N (1) = N (xy) = N (x)N ( y√) and N ( x√) = 1 = N ( y )√ If N (a + b unit. d) = 1, then ±1 = a2 db2 = (a + b d)(a b d) and a + b d is a
4. This part follows directly from 2 and 3.
2. Say a = bu where u is a unit. Then ra = rbu = (ru)b ∈ ⟨b⟩ so that ⟨a⟩ ⊆ ⟨b⟩. By symmetry, ⟨b⟩ ⊆ ⟨a⟩. If ⟨a⟩ = ⟨b⟩, then a = bu and b = av. Thus, a = avu and
uv = 1.
3. Let I = ∪Ii. Let a, b ∈ I and r ∈ R. Then a ∈ Ii for some i and b ∈ Ij for some j. Thus a, b ∈ Ik, where k = max{i, j} So, a b ∈ Ik ⊆ I and ra and ar ∈ Ik ⊆ I
4. Say r is irreducible and u is a unit. If ru = ab where a and b are not units, then r = a(bu 1) where a and bu 1 are not units.
5. Clearly, ⟨ab⟩ ⊆ ⟨b⟩. So the statement is equivalent to ⟨ab⟩ = ⟨b⟩ if and only if a is a unit. If ⟨ab⟩ = ⟨b⟩ there is an r in the domain such that b = rab, so that 1 = ra and a is a unit. If a is a unit then b = a 1(ab) belongs to ⟨ab⟩ and therefore ⟨b⟩ ⊆ ⟨ab⟩.
6. a a since a = a 1; if a b, say a = bu where u is a unit, then b = au 1 so b a; if a b, say a = bu where u is a unit and b c, say b = cr where r is a unit, then a = bu = cru where ru is a unit. ⟨ab⟩ = ⟨b⟩. So, a is not a unit.
7. Say x = a + bi and y = c + di. Then xy = (ac bd) + (bc + ad)i.
On the other hand, d(x)d(y) = (a2 + b2 )(c2 + d2 ) = a2 c2 + b2 d2 + b2 c2 + a2d2
8. First, observe that for any r D, d(1) d(1 r) = d(r) so that d(1) is the minimum value of d Now if u is a unit, then d(u) d(uu 1) = d(1) so that d(u) = d(1). If d(u) = d(1), then 1 = uq + r where r = 0 or d(r) < d(u) = d(1). So r = 0.
9. Suppose a = bu, where u is a unit. Then d(b) ≤ d(bu) = d(a). Also, d(a) ≤ d(au 1) = d(b). | |
10. Mimic the proof of Theorem 17.5.
11. Suppose that x = a + b √ d is a unit in Z[ √ d]. Then 1 = N (x) = a2 + ( d)b2 . But d > 1 implies that b = 0 and a = ±1.
12. Suppose that D has a proper ideal I1 that is not contained in a maximal ideal. Let a1 be in D but not in I1 Then I2 = ⟨I1, a1⟩ is a proper ideal that properly contains I1 and is not maximal. Repeating this argument we have a strictly increasing chain of ideals I1 I2 . So, by the Ascending Chain Condition, this chain is finite. But then the last ideal in the chain is maximal. Alternate solution. Suppose that D has a proper ideal I1 that is not contained in a maximal ideal. By definition, there is some proper ideal I2 that properly contains I1 but is not a maximal ideal. Repeating this argument we have a strictly increasing chain of ideals I1 I2 . So, by the Ascending Chain Condition, this chain is finite. But then the last ideal in the chain is maximal.
13. First, observe t h a t √ 2 1 = 3 · 7 and that 21 = (1 + 2 √ 5)(1 2 √ 5). To prove that 3 is irreducible in Z[ 5] suppose that 3 = xy, where x, y ∈ Z[ √ 5] and x and y are not units. Then 9 = N (3) = N (x)N (y) and, therefore, N (x) = N (y) = 3. But there are no integers a a √ n d b such that a2 + 5b2 = √ 3. The same argument s h o√w s that 7 is irreducible√o ver Z[ 5]. To show t h a√t 1 + 2 5 is irreducible over Z[ 5] suppose that 1 + 2 √ 5 = xy, where x, y ∈ Z[ 5] and x and y are not units. Then 21 = N (1 + 2 impossible. 5) = N (x)N (y). Thus N (x) = 3 or N (x) = 7, both of which are
14. Suppose that there are Gaussian integers z1 and z2 such that 1 i = z1z2. Then 2 = d(1 i) = d(z1)d(z2). This forces d(z1) = 1 or d(z2) = 1. Now use Exercise 8.
15. First, observe t h a t √ 1 0 = 2 · 5 and that 10 = (2 √ 6)(2 + √ √6). To see that 2 is irreducible over Z[ 6], assume that 2 = xy, where x, y ∈ Z[ 6] and x and y are not units. Then 4 = N (2) = N (x)N (y) so that N (x) = 2. But 2 c a√n n o t be written in the form a + 6b2 . √ A similar argument applies√t o 5. To see that 2 6 is irreducible, suppose that√2 6 = xy where x, y ∈ Z[ 6] and x and y are not units. Then 1 0 √= N (2 6) = N (x)N (y) and as before, this is impossible. We know that Z[ 6] is not a principle ideal domain because a PID is a UFD (Theorem 18.3).
16. C[x] is a UFD but contains Z[ √ 6].
17. Suppose 3 = αβ, where α, β Z[i] and neither is a unit. Then 9 = d(3) = d(α)d(β), so that d(α) = 3. But there are no integers such that a2 + b2 = 3. Observe that 2 = i(1 + i)2 and 5 = (1 + 2i)(1 2i) and 1 + i, 1 + 2i, and 1 2i are not units.
18. If 7 = (a + b √ 6)(c + d √ 6) and neither factor is a unit, then |a2 6b2 | = 7 and, modulo 7, a2 = 6b2 . H o w e v e √ r the only solutions to this equation modulo 7 are a = b = 0. In this case c + d 6 is a unit.
19. Use Exercise 1 with d = 1. 5 and 1 + 2i; 13 and 3 + 2i; 17 and 4 + i.
20. Use Example 1 and Theorem 18.2.
21. Suppose that √ 1 + 3 √ 5 = xy, where x, y ∈ Z[ √ 5] and x and y are not units. Then 46 = N (1 + 3 5) = N (x)N (y). Thus, N (x)√= 2 or N (x) = 23. But n √ e i t h e r 2 nor 5 can be w r i√ t t en in the form a2 + 5b2 , so 1 + 3 √ 5 is i r r e du c i√ b l e over Z[ 5]. To see that 1 + 3√ 5 is not prime, observe t h √ a t (1 + 3 5)(1 3 5) = 1 + 45 = 4 6√ s o that 1 + 3 5 divides 2 · 23. For 1 + 3 5 to divide 2, we need 46 = N (1 + 3 5)
divides N (2) = 4. Likewise, for 1 +√3 √ 5 to divide 23 we need that 46 divides 232 . Since neither of these is true, 1 + 3 5 is not prime.
22. To show the two elements are irreducible mimic the solution given for Exercise 18 but use modulo √ 4. To show√the two elements are not prime use the o b √ s e rvation that 2 2 = 4 = (1 + 5)( 1 + 5). A l √ s o note that 2 does not √ d i v i d e (1 + 5) for if so, there would be an element z in Z[ 5] √such that 2z = 1 + 5. But N (2z) = N ( 2 ) N√ ( z ) = 4N (z) = N (1 + 5) = 4. Thus N (z) = 1. But the only elements of Z[ 5√ ] with norm 1 are 1 √ a n d 1. A s i m√i lar argument shows that 2 does not divide 1 = 5 and, neither 1 + 5 nor 1 + 5 divides 2.
23. First, observe that ( 1√+ √ 5)(1 + √ 5) = 4 = √ 2 · 2 and by Exercise 22 , √1 + √ 5 and 2 are i r r e du c√ i b le over Z[ 5]. To see t√ h a t 1 + 5 is irreducible over Z[ 5], s up p o√ s e that 1 + 5 = xy√ where x, y ∈ Z[ 5] and x and y are not units. Let x = a + b 5. Then 4 = N ( 1 + 5) = N (x)N (y) so that a2 5b2 = ± 2. Viewing this equation modulo 5 gives us a2 = 2 or a2 = 2 = 3. However, every square in Z5 is 0, 1, or 4.
24. Let I be a non-zero prime ideal in F [x] and let M be a proper ideal that contains I. By Theorem 16.3 there are elements f (x) and g(x) in F [x] such that M = f (x) and I = g(x) . By definition g(x) is a prime and by Theorem 18.2 it is irreducible over F . Since M contains I there is an h(x) in F [x] such that g(x) = f (x)h(x) and since g(x) is irreducible over F, h(x) is a unit. Thus by Exercise 2 in this chapter M = I.
28. To see that 2 is prime in Z12 note that 2 is not a unit in Z12 and suppose that bc = 2t where b, c and t belong to Z12. Thus there is an integer k such that bc 2t = 12k in the ring Z. This implies that 2 divides bc in Z. Thus one of b or c is divisible by 2 in Z12 The same argument applies when 2 is replaced by 3. To see that 2 is irreducible in Z12 , suppose that 2 = bc in Z12 . Then there is an integer k such that 2 bc = 12k in Z. By Euclid’s Lemma this implies that 2 divides b or c in Z Say, b = 2d Then in Z we have 2 2dc = 12k which reduces to 1 = dc + 6k This implies that c is relatively prime to 6 and therefore a unit in Z12. But the only solutions to the equation 2 = 2x in Z12 are x = 1 and x = 7. If dc = 1 then c is a unit and if dc = 7 then in Z12, 7dc = 7 · 7 mod 12 = 1 and c is again a unit. To see that 3 is reducible in Z12 note that 3 = 3 9 mod 12 and neither 3 nor 9 is a unit.
29. Suppose that bc = pt in Zn Then there exists an integer k such that bc = pt + kn This implies that p divides bc in Z and by Euclid’s Lemma we know that p divides b or p divides c.
30. If p2 does not divide n then there are integers s and t such that 1 = ps + (n/p)t. Thus, p = p(ps) mod n. Since both p and ps are zero-divisors in Zn they cannot be units. This contradicts our assumption that p is irreducible in Zn.
31. See Example 3.
32. If (a + bi) is a unit, then a2 + b2 = 1. Thus, ±1, ±i.
33. Note that p|(a1a2 · · · an 1)an implies that p|a1a2 · · · an 1 or p|an Thus, by induction, p divides some ai.
34. 4x + 1 and 2x + 3 are associates of 3x + 2 and x + 4.
35. By Exercise 10, ⟨p⟩ is maximal and by Theorem 14.4, D/⟨p⟩ is a field.
36. Let a be a nonzero element of the domain. It suffices to show that a is a unit. Consider the chain ⟨a⟩ ⊇ ⟨a2 ⟩ ⊇ ⟨a3 ⟩ ⊇ ⟨a4 ⟩ ⊇ . By hypothesis, we have ⟨an⟩ = ⟨an+1⟩ for some n. Thus an = an+1b for some b and 1 = ab.
37. Suppose R satisfies the ascending chain condition and there is an ideal I of R that is not finitely generated. Then pick a1 ∈ I. Since I is not finitely generated, ⟨a1⟩ is a proper subset of I, so we may choose a2 ∈ I but a2 ∈ / ⟨a1⟩ As before, ⟨a1, a2⟩ is proper, so we may choose a3 ∈ I but a3 ∈ / ⟨a1, a2⟩. Continuing in this fashion, we obtain a chain of infinite length ⟨a1⟩ ⊂ ⟨a1, a2⟩ ⊂ ⟨a1, a2, a3⟩ ⊂ · · ·
Now suppose every ideal of R is finitely generated and there is a chain I1 ⊂ I2 ⊂ I3 ⊂ · · · Let I = ∪Ii Then I = ⟨a1, a2, , an⟩ for some choice of a1, a2, . . . , an. Since I = Ii, each a i belongs to some member of the union, say Ii′ . Letting k = max i′ i = 1, , n , we see that all a i Ik Thus, I Ik and the chain has length at most k.
38. The set of c o m√p le x numbers is a Euclidean domain (take d(a) = 0 for all a /= 0) containing Z[ Theorem 18.4. 5] as a subdomain. Now use Example 7 and the corollary to
39. Say I = ⟨a + bi⟩. Then a2 + b2 + I = (a + bi)(a bi) + I = I and therefore a2 + b2 ∈ I For any c, d ∈ Z, let c = q1(a2 + b2) + r1 and d = q2(a2 + b2) + r2, where 0 ≤ r1, r2 < a2 + b2 . Then c + di + I = r1 + r2 i + I.
40. 1 + √ 2; infinite.
41. N (6 + 2 √ 7) = 64 = N (1 + 3 √ 7). The other part follows directly from Exercise 25.
42. Let Ii = {(a1, a2, , ai, 0, 0, 0, )}, where the a’s are integers.
43. Theorem 18.1 shows that primes are irreducible. So, assume that a is an irreducible in a UFD R and that a bc in R. We must show that a b or a c. Since a bc, there is an element d in R such that bc = ad Now replacing b, c, and d by their factorizations as a product of irreducibles, we have by the uniqueness property that a (or an associate of a) is one of the irreducibles in the factorization of bc Thus, a is a factor of b or a is a factor of c.
44. Observe that both x2 and x3 are irreducible over F but x3x3 = x2x2x2 .
45. See Exercise 23 in Chapter 0
46. The argument given in Example 7 of Chapter 18 applies in both cases.
CHAPTER 19
Extension Fields
1. {a52/3 + b51/3 + c | a, b, c ∈ Q}.
2. Observe that √ 1 √ = 1 √ 2 √ 3 = √ 2 + √ 3 is in Q( √ 2 + √ 3). So,
(√2 + √3 + ( √ 2 +√ 3) = 2 √ 3 √ i s in Q√ ( 2 + 3). Then, 3 and ( 2 + 3) 3 = 2 are in Q( 2 + 3).
Alternate s o l u t i o n √O√ b s e r v e that ( √ 2 + √ 3)2 = 5 + 2 √ 2 √ 3 is in Q( √ 2 + √ 3). From t√ his √we h√ave th√at 2 √ 3 is in √Q ( √ 2 + √ 3). √ T h e n , √ √ ( √ 2 3)(√ 2 + 3√) = 2 √3 + 3 √2 = 3 √ 2 + √2 3 i√ s in Q( 2 + √ 3). √ Mo r e o v√ e r , √ 3( 2 √ + 3)√ (3 2 + 2 √3 ) = √3 is in Q( √2 +√ 3). Then, ( 2 + 3) 3 = 3 is in Q( 2 + 3). Finally, 2 + 3 is in Q( 2, 3) by closure.
3. Since x3 1 = (x 1)(x2 + x + 1), the zeros of x3 1 are 1, ( 1 + √ 3)/2, and
⊆ ⊆ ( 1 √ 3)/2. So, the splitting field is Q( √ 3).
4. F (a, b) F (a)(b) since F (a)(b) is a field that contains F , a and b. Also, F (a)(b) F (a, b) since F (a, b) contains F (a) and b So, F (a, b) = F (a)(b) and, by symmetry, F (a, b) = F (b, a).
5. Since√the zeros of x2 + x + 1 are ( 1 √± √ 3)/2 and the zeros of x2 x + 1 are (1 ± 3)/2, the splitting field is Q( 3).
6. 22/3 = (21/3)2 ∈ Q(21/3) by closure. So, Q(22/3) ⊆ Q(21/3). By closure, 2 · 21/3 = (22/3)2 ∈ Q(22/3). So, Q(21/3) ⊆ Q(22/3).
7. Since ac + b ∈ F (c), we have F (ac + b) ⊆ F (c). Also, c = a 1(ac + b) a 1b, so F (c) ⊆ F (ac + b).
8. 8. Use Theorem 19.3. To construct the multiplication table, observe that a3 = a + 1.
9. Since a3 + a + 1 = 0, we have a3 = a + 1. Thus, a4 = a2 + a; a5 = a3 + a2 = a2 + a + 1. To compute a 2 and a100 , we observe that a7 = 1, since F (a)∗ is a group of order 7. Thus, a 2 = a5 = a2 + a + 1 and a100 = (a7)14a2 = a2 .
10. Use the fact that a3 = a + 1, a4 = a2 + a, a5 = a2 + a + 1, a6 = a2 + 1 and a7 = 1.
11. Q(π) is the set of all expressions of the form (anπn + an 1πn 1 + · · · + a0)/(bm πm + bm 1πm 1 + · · · + b0), where bm /= 0.
12. A basis is {1, π, π2 }. 1/π = (1/π3 )π2 ; π11 = ((π3)3)π2 .
13. x7 x = x(x6 1) = x(x3+1)(x3 1) = x(x 1)3(x+1)3; x10 x = x(x9 1) = x(x 1)9 (see Exercise 49 of Chapter 13).
14. Suppose that φ is an automorphism of Q( √ 5). Since φ(1) = 1, we have φ(n) = φ(n 1) = nφ(1) = n. Also, 1 = φ(n/n) = nφ(1/n) gives φ(1/n) = 1/n. Thus, φ(m/n) = mφ(1/n) = m/n. So φ is the identity map on Q. Lastly, √ 2 √ 2
5 = φ√(5 ) = φ( 5 ) = (φ( 5))√ , so φ( 5) = ± 5. So there are two automorphisms of Q( 5). For the case of Q( 3 5) we have that φ is the identity map on Q and 5 = φ(5) = φ ( √ 3 5 ) =√( φ ( √ 3 5))3 , so φ ( √ 3 5) = √ 3 5. So there is only the identity automorphism of Q( 3 5).
15. If f (x) is irreducible over F we are done. If f (x) does not split in F , then f (x) = g(x)h(x) where g(x), h(x) F [x], 1 deg g(x) < p, and g(x) is irreducible over F . Let b be a zero of f (x) in some extension of F . Then bp = a and f (x) = xp bp = (x b)p (see Exercise 49 of Chapter 13). If b F , then f (x) splits in F ; if b F , then deg g(x) > 1 and has multiple zeros. So, by Theorem 19.6, we know that g(x) = k(xp ) for some k(x) in F [x]. But then g(x) has degree at least p. Alternate proof. Let g(x) and b be as in the first proof. Because f (x) = (x b)p we know that g(x) = (x p)k for some 1 < k < p. In the expanded product, the coefficient of xk 1 is kb and since g(x) F [x] we have that kb F . Since k < p, we know that k is in F Then k 1 kb = b is in F and f (x) splits in F
3 2) for a, b, and c yields a = 4/3, b = 2/3, and c = 5/6.
18. a = 3/23, b = 4/23.
19. By Exercise 7, Q(4 i) = Q and Q(1 + i) = Q. Alternate solution. Since 1 + i = (4 i) + 5, Q(1 + i) ⊆ Q(4 i); conversely, 4 i = 5 (1 + i) implies that Q(4 i) ⊆ Q(1 + i).
20. Note that a = 1 + √ 5 implies that a4 2a2 4 = 0. Then p(x) = x4 2x2 4 is irreducible over Q. To see this, use the mod 3 test on x4 2x2 + 2 Substitution shows this has no zeros. By Example 8 in Chapter 17, the only quadratic factors we need check as factors are x2 + 1, x2 + x + 2 and x2 = 2x + 2. Long division rules these out.
21. If the zeros of f (x) are a1, a2, . . . , an, then the zeros of f (x + a) are a1 a, a2 a, . . . , an a So, by Exercise 7, f (x) and f (x a) have the same splitting field.
22. Since f (x) and g(x) are relatively prime in F [x] there are polynomials t(x) and s(x) in F [x] such that 1 = f (x)t(x) + g(x)s(x). (See Exercise 43 for Chapter 16.) If f (x) and g(x) had a nonconstant factor in common this factor would divide 1.
23. C l e√ a r l y , Q and Q( √ 2) are subfields o f √Q ( √ 2). Assume that there is a subfield F of Q(√2) that contains an element a + b 2 with b /= 0. Th√en, since every√subfield of Q( 2) m√ u s t contain Q, we have by Exercise 20 that Q( 2) = Q(a + b 2) ⊆ F So, F = Q( 2).
24. They are of the form a + b √ 4 2 where a, b ∈ Q( √ 2).
25. It is 64. To see this, we let a be a zero in the splitting field of x2 + x + 1 over Z2 Then x2 + x + 1 = (x a)(x a 1). Checking to see that none of the four elements of F (a) is a zero of x3 + x + 1, we know that x3 + x + 1 is irreducible over F (a).
Then letting b be a zero of x3 + x + 1 in the splitting field, we know that F (a)(b) is a field of order 64 in which x3 + x + 1 splits.
26. Following Example 8, we first observe that i is a primitive 4th root of unity. Then the splitting field is Q( 4 1, i) = Q( 4 1) since ( 4 1)2 = i.
27. Let a be a zero of x3 + 2x + 1 in some extension of Z3 By Theorem 19.3, F (a) = {c2a2 + c1a + c0 | where c2, c1, c0 ∈ Z3} and we know that a3 + 2a + 1 = 0. Using long division, we obtain x3 + 2x + 1 = (x a)(x2 + ax + (2 a2)). By trial and error, we discover that a + 1 is a zero of x2 + ax + (2 a2) and, by long division, we deduce that 2a 1 is the other zero of x2 + ax + (2 a2). So, we have x3 + 2x + 1 = (x a)(x a 1)(x + 2a + 1).
28. Over Z2, x6 + 1 = x6 1 = (x3 1)(x3 + 1) = (x3 1)2 = (x 1)2(x2 + x + 1)2 = (x + 1)2(x2 + x + 1)2 . Over Z3, we use Exercise 49 in Chapter 13 to obtain x6 + 1 = (x2)3 + 13 = (x2 + 1)3 .
29. Suppose that φ: Q( √ 3) → Q( √ 3) is√a n i s√o m o r ph is m √Since φ(1) = 1, we have 2 φ( 3) =√ 3. Then 3 = φ( 3) = φ( 3 3) = (φ( 3)) This is impossible, since φ( 3) is a real number.
30. The field of quotients of Zp [x] is not perfect.
31. By long division we obtain x2 + x + 2 = (x β)(x + β + 1), so the other zero is β 1 = 4β + 4.
32. Use Theorem 19.5.
33. Since f (x) = x21 + 2x8 + 1 and f ′(x) = x7 have no common factor of positive degree, we know by Theorem 19.5 that f (x) has no multiple zeros in any extension of Z3
34. Observe that because 1 is a zero of both f (x) = x19 + x8 + 1 and f ′(x) = 19x18 + 8x7 = x18 + 2x7 we know by Corollary 2 of Theorem 16.2 (the Factor Theorem) and Theorem 19.5 that f (x) has multiple zeros in some extension of Z3 By the proof on Theorem 19.5, 1 is a multiple zero.
35. Since f (x) = xp n x and f ′(x) = 1 have no common factor of positive degree, we know by Theorem 19.5 that f (x) has no multiple zeros in any extension of Z3
36. The splitting field is F = Z3[x]/⟨x2 + x + 2⟩ and β is a zero of x2 + x + 2 in F F has nine elements and f (x) = (x β)(x (2β + 2))(x 2β)(x (β + 1)).
37. Let K be the intersection of all the subfields of E that contain F and the set a1, a2, . . . , an . It follows from the subfield test given in Exercise 29 Chapter 13 that K is a subfield of E and, by definition, K contains F and the set {a1, a2, . . . , an}. Since F (a1, a2, . . . , an) is the smallest such field, we have F (a1, a2, . . . , an) ⊆ K. Moreover, since the field F (a1, a2, . . . , an) is one member of the intersection, we have K ⊆ F (a1, a2, . . . , an ) ⊆ K.
}
38. Observe that x4 x2 2 = x4 + 2x2 + 1 = (x2 + 1)2 . So the splitting field is Z3[x]/⟨x2 + 1⟩
39. Since (Z2[x]/ f (x) )∗ = 31 is prime and the order of every element must divide it, every nonidentity is a generator.
40. Observe that x4 6x2 7 = (x2 7)(x2 + 1).
41. Use the Fundamental Theorem of Field Theory (Theorem 19.1) and the Factor Theorem (Corollary 2 of Theorem 16.2).
42. By the proof of the Fundamental Theorem of Field Theory p(x) has a zero in some extension E of F and [E : F ] n Now write p(x) = (x a)g(x) where g(x) E[x]. By induction on the degree of the polynomial, g(x) splits in some extension K of E and [K : E] ≤ (n 1)!. Thus p(x) splits in K and [K : F ] = [K : E][E : F ] ≤ (n 1)!n = n!.
43. Proceeding as in Example 9 we suppose that h(t)/k(t) is a zero in Zp(t) of f (x) where deg h(t) = m and deg k(t) = n. Then (h(t)/k(t))p = t, and therefore (h(t))p = t(k(t))p . Then by Exercise 49 of Chapter 13 we have h(tp) = tk(tp). Since deg h(tp) = pm and deg tk(tp) = 1 + pn, we have pm = 1 + pn But this implies that p divides 1, which is false. So, our assumption that f (x) has a zero in Zp(x) has led to a contradiction. That f (x) has a multiple zero in K follows as in Example 9.
44. By the corollary to Theorem 19.9, deg f (x) has the form nt where t is the number of distinct zeros of f (x).
45. Since 1 = 1, xn x would have 1 as a multiple zero. But then, by Theorem 19.5, xn x and its derivative, which is 1 = 1, must have a common factor of positive degree. This is impossible.
46. Since x2 + x + 1 is the only irreducible over F , it is the only possible quadratic factor in the product. If x2 + x + 1 appeared more than once in the product, then in the irreducible factorization of f (x) in the splitting field of f (x) over F the linear factors of f (x) would have multiplicity at least 2, in violation of the corollary of Theorem 19.9.
47. From E x a m p√ l e 8 we know that the splitting field of x 3 √ 2 over Q is Q ( √ 3 2, ω) where ω = 1/2 + √ 3i/2. So, the splitting field over F = √ Q ( 3 2) is F√(ω) where 3 ω = 1/2 + 3i/2. The splitting field over F = Q( 3i) is F ( 2).
48. 2a2 + 1.
49. Observe that the polynomial x2 2x 1 is irreducible over Z5 . So Theorems 19.1 and 19.3 shows that such a field exists.
50. No, because α2 = α + 2 implies that 0 = α2 α 2 = (α + 1)(α 2) and neither α + 1 nor α 2 is 0.
51. If α F (β), then we have α = aβ + b for some a, b F . Squaring both sides, replacing β2 with β 1, and solving for β, we find that β F For the second part, if β F (α) we have β = aα + b for some a, b F . Solving for α in terms of β and proceeding as before, we get that β is in F .
53. Because the derivative of xp n x is 1, xp n x has no multiple zeroes in any extension of Zp
54. It follows from Theorem 19.9 that f (x) is reducible over F By Theorem 19.9, one of a1 or a2, is in F . If a2 is in F , then so is a1. For if x a2 is in F [x], when we long divide f (x) by (x a2)2, the quotient x a1 is also in F [x].
55. Let f (x) = cnxn + cn 1xn 1 + + c1x + c0. Then, using Exercise 49 in Chapter 13 and that the fact that 0 = 02 and 1 = 12, we have f (a2 ) =
(
)2 = 0. By repeating the first part of this problem, we know that the five zeros are a, a2 , a4 , a8 , a16 . If any two of these were equal, then we would have ak = 1 for some positive integer k less than 31, whereas |a| = 31 because |(Z2 [
]⟨f (x)⟩)∗| = 31.
56. It is a zero of f (x) of multiplicity at least 3.
CHAPTER 20
Algebraic Extensions
1. It follows from Theorem 20.1 that if p(x) and q(x) are both monic irreducible polynomials in F [x] with p(a) = q(a) = 0, then deg p(x) = deg q(x). If p(x) /= q(x), then (p q)(a) = p(a) q(a) = 0 and deg (p(x) q(x)) < deg p(x), contradicting Theorem 20.1.
To prove Theorem 20.3 we use the Division Algorithm (Theorem 16.2) to write f (x) = p(x)q(x) + r(x), where r(x) = 0 or deg r(x) < deg p(x). Since 0 = f (a) = p(a)q(a) + r(a) = r(a) and p(x) is a polynomial of minimum degree for which a is a zero, we may conclude that r(x) = 0.
2. If f (x) ∈ F [x] does not split in E, then it has a nonlinear factor q(x) which is irreducible over E But then E[x]/⟨q(x)⟩ is a proper algebraic extension of E
3. Let F = Q( √ 2, √ 3 2, √ 4 2, . . .). Since [F : Q] ≥ [Q( √ n 2) : Q] = n for all n, [F : Q] is infinite. To prove√t h a√ t F √ i s an a l g√ e b r a ic extension of Q, let a ∈ F . There is some k k such that a ∈ Q( 2, 3 2, 4 2, . . . , 2). It follows from Theorem 20.5 that [Q( √ 2, √ 3 2, √ 4 2, , √ k 2) : Q] is finite and from Theorem 20.4 that Q( √ 2, √ 3 2, √ 4 2, , √ k 2) is algebraic.
4. Suppose E(a) is an algebraic extension of E. Then a is algebraic over F (Theorem 20.7). So if f (x) is the minimal polynomial for a over F then f (x) can be written in the for (x a1)(x a2) (x an) where each ai ∈ E. But then f (a) = 0 implies a = ai for some i and a ∈ E.
5. Since every irreducible polynomial in F [x] is linear, every irreducible polynomial in F [x] splits in F . So, by Exercise 4, F is algebraically closed.
6. Suppose g(x) = h(x)k(x) where h(x) is irreducible over F (a) and deg h(x) < deg g(x). Let b be a zero of h(x) in some extension of F (a). Then [F (a, b) : F (a)] < deg g(x) = [F (b) : F ]. Since [F (a, b) : F ] = [F (a, b) : F (a)][F (a) : F ] = [F (a, b) : F (b)][F (b) : F ] we know that [F (a, b) : F ] is divisible by [F (a) : F ][F (b) : F ]. But [F (a, b) : F ] = [F (a, b) : F (a)][F (b) : F ] < [F (b) : F ][F (a) : F ]. This is a contradiction.
7. √ S up p o s e Q( √ a) = Q( √ b). If √ b ∈ Q, t h e√n √ a ∈ Q and we may take c = √ a/ √ b. If b ∈ / Q, then √ a ∈ / Q. Write √ a = r + s b where r and s belong to Q. Then r = 0 for, if not, then a = r2 + 2rs √ b + b a√nd therefore (a r2 b)/2r = s √ b. But (a r2 b)/2r is rational whereas s b is irrational. Conversely, if there is an element c ∈ Q such that a = bc2 (we m√a y assume th√at c is po√sitive) then, by Exercise 7 in Chapter 19, Q( √ a) = Q( bc2) = Q(c b) = Q( b).
8. Since ( √ 3 + √ 5)2 ∈ Q( √ 15), [Q( √ 3 + √ 5) : Q( √ 15)] = 2. A basis is {1, √ 3 + √ 5}. For the s e c o n√ d q u√e s t i √ o n , note t h√ a t because 21/12 = 24/12 · 2 3/12 = 21/3 · 2 1/4 , we have that Q( 2, 3 2, 4 2) = Q( 12 2). Since 1 √ 2 2 is a√z ero of x12 2 and x12 2 is irreducible over Q by Eisenstein, we know that Q( 12 2) has degree 12 over Q and a basis is 1, 21/12 , 22/12 , . . . , 211/12 .
Alternate solution. First observe that Q( √ 2, √ 3 2, √ 4 2) = Q ( √ 3 2, √ 4 2). Bec ause [Q( 3 2, 4 2) : Q] = [Q( 3 2, 4 2) : Q( 4 2)][Q( 4 2) : Q] and a basis for Q( 3 2, 4 2) over Q ( √ 4 2) is X = {1, 21/3 , 22/3} and a basis for Q ( √ 4 2) over Q i√ s √ Y = {1, 21/4 , 22/4 , 23/4}, we know that XY is a basis for Q( 3 2, 4 2) over Q.
9. Since [F (a) : F ] = 5, 1, a, a2 , a3 , a4 is a basis for F (a) over F . Also, from 5 = [F (a) : F ] = [F (a) : F (a3)][F (a3) : F ] we know that [F (a3) : F ] = 1 or 5.
However, [F (a3) : F ] = 1 implies that a3 ∈ F and therefore the elements
1, a, a2 , a3 , a4 are not linearly independent over F because a3 can be written as 1 · a3 and as a3 1. So, [F (a3) : F ] = 5.
10. Since [E : F ] = [E : F (a)][F (a) : F ] we have [F (a) : F ] = [E : F ], in which case F (a) = E, or [F (a) : F ] = 1, in which case F (a) = F .
11. If a is a zero of f (x) in E then n = [E : F ] = [E : F (a)][F (a) : F ] = [E : F ](deg f (x)).
12. Observe that [F (a) : F (a2)] = 1 or 2 and that n = [E : F ] = [E : F (a)][F (a) : F (a2)][F (a2) : F ].
13. g(x) = f (x/b c/b)
14. Since β is a zero of xn βn , F (β) is an algebraic extension of F (βn), and 1, β, β2 , . . . , βn 1 is a basis for F (β) over F (βn), we have [F (β) : F (βn)] = n. For n = 4, we have 3β5 + 2β 3 = (3β2 )β3 + 2β 4β.
15. By the Primitive Element Theorem there is an element b in F (a1, a2) such that F (a1, a2) = F (b). Then, by induction on n, there is an element c in F (b, a3, . . . , an) such that F (c) = F (b, a3, , an) = F (a1, a2, , an).
20. If b = 0, then x c is minimal. If b = 0 for g(x) = f ((x c)/b) we have g(ab + c) = f ((ab + c c)/b) = f (a) = 0. That g(x) has minimum degree follows from the fact that F (a) = F (ab + c) (see Exercise 7 Chapter 19). Or that g(bx + c) = f (x).
21. Elements of Q(π) have the form (am πm + am 1 πm 1 + + √ a 0 ) / ( b n π n + bn 1 φn 1 + + b0 ), where the a’s and b’s are rational numbers. So, if 2 ∈ Q(π), we have an expression of the form 2(bn πn + bn 1 φn 1 + + b0)2 = (amπm + am 1πm 1 + + a0 )2 Equating the lead terms of both sides, we have 2b2 π2n = a2 π2m . But then we have m = n, and √ n m 2 is equal to the rational number am /bn This argument works for any real number of the form k a not in Q
22. A splitting field of a polynomial in Q[x] is an algebraic extension of Q whereas C is a transcendental extension of Q
23. Let {x1, x2, , xn } be a basis for K over F Then K = F (x1, x2, , xn ) for some x1 , x2 , . . . , xn in K.
24. If an irreducible polynomial p(x) in R[x] has degree n and a is a zero in C of p(x) then 2 =[C:R]= [C: R(a)][R(a):R] = [C: R(a)]n So, n = 1 or 2.
25. By Theorem 20.5, we have that [F3 : F1] = [F3 : F2][F2 : F1] and [F4 : F1] = [F4 : F3][F3 : F1] = [F4 : F3][F3 : F2][F2 : F1]. The general case that [Fn : F1 ] = [Fn : Fn 1 ][Fn 1 ] : Fn 2 ] · · · [F2 : F1 ] follows by induction.
26. Suppose that [E : F ] = 1. Because {1} is a linearly independent set over F , it is a basis for E over F . So every element of E has the form a · 1 = a for some a in F . Now suppose that E = F . Then {1} is a basis for E over F .
27. Since [L : F ] = [L : K][K : F ] we have [K : F ] = 1. It follows that F = K.
28. Pick a in K but not in F . Now use Theorem 20.5.
29. If a is a zero of p(x) Q[x], then for k 2, Alternate solution. Observe that √ k a is a zero of p(xk) ∈ Q[x]. [Q( √ k a : Q] = [Q( √ k a : Q(a)][Q(a) : Q] k[Q(a) : Q] is finite. For the case of am/n where m and n are positive integers, note that a is algebraic over Q implies that am is algebraic over Q So ( n (am) = am/n is algebraic over Q
30. Note that if c ∈ Q(β) and c /∈ Q, then 5 = [ Q√( β ) : Q] = [Q(β)√: Q(c)][Q(c) : Q] so that [ Q(c) : Q] = 5. On the other hand, [Q( 2) : Q] = 2, [Q( 3 2) : Q] = 3, and [Q( 4 2) : Q] = 4.
31. Since √ 2 = ( √ 6 2)3 and √ 3 2 = ( √ 6 2)2 w√e know t√ hat Q√( √ 2 , √ 3 2) ⊆ Q ( √ 6 2). Because 21/6 = 21/2(21/3) 1 , we a l √ s o have√Q( 6 2) ⊆√Q ( 2, 3 2). A l t√ e r n a√ t e solutio√ n. Since 2 = ( 2) and 2 = ( √ 2) we k n o√w that 6 3 3 6 2 Q( √2, 3 2) ⊆ Q( 6 2). Also, [Q( √ 2, √ 3 2) : Q] is divisible by [Q( 2) : Q] = 2 and [Q( 3 2) : Q] = 3 It follows that [Q( √ 2, √ 3 2) : Q] = 6 = [ Q ( √ 6 2) : Q]. √ √
√ 1 1 √ √ a √ b √ a √ b
32. By closure, Q( a + b) ⊆ Q( a, b). Since ( a + b) = √ a + b √ a √ b = a b √ √ √ √ √ √ is trivial.) It follows that √ a = 1 (( √ a + √ b) + ( √ a √ b)) and √ 1 √ √ √ √ 2 √ √ √
33. 9
b = 2 (( a + b) ( a b)) are in Q ( a, b). So, Q ( a, b) ⊆ Q ( a + b).
34. Suppose E1 ∩ E2 /= F . Then [E1 : E1 ∩ E2 ][E1 ∩ E2 : F ] = [E1 : F ] implies [E1 : E1 ∩ E2 ] = 1, so that E1 = E1 ∩ E2 . Similarly, E2 = E1 ∩ E2 .
35. First observe that √ 4 3 is a z e r√ o of x2 √ 3, which is in Q ( √ 3). If this polynomial is √ √
and a b ∈ Q( a + b) we have a b ∈ Q( a + b). (The case that a b = 0 not a min imal polynomial of 4 3 over Q( √ 3), then [ Q ( √ 4 3) : Q( √ 3)] = 1. But then,
4 = [Q( 4 3) : Q] = [Q( 4 3) : Q( 3)][Q( 3) : Q] = 1 2, which is a contradiction.
36. Observe that F (1 + a 1 ) = F (a 1 ) = F (a).
37. If ab is a zero of cnxn + + c1x + c0 F [x], then a is a zero of cnbnxn + + c1bx + c0 F (b)[x].
Alternate proof. Observe that because both ab and b are algebraic over F (b), so is a = (ab)b 1 (Corollary of Theorem 20.7.)
38. It suffices to show that for every non-zero element a in R a 1 is also in R. Since a is in E, it is the zero of some minimal polynomial in F [x] of degree d By closure under multiplication, we know that the basis {1, a, a2 , . . . , ad 1} of F (a) is contained in R. So, F (a) ⊆ R and a 1 ∈ F (a).
39. Note that x 3, x3 (π3)2 and x π3 are in Q(π3)[x] and have 3, π2, and π as zeros, respectively. So, 3, π2, and π are algebraic over Q(π3). Then, we have by the corollary to Theorem 20.7 that 3π2 + π is algebraic over Q(π3).
40. Every element of F (a) can be written in the form f (a)/g(a), where f (x), g(x) F [x]. If f (a)/g(a) is algebraic and not a member of F , then there is some h(x) F [x] such that h(f (a)/g(a)) = 0. By clearing fractions and collecting like powers of a, we obtain a polynomial in a with coefficients from F equal to 0. But then a would be algebraic over F
41. Note that [F (a, b) : F ] is divisible by both m = [F (a) : F ] and n = [F (b) : F ] and [F (a, b) : F ] = [F (a, b) : F (b)][F (b) : F ] [F (a) : F ][F (b) : F ] mn. So, [F (a, b) : F ] = mn.
42. Since a is a zero of x3 a3 over F (a3), we h √ a v e [F (a) : F (a3)] ≤√3. For the second
43. Take F = Q, a = √ 4 2, b = √ 6 2. Then [F (a, b) : F ] = 12 and [F (a) : F ][F (b) : F ] = 24.
44. Since E must be an algebraic extension of R, we have E C and so [C : E][E : R] = [C : R] = 2. If [C : E] = 2, then [E : R] = 1 and therefore E = R. If [C : E] = 1, then E = C.
45. Use the quadratic formula and Exercise 7 of Chapter 19.
46. Let a be a zero of p(x) in some extension of F . First note [E(a) : E] [F (a) : F ] = deg p(x). Then observe that [E(a) : F (a)][F (a) : F ] = [E(a) : F ] = [E(a) : E][E : F ]. This implies that deg p(x) divides [E(a) : E], so that deg p(x) = [E(a) : E]. It now follows from Theorem 19.3 that p(x) is irreducible over E.
47. If E is a finite extension of F , pick a1 ∈ E but not in F . Then [E : F ] ≥ [F (a1) : F ]. If F (a1) = E, pick a2 E not in F (a1). Then [E : F ] [F (a1, a2) : F ]. Since [E : F ] is finite, we can continue this process only a finite number of times. The converse follows from Theorem 20.5.
part, take F = Q, a = 1; F = Q, a = ( 1 + i 3)/2; F = Q, a = 3 2. (α + β)2 4
48. √ Suppose that α + β a√n d αβ are algebraic√o ver Q and that α ≥ β. Then Also, α = ((α + β) (α β))/2 is algebraic over Q, which is a contradiction.
is also algebraic over Q.
49. If f (a) is a zero of g(x) ∈ F [x], then a is a zero of (g ◦ f )(x) ∈ F [x].
50. It follows from the Quadratic Formula that √ b2 4ac is a primitive element.
51. Let deg f (x) = m, deg g(x) = n and suppose that f (x) is irreducible over F (b). Then m = [F (a, b) : F (b)] and [F (a, b) : F ] = [F (a, b) : F (b)][F (b) : F ] = mn = [F (a, b) : F (a)][F (a) : F ] = [F (a, b) : F (a)]m Thus, n = deg g(x) = [F (a, b) : F (a)] and therefore g(x) is irreducible over F (a). The other half follows by symmetry.
52. Because a ∈√Q( √ a) it suffices to show that √ a ∈ Q(a). Since a3 = 1 we have a4 = a. Then a2 = a ∈ Q(a).
53. By the Factor Theorem (Corollary 2 of Theorem 16.2), we have f (x) = (x a)(bx + c), where a, b, c ∈ F (a).
54. Say a is a generator of F ∗ . F cannot have characteristic 0 because the subgroup of rationals is not cyclic. Thus F = Zp(a), and by Theorem 20.3 it suffices to show that a is algebraic over Zp If a Zp, we are done. Otherwise, 1 + a = ak for some k = 0. If k > 0, we are done. If k < 0, then a k + a1 k = 1 and we are done.
55. Suppose that there are such rationals. Let s = s1/s2 and t = t1/t2 where s1, s2, t1, t2 are positive integers. Then (rs)s2 t2 = (πt)s2 t2 But the left side is rational and the right side is not.
56. If [K : F ] = n, then there are elements v1, v2, , vn in K that constitute a basis for K over F . The mapping a1v1 + + an vn (a1, . . . , an ) is a vector space isomorphism from K to F n . If K is isomorphic to F n , then the n elements in K corresponding to (1, 0, . . . , 0), (0, 1, . . . , 0), . . . , (0, 0, . . . , 1) in F n constitute a basis for K over F .
57. A counterexample is [Q(41/4) : Q] = 2; another is [Q(41/6 : Q] = 3.
58. Observe that [F (a, b) : F (a)] ≤ [F (a, b) : F (a)][F (a) : F ] = [F (a, b) : F ].
59. First note that c0 /= 0, for otherwise cd 1xd 1 + cd 2xd 2 + + c1x + c0 has x as a factor over field F . So from cd 1ad 1 + cd 2ad 2 + · · · + c1a + c0 = 0 we have c0 = a(cd 1ad 2 + cd 2ad 3 + + c1). Multiplying both sides by a 1( c0) 1 we obtain a 1 = c0 1cd 1ad 2 c0 1cd 2ad 3 c0 1c1.
60. Observe that K = F (a1, a2, . . . , an), where a1, a2, . . . , an are the zeros of the polynomial. Now use Theorem 20.5.
3. The lattice of subfields of GF(64) looks like Figure 20.3 with GF(2) at the bottom, GF(64) at the top, and GF(4) and GF(8) on the sides.
4. First observe that 0 = a3 + a2 + 1 = a2(a + 1) + 1 so (a + 1) 1 = a2 . Then solving the equation for x we have that x = a2 + a.
5. GF(26)
6. For any a in GF(32)∗ |a| must divide 31.
7. By Theorem 21.3, the other solution is a3 = 2a + 1.
8. Since a is a zero and the prime subfield has characteristic 2, we have from Theorem 21.3 that the other zeros are a2 and a4 .
9. Since each binomial coefficient pi other than j = 1 and j = pi is divisible by p, we have (a + b)p i = ap i + bp i . Clearly (ab)p i = ap i bp i . Since GF(pp ) is a field, ap i = 0 only when a = 0, so the Ker φ = {0}.
j
i
10. It follows from the Finite Subgroup Test (Theorem 3.3)) that S is a group under addition and the non-zero elements form a group under multiplication.
11. Observe that x2 + 1 = (x + 1)2
Alternate solution. Any solution a has the property that a2 = 1 = 1. But because |GF(2n)∗| = 2n 1, there is no element in GF(2n)∗ of order 2.
12. Note the |a5| = 3.
13. Let a be a zero of a cubic irreducible f (x) in Z2[x] in some extension of Z2 Then Z2(a) has order 8, and by Theorem 21.3, the other two zeros of f (x) are a2 and a4 , which also belong to Zp(a). So, f (x) splits over Zp(a).
14. Use Theorem 21.1.
15. Because GF(64)∗ is a cyclic group of order 63, we have by Theorem 4.4 that there are φ(63) = φ(9)φ(7) = 36 elements with the property.
16. According to Theorem 21.3, the zeros are: a, a2 , a4 , a8 for the first part and a7 , (a7)2 , (a7)4 , (a7)8 = a7 , a14 , a13 , a11 for the second part.
17. Let F = pn Then n must be divisible by both 4 and 6. So, by Theorem 21.4, F must also have subfields of order p12 , p3 , p2 and p.
18. By Theorem 22.3, m divides 3.
19. By observation, the linear factors are x and x 1. Since x8 x has no multiple zeros in GF(8), it has no multiple irreducible factors over GF(2). Because the degrees of the irreducible factors must divide 3 and sum to 8 it follows that the nonlinear irreducible factors must be two cubics. Noting that the only cubic irreducibles over Z2 are x3 + x + 1 and x3 + x2 + 1, these are the remaining factors of x8 x.
20. Note that by Theorem 21.4, the only proper subfield of GF(32) is GF(2). So, the degree of a over GF(2) is 5.
21. Observe that a16 = a4 implies that 0 = a16 a4 = (a4 )4 a4 = (a4 a)4 .
22. Observe that Zp(b) is a subfield field of Zp(a) and |Zp(b)| = pn .
23. First note that x, x 1 and x 2 are factors and they are the only linear factors. If p(x) is an irreducible factor of x27 x and a is a zero of an irreducible factor x27 x of degree d, then Z3 (a) is a subfield of GF(27) of order 3d , and by Theorem 21.4 we have that d = 1 or 3. So, the irreducible factorization of x27 x consists of three linear factors and eight cubic factors.
24. Observe that the mapping from F to itself that takes x to x2 is a field automorphism with kernel {0}. Since a 1-1 mapping from a finite set to itself is onto we are done.
25. By Theorem 21.4, GF(pn) is properly contained in GF(pm ) when n is a proper divisor of m So the smallest such field is GF(p2n).
26. Observe that every element of the GF(pn) is a zero of xp n x So evaluating xp n x + 1 for any element of GF(pn) gives 1.
27. To show that F is a field, let a, b ∈ F Then a ∈ Fi for some i and b ∈ Fj for some j and a, b ∈ Fk where k is the maximum of i and j. It follows that a b ∈ Fk, ab ∈ Fk and a 1 ∈ Fk when a /= 0.
28. GF(210 ), GF(215 ), GF(225).
29. Since 0, 1, and 2 are zeros, we know x, x 1 and x 2 are factors and by Theorem 19.8 these each have multiplicity 1. Theorem 21.5 tells us that all other irreducible factors have degree 2 and, since their degrees must sum to six, there are three of them. Finally, from the proof of Theorem 21.1 we know that each of the nine elements of GF(9) are zeros of x9 x. So no two quadratic reducibles can be the same. The factorization x9 x = x(x 1)(x 2)(x2 + 1)(x2 + x + 2)(x2 + 2x + 2).
30. It follows from Theorem 21.3 that the desired field is GF(212).
31. Since (Z3[x]/ x3 + 2x + 1 )∗ = 26, we need only show that x = 1, 2 or 13. Obviously, x = 1 and x2 = 1. Using the fact that x3 + 2x + 1 = 0 and doing the calculations we obtain x13 = 2.
32. The argument in the proof of Theorem 21.3 shows that 4 divides 5n 1. Since GF(5n)∗ is a cyclic group of order 5n 1 and 2 = 4, we know that 2 is the unique subgroup of order 4. We also know that a(5n 1)/ 4 is an element of order 4. By Corollary 3 of Theorem 4.3, if b is an element of a finite cyclic group of order 4 then the only other element in the group of order 4 is b3 . So, k = (5n 1)/4.
33. From x3 = x + 1 we get x9 = (x + 1)3 = x3 + 1 = x + 2 and x4 = x2 + x. So, x13 = (x + 2(x2 + 2) = 1. Then |2x| = 26 and is a generator.
34. Since 1, x and x2 form a basis for Z3[x]/⟨f (x)⟩ over Z3 we have that |x| /= 2; for otherwise x2 1 = (x 1)(x + 1) = 0 and therefore x = ±1. By Lagrange’s Theorem |x| is 13 or 26. If |x| = 13, then |2x| = 26.
35. Note that if K is any subfield of GF(pn), then K∗ is a subgroup of the cyclic group GF(pn)∗ So, by Theorem 4.3, K∗ is the unique subgroup of GF(pn)∗ of its order.
36. Observe that |ab| = |⟨a⟩⟨b⟩| = |⟨a⟩||⟨b⟩|/|⟨a⟩ ∩ ⟨b
| = |⟨a⟩||⟨b⟩| = 5 · 16 = 80.
37. Let a, b K. Then, by Exercise 49b in Chapter 13, (a b)p m = ap m bp m = a b. Also, (ab)p m = ap m bp m = ab So, K is a subfield.
38. Mimic the proof of the case where the field is GF(pn) and the group is GF(pn)∗ given in Theorem 21.2.
39. By Corollary 4 of Lagrange’s Theorem (Theorem 7.1), for every element a in F ∗ we have ap n 1 = 1. So, every element in F ∗ is a zero of xp n x
40. Theorem 21.3 reduces the problem to constructing the subgroup lattices for Z18 and Z30.
41. They are identical to the lattice of Z30.
42. Without loss of generality, we may assume that p(x) is monic. In some splitting field K of p(x) we can write p(x) = (x a1)(x a2) (x am) where the ai are distinct and K = GF(p)(a1, a2, . . . , am). Then K = GF (pn) for some n and since every element of K is a zero of xp n x we have that p(x) = (x a )(x a ) (x a ) divides xp n x. 1 2 m
43. The hypothesis implies that g(x) = x2 a is irreducible over GF(p). Then a is a square in GF(pn) if and only if g(x) has a zero in GF(pn). Since g(x) splits in GF(p)[x]/ g(x) GF(p2), g(x) has a zero in GF(pn) if and only if GF(p2) is a subfield of GF(pn). The statement now follows from Theorem 21.3.
44. Let a be a zero of f (x) in E = Zp[x]/⟨f (x)⟩. Then |Zp(a)| = p3 and, by the corollary of Theorem 21.3, Zp(a) is the splitting field of f (x).
46. Note that GF(p2n) is an algebraic extension of GF(pn). Alternate proof. Let F = {a1, a2, , an} be a finite field. Then f (x) = (x a1)(x a2) · · · (x an) + 1 ∈ F [x] but has no zeros in F .
47. Since both a62 and 1 have order 2 in the cyclic group F ∗ and a cyclic group of even order has a unique element of order 2 (see Theorem 4.4), we have a62 = 1.
48. pk where k = gcd(s, t).
49. If K is a finite extension of a finite field F , then K itself is a finite field. So, K∗ = ⟨a⟩ for some a ∈ K and therefore K = F (a).
50. Suppose b is one solution of xn = a. Since F ∗ is a cyclic group of order q 1, it has a cyclic subgroup of order n, say ⟨c⟩. Then each member of ⟨c⟩ is a solution to the equation xn = 1. It follows that b⟨c⟩ is the solution set of xn = a
51. Observe that p 1 = 1 has multiplicative order 2 and a(p 1)/2 is the unique element in ⟨a⟩ of order 2.
52. Since d divides n, GF(pn) contains the subfield GF(pd), and by Corollary 2 of Theorem 21.2, GF(pd) has the form GF(p)(a) for some element a in GF(pd) of degree d Moreover, by Theorem 21.1, GF(p)(a) has the form GF(p)[x]/ p(x) where p(x) is an irreducible polynomial over GF(p), the degree of p(x) is d, and p(a) = 0. Since p(x) is irreducible over GF(p) the only common divisor of p(x) and xp n x is 1 or p(x). If the common divisor is 1 then by Exercise 43 of Chapter 16 there are elements h(x) and k(x) of GF(p)[x] such that p(x)h(x) + (xp n x)k(x) = 1. But since x a divides the left side we have a contradiction.
53. Observe that the mapping from the cyclic group GF(pn)∗ to itself that takes x to x2 is a group homomorphism with the kernel {±1}. So, the mapping is 2-1.
54. Since 5 mod 4 = 1, we have that 5n 1 is divisible by 4 for all n. Now observe that 2 has multiplicative order 4 and a(5n 1)/4 has order 4. (The only other element of order 4 is a3(5n 1)/ 4 .)
55. Since p = 1 (mod 4) we have pn = 1 mod 4 and GF(pn)∗ is a cyclic of order pn 1. So, by Theorem 4.4 there are exactly two elements of order 4.
56. When n is odd, pn = 3 mod 4 and therefore pn 1 is not divisible by 4. By Theorem 4.3 a has no element of order 4. When n = 2m, pn = (p2)m = (32)m = 1 mod 4 and therefore pn 1 is divisible by 4. Thus, by Theorem 4.4, GF(pn)∗ has exactly two elements of order 4.
57. First note that a is not in Z5 , for then x a would be a factor of the irreducible. Then taking b = 0 and we solve for c = (4a + 1)(3a + 2) 1 .
58. Since (Z2[x]/ x4 + x2 + 1 )∗ = 15 we need only check to see that a3 and a5 are not 1. a3 = 1 because a has degree 4. a4 = a2 + 1 implies that a5 = a3 + a = 1 because 1, a, a2, and a3 are basis elements.
59. It is a field of order 45 .
60. Because 2 = [GF(64):GF(8)] = [GF(64):GF(8)(a)][GF(8)(a)][GF(8)] we have deg f (x) = 2. Let GF(64)∗ = ⟨a⟩ Then GF(8)∗ = ⟨a9⟩ and f (x) = x2 a9 ∈ GF(8) and is irreducible over GF(8) (its zeros are ± a3).
61. For the case GF(pn ), observe that if 1 + a + a2 + a3 + + ai = 1 + a + a2 + a3 + + aj for some i < j, then 0 = ai+1 + + aj = ai+1 (1 + a + a2
aj i 1) and therefore ai+1 is a zero-divisor.
62. By Theorem 21.4 the zeros are a, a2 , a(22 ) , a(23 ) , a(24 )
63. Let F1 = GF(pn), F2 = GF(p2n), F3 = GF(p4n), F4 = GF(p8n),
64. Since every 5th degree monic irreducible polynomial over Z3 splits in GF(35 ), they all divide x3 x. No such factor can appear more than once in the factorization because x3 x has no multiple zeros. By Theorem 21.5, the monic irreducible factors of x3 x have degrees 1 or 5 and exactly three have degree 1. So, the remaining irreducible factors have degree 5 and there must be exactly 48 of them in order for the degree of the product of all the irreducible monic factors to be 243.
65. The algebraic closure of Z2.
66. The only finite subfield of Z2(x) = {f (x)/g(x) |f (x), g(x) ∈ Z2[x], g(x) /= 0} (the field of quotients of Z2[x]) is Z2
67. By Theorem 21.4, for each prime q the only proper subfield of GF(pq ) is GF(p).
68. Since [F (a, b) : F ] = [F (a, b) : F (a)][F (a) : F ] is finite we know that F (a, b) is finite and therefore F (a, b)∗ is a cyclic group. Then, for any generator c of F (a, b)∗ , we have F (a, b) = F (c).
69. To see that GF(p∞ ) is a field, let a and b be elements in GF(p∞). Then a belongs to GF(pm!) and b belongs to GF(pn!) for some positive integers m and n. Thus, a and b belong to the larger of these two fields, say GF(pm!), which means that a b and ab 1(b = 0) also belong to GF(pm!), which is a subfield of GF(p∞). Lastly, if f (x) is any polynomial in GF(p∞ )[x], then f (x) belongs GF(ps![x] for some positive integer s Thus, the splitting field of f (x) over GF(ps !) is a finite field of characteristic p and so is contained in GF(pt) for some positive integer t. This means this splitting field is a subfield of GF(p∞ ).
70. It follows from the Primitive Element Theorem and Theorem 21.2 that F is an infinite field of prime characteristic.
71. Since x16 x has 16 distinct zeros in its splitting field E, no irreducible factor p(x) of it can occur in the factorization more than once. Moreover, we know from Theorem 16.5 that the possible degrees of p(x) are 1, 2, and 4. Because the factorization as irreducibles includes x and x 1, the sum of the degrees of the nonlinear factors is 14. Since 14 is not a multiple of 4, x16 x must have has a quadratic factor. Thus it is x2 + x + 1 because it is the only irreducible quadratic over Z2. So, that leaves three irreducible fourth- degree factors because the sum of the degrees of the irreducible factors is 16.
72. Suppose there is such an element a in some field extension of (GF)(p). Then both 1 and a are zeros of xk 1 F [x]. So, we can write xk 1 = (x 1)g(x). This gives us that g(a) = 0, which means deg a < d over F .
73. They are the same.
74. If there is some other element a in GF(3n) that is a zero, then a3 = 1 and therefore |a| = 6. But 6 does not divide 3n 1, which is the order of the cyclic group GF(3n)∗ .
75. GF(53k) where k is any positive integer.
76. By Theorem 21.3, we have that p(x) = x(x a)(x ap)(x ap 2 ) · · · (x ap d ).
77. By the proof of Theorem 21.1, we have that xn x = x(x a)(x a2)(x a3) (x an 1).
CHAPTER 22
Geometric Constructions
1. To construct a + b, first construct a Then use a straightedge and compass to extent a to the right by marking off the length of b. To construct a b, use the compass to mark off a length of b from the right end point of a line of length a The remaining segment has length a b.
2. Let x denote the length of the long side of the triangle. Then a = x . 1 b
3. Let y denote the length of the hypotenuse of the right triangle with base 1 and x denote the length of the hypotenuse of the right triangle with the base |c|. Then y2 = 1 + d2 , x2 + y2 = (1 + |c|)2 and |c|2 + d2 = x2 So, 1 + 2|c| + |c|2 = 1 + d2 + |c|2 + d2 , which simplifies to |c| = d2 .
4. Let x denote the length of the side of along the base of the small triangle in figure in the text. Then x is constructible and x = a . 1 b
5. Suppose that sin θ is constructible. Then, by Exercises 1, 2, and 3, 1 sin2 θ = cos θ is constructible. Similarly, if cos θ is constructible then so is sin θ.
6. Look at Figure 22.1.
7. From the identity cos 2θ = 2 cos2 θ 1 we see that cos 2θ is constructible if and only if cos θ is constructible.
8. By Exercises 5 and √ 7, it suffices to show that a 30◦ is constructible. This true because cos(30◦) = 3/2.
9. Use Exercise 6.
10. The discussion in this chapter shows that cos 20◦ is not constructible. Now use Exercises 5, 6, and 7.
11. Note that solving two linear equations with coefficients in F involves only operations under which F is closed.
12. Say the line is ax + by + c = 0 and the circle is x2 + y2 + dx + ey + f = 0. We seek the simultaneous solution of these two equations. If a = 0, then y = c/b and the equation of the circle reduces to a quadratic in x with coefficients from F Since the solution of a quadratic involves only the operations of F and a square root of an element from F , the value for x lies in F or in F ( α) where α F and α > 0. If a = 0, the x terms in the circle can be replaced by ( b/a)y c/a to obtain a quadratic equation. Then, as in the previous case, y and therefore x lie in F or in F ( α). For the second portion, consider x2 + y2 = 1 and x y = 0.
13. This follows from Theorem 17.1 and the mod 5 irreducibility test. (Theorem 17.3.)
14. If the polygon is constructible, so is cos(2π/7). Thus, it suffices to show that 8x3 + 4x2 4x 1 is irreducible over Q This follows from mod 5 irreducibility test.
15. If a regular 9-gon is constructible, then so is the angle 360◦/9 = 40◦ . But Exercise 10 shows that a 40◦ angle is not constructible.
16. Use Exercises 5, 6, and 7.
17. This amounts to showing √ π is not constructible. But if √ π is constructible, so is π. However, [Q(π) : Q] is infinite.
18. It suffices to show that 2π/5 is constructible. Clearly, [Q(cos 2π/5) : Q] = 2 so that cos 2π/5 is constructible. Now use Exercise 6.
19. “Tripling” the cube is equivalent to constructing an edge of length √ 3 3. But [Q( 3 3) : Q] = 3, so this can’t be done. No, for 4V since [Q( 3 4) : Q] = 3. Yes, for 8V since [Q( 3 8) : Q] = 1, which is a power of 2.
20. “Cubing” the circle is equivalent to constructing the length √ 3 π. But [ Q ( √ 3 π) : Q] is infinite.
21. Use Exercises 1-4.
CHAPTER 23
Sylow Theorems
1. a = eae 1; cac 1 = b implies a = c 1bc = c 1b(c 1) 1; a = xbx 1 and b = ycy 1 imply a = xycy 1x 1 = xyc(xy) 1 .
2. Observe that bab 1 = φb(a) where φb is the inner automorphism induced by b.
3. Note that |a2| = |a|/2 and appeal to Exercise 2.
4. {e}, {a2}, {a, a3}, {b, ba2}, {ba, ba3}
5. Observe that T (xC(a)) = xax 1 = yay 1 = T (yC(a)) y 1xa = ay 1x y 1x C(a) yC(a) = xC(a). This proves that T is well defined and one-to-one. Onto is by definition.
6. cl(a) = {a} if and only if for all x in G, xax 1 = a. This is equivalent to a ∈ Z(G).
7. Say cl(e) and cl(a) are the only two conjugacy classes of a group G of order n Then cl(a) has n 1 elements and, by Theorem 24.1 C(a)) = n 1. Because C(a) is a subgroup of G, we have that n 1 divides n But the only time n 1 divides n is when n = 2.
8. By Sylow’s Third Theorem the number of Sylow 7 subgroups is 1 or 8. So the number of elements of order 7 is 6 or 48.
9. It suffices to show that the correspondence from the set of left cosets of N (H) in G to the set of conjugates of H given by T (xN (H)) = xHx 1 is well defined, onto, and one-to-one. Observe that xN (H) = yN (H) y 1xN (H) = N (H) y 1x N (H) y 1xH(y 1x) 1 = y 1xHx 1y = H xHx 1 = yHy 1 . This shows that T is well defined and one-to-one. By observation, T is onto.
10. Let r denote the number of conjugates of H By Exercise 9 we have r = G : N (H) . Since each conjugate of H has the same order as H and contains the identity and H ⊆ N (H), we know that the union of all the conjugates of H has fewer than |G : N (H)||H| ≤ |G : H||H| = |G| elements because the identity occurs r times in
11. Say cl(x) = {x, g1xg1 1 , g2xg2 1 , . . . , gkxg 1}. If x 1 = gixg 1 , then for each gjxg 1 in cl(x) we have (gjxg 1) 1 = gjx 1g 1 = gj(gixg 1)g 1 ∈ cl(x). Because |G| has odd order, gjxgj 1 /= (gjxgj 1) 1 It follows that |cl(x)| is even. But this contradicts the fact that |cl(x)| divides |G|.
12. By Theorem 9.3, we know that in each case the center of the group is the identity. So, in both cases the first summand is 1. In the case of 39, all the summands after the first one must be 3 or 13. In the case of 55, all the summands after the first one must be 5 or 11. Thus the only possible class equations are 39 = 1 + 3 + 3 + 3 + 3 + 13 + 13; 55 = 1 + 5 + 5 + 11 + 11 + 11 + 11.
13. Part a is not possible by the Corollary of Theorem 23.2. Part b is not possible because it implies that the center would have order 2 and 2 does not divide 21. Part c is the class equation for D5 Part d is not possible because of Corollary 1 of Theorem 23.1.
14. Since Z(G) = {R0, R180} we have two occurrences of 1 in the class equation.
15. Let H and K be distinct Sylow 2-subgroups of G By Theorem 7.2, we have 48 ≥ |HK| = |H||K|/|H ∩ K| = 16 16/|H ∩ K|. This simplifies to |H ∩ K| > 5. Since H and K are distinct and |H ∩ K| divides 16, we have |H ∩ K| = 8.
16. This is immediate by induction.
17. By Example 5 of Chapter 9, ⟨x⟩K is a subgroup. By Theorem 7.2, |⟨x⟩K| = |⟨x⟩||K|/|⟨x⟩ ∩ K| Since K is a Sylow p-subgroup it follows that ⟨x⟩ = ⟨x⟩ ∩ K. Thus ⟨x⟩ ⊆ K.
19. By Theorem 23.5, np, the number of Sylow p-subgroups has the form 1 + kp and np divides G . But if k 1, 1 + kp is relatively prime to pn and does not divide m. Thus k = 0. Now use the corollary to Theorem 23.5.
| | ≥
20. Use Exercise 17.
21. r does not divide st; s does not divide rt; and t does not divide rs.
22. By Sylow, n7 = 1 or 8. If n7 = 8, the Sylow 7-subgroups contain 48 elements of order 7. This means all the elements whose orders are a power of 2 belong to a single Sylow 2-subgroup. So n2 = 1.
23. There are two Abelian groups of order 4 and two of order 9. There are both cyclic and dihedral groups of orders 6, 8, 10, 12, and 14. So, 15 is the first candidate. And, in fact, Theorem 23.5 shows that there is only one group of order 15.
24. n3 = 7, otherwise the group is the internal direct product of subgroups of orders 3 and 7 and such a group is cyclic.
25. Let np denote the number of Sylow p-subgroups of G. By Sylow’s Third Theorem, np = 1, q or q2 Clearly, p does not divide q 1, so np /= q If p is a divisor of q2 1 = (q 1)(q + 1), then p divides q 1 or q + 1. But p > q implies p ≥ q + 2 > q + 1 > q 1. So, np /= q2 Thus, np = 1.
26. Let H be a subgroup of order 3 and K a subgroup of order 5 in one of the six groups. If one of H or K is normal in the whole group then HK is a cyclic subgroup (Theorem 23.6) of order 15. But a permutation of order 15 written in disjoint cycle form requires at least one 3-cycle and one 5-cycle.
27. A group of order 100 has 1, 5 or 25 subgroups of order 4; exactly one subgroup of order 25 (which is normal); at least one subgroup of order 5; and at least one subgroup of order 2.
28. Mimic Example 6.
29. Let H, K, and L be Sylow 5-, 7-, and 17- subgroups of G, respectively. By the Sylow theorems, H is normal in G and by Example 5 in Chapter 9 and Theorem 23.6, HK and HL are cyclic subgroups of G of orders of 35 and 85. Thus, C(H) is divisible by 5, 7, and 17. So, C(H) = G. This means that H is a subgroup of Z(G).
Alternate solution. Let H be a Sylow 5-subgroup. Since the number of Sylow 5-subgroups is 1 mod 5 and divides 7 · 17, the only possibility is 1. So, H is normal in G. Then by the N/C Theorem (Example 16 of Chapter 10), |G/C(H)| divides both 4 and |G|. Thus C(H) = G.
30. 2|b| =
|a|, 16|b| = 1
|a|, 11 16 = 1 mod |a|, 112 = 16 mod |a|.
31. By Theorem 23.6, G/Z(G) would be cyclic and therefore by Theorem 9.3 G would be Abelian. But then G = Z(G).
32. H = 1 or p where p is a prime. To see this, note that if an element g of G has order pn where p is prime, then gn = p. Thus H must divide p. So the only case when H = p is when G has order pk where p is prime. For if G = pkm where p does not divide m, then by Sylow’s First Theorem, G would have an element of some prime order q /= p But then |H| would divide both p and q
33. By Theorem 23.5, np has the form 1 + kp and divides q. Thus, k = 0, and by Corollary 1 of Theorem 23.5, the Sylow p-subgroup of G is normal.
34. By Theorem 23.4 the groups have subgroups H and K of orders 3 and 5, respectively. By Theorem 23.5 n1 is 1 or 25. If n3 = 1, then |HK| = 15. (See Example 5 of Chapter 9.) If n3 = 25, then by Exercise 9 |G : N (H)| = 25 and therefore |N (H)| = 15.
35. Sylow’s Third Theorem (Theorem 23.5) implies that the Sylow 3- and Sylow 5-subgroups are unique. Pick any x not in the union of these. Then |x| = 15.
36. By Sylow, n7 = 1 or 15, and n5 = 1 or 21. Counting elements reveals that at least one of these must be 1. Let H be a Sylow-7 subgroup and K be a Sylow-5 subgroup. Then HK is a subgroup of order 35 and is cyclic by Theorem 23.6. So, HK has φ(5)φ(7) = 24 elements of order 35. If H is normal, then G has 21 4 = 84 elements of order 5. This gives at least 108 elements in G. If K is normal, then G has 15 · 6 = 90 elements of order 7. This gives at least 114 elements in G.
37. By Sylow’s Third Theorem, n17 = 1 or 35. Assume n17 = 35. Then the union of the Sylow 17-subgroups has 561 elements. By Sylow’s Third Theorem, n5 = 1. Thus, we may form a cyclic subgroup of order 85 (Example 5 of Chapter 9 and Theorem 23.6). But then there are 64 elements of order 85. This gives too many elements for the group.
38. By Sylow, n5 = 1 or 6. A5 has 24 elements of order 5.
39. If G = 60 and Z(G) = 4, then by Theorem 23.6, G/Z(G) is cyclic. The “G/Z” Theorem (Theorem 9.3) then tells us that G is Abelian. But if G is Abelian, then Z(G) = G.
40. a. Form a factor group G/N of order 30. The discussion preceding Theorem 23.6 shows G/N has normal subgroups of orders 3, 5, and 15. Now pullback.
b. Let H be a Sylow 2-subgroup containing N . The product of H with the subgroups of orders 6 and 10 have orders 12 and 20.
c. The product of N and the subgroup of order 15 has order 30 and is an internal direct product.
41. Let H be the Sylow 3-subgroup and suppose that the Sylow 5-subgroups are not normal. By Sylow’s Third Theorem, there must be six Sylow 5-subgroups, call them K1, , K6 These subgroups have 24 elements of order 5. Also, the cyclic subgroups HK1, . . . , HK6 of order 15 each have eight generators. Thus, there are 48 elements of order 15. This gives us more than 60 elements in G.
42. Let N be the normal subgroup of order 4. Then by Sylow, G/N has a normal Sylow 7-subgroup whose pullback is normal.
43. We proceed by induction on |G|. By Theorem 23.2 and Theorem 9.5, Z(G) has an element x of order p By induction, the group G/⟨x⟩ has normal subgroups of order pk for every k between 1 and n 1, inclusively. By Exercise 51 in Chapter 10 and Exercise 59 of Chapter 9, every normal subgroup of G/⟨x⟩ has the form H/⟨x⟩, where H is a normal subgroup of G Moreover, if |H/⟨x⟩| = pk, then |H| has order pk+1
44. Let x G have maximum order, x = pt Now let y belong to G Then y = ps pt . Since x has a subgroup of order ps , we have y x . Alternate solution. Suppose that G is not cyclic. Then every element of G belongs to a subgroup of order less than pn . But summing the number of elements in the n proper subgroups we get less that 1 + p + p2 + + pn 1 < pn , which is a contradiction
45. Pick x ∈ Z(G) such that |x| = p. If x /∈ H, then x ∈ N (H) and we are done. If x ∈ H, by induction, N (H/⟨x⟩) > H/⟨x⟩, say y⟨x⟩ ∈ N (H/⟨x⟩) but y⟨x⟩ /∈ H/⟨x⟩ Then y /∈ H and for any h ∈ H we have yhy 1⟨x⟩ = y⟨x⟩h⟨x⟩y 1⟨x⟩ ∈ H/⟨x⟩. So, yhy 1 ∈ H and therefore y ∈ N (H).
46. Since both H and xHx 1 have the same set of conjugates this statement follows directly from Exercise 9.
47. By Sylow’s Second Theorem, K is contained in some Sylow p-subgroup H. Then for any x in G we have K = xKx 1 xHx 1 By Sylow’s Theorem every Sylow p-subgroup has the form xHx 1 . Alternate proof. Let K be a Sylow p-subgroup. Then by Example 5 of Chapter 9 HK is a subgroup of G and by Theorem 7.2 HK = H K / H K . Since this is a power of p, we must have HK = K and because K is contained in HK , we have HK = K Thus H is contained in K
48. If there is only one Sylow 2-subgroup it is normal by the corollary of Theorem 23.5. Otherwise, let H and K be distinct Sylow 2-subgroups. Then, by Lagrange, H K = 1, 2, 4, or 8. If H K = 1, 2 or 4 then the set HK has at least H K / H K 16 16/4 = 64, which is impossible in a group of order 48. If H K = 8, then because H K is a subgroup of both H and K, by Exercise 45, N (H K) contains both H and K. But then 16 divides N (H K) and N (H K) has more than 16 elements. So N (H K) is the entire group and therefore H K is normal in G
49. Sylow’s Third Theorem shows that all the Sylow subgroups are normal. Then Theorem 7.2 and Example 5 of Chapter 9 ensure that G is the internal direct product of its Sylow subgroups. G is cyclic because of Theorems 9.6 and 8.2. G is Abelian because of Theorem 9.6 and Exercise 4 in Chapter 8.
1 2 k
50. Let pn 1 pn2 pn k be the prime power decomposition of |G|. For each pi let Sp be the unique Sylow p-subgroup of G By Corollary 1 of Theorem 23.5 and Exercise 44 we know that each Sp is normal in G and cyclic. It follows from Corollary 1 of Theorem 8.2 and Theorem 9.6 that G is cyclic.
51. Since automorphisms preserve order, we know α(H) = H . But then the corollary of Theorem 23.5 shows that α(H) = H.
52. Clearly, N (H) N (N (H)). Let x belong to N (N (H)). Since H N (H), for any h in H we have that xhx 1 belongs to N (H). By Theorem 7.2, xhx 1 H = xhx 1 H / xHx 1 H Since xhx 1 is a power of p and H is a Sylow p-subgroup we must have xhx 1 = xhx 1 H H. Thus, xhx 1 is in H and x is in N (H).
53. That N (H) = N (K) follows directly from the last part of Sylow’s Third Theorem and Exercise 9.
54. Mimic Example 6. Three pairs are: 5, 7; 7, 11; 11, 13.
55. Normality of H implies cl(h) H for h in H. Thus the conjugacy classes of H obtained by conjugating by elements from G are subsets of H. Moreover, since every element h in H is in cl(h) the union of the conjugacy classes of H is H This is true only when H is normal.
56. By Sylow’s Second Theorem we know H is a subgroup of a Sylow p-subgroup of G, call it K. If H is a proper subgroup of K, then by Exercise 43 there is an element a in the normalizer of H restricted to K that is not in H. Then H a is a p-subgroup of N (H) of order larger than H. But then H is not a Sylow p-subgroup of N (H).
57. Suppose that G is a group of order 12 that has nine elements of order 2. By the Sylow Theorems, G has three Sylow 2-subgroups whose union contains the identity and the nine elements of order 2. If H and K are both Sylow 2-subgroups, by Theorem 7.2 H K = 2. Thus the union of the three Sylow 2-subgroups has at most 7 elements of order 2 since there are 3 in H, 2 more in K that are not in H, and at most 2 more that are in the third but not in H or K.
58. By way of contradiction, assume that H is the only Sylow 2-subgroup of G and that K is the only Sylow 3-subgroup of G Then H and K are normal and Abelian (corollary to Theorem 23.5 and corollary to Theorem 23.2). So, G = H × K ≈ H ⊕ K and, from Exercise 4 of Chapter 8, G is Abelian.
59. By Lagrange’s Theorem, any nontrivial proper subgroup of G has order p or q It follows from Theorem 23.5 and its corollary that there is exactly one subgroup of order q which is normal (for otherwise there would be (q + 1)(q 1) = q2 1 elements of order q). On the other hand, there cannot be a normal subgroup of order p, for then G would be an internal direct product of a cyclic group of q and a cyclic group of order p, which is Abelian. So, by Theorem 23.5 there must be exactly q subgroups of order p.
60. Mimic Example 6.
61. Note that any subgroup of order 4 in a group of order 4m where m is odd is a Sylow 2-subgroup. By Sylow’s Third Theorem, the Sylow 2-subgroups are conjugate and therefore isomorphic. S4 contains both the subgroups ⟨(1234)⟩ and {(1), (12), (34), (12)(34)}.
62. By the “G/Z Theorem” (9.3), |N (H)/C(H)| divides |Aut(H)|. Since H is cyclic we know that C(H) ⊇ H and therefore |N (H)/C(H)| is relatively prime to p Letting k k |H| = p , we have by Theorem 6.5 that Aut(H) is isomorphic to U (p ) and by the formula given in Chapter 8 we have |U (pk)| = pk 1(p 1). Since the smallest prime divisor of |N (H)/C(H)| is greater than p, we must have |N (H)/C(H)| = 1.
63. By Sylow’s Third Theorem, the number of Sylow 13-subgroups is equal to 1 mod 13 and divides 55. This means that there is only one Sylow 13-subgroup, so it is normal in G. Thus N (H)/C(H) = 715/ C(H) divides both 55 and 12. This forces 715/ C(H) = 1 and therefore C(H) = G. This proves that H is contained in Z(G). Applying the same argument to K we get that K is normal in G and N (K)/C(K) = 715/ C(K) divides both 65 and 10. This forces 715/ C(K) = 1 or 5. In the latter case K is not contained in Z(G).
64. In each case the number of Sylow 3-subgroups is the number of elements of order 3 divided by 2 since a group of order 3 has two elements of order 3 and no two of subgroups of order 3 share an element of ordder 3. So, In S4 there are 4 · 3 · 2/3 = 8 elements of order 3 and 4 subgroups of order 3; in S5 there are 5 4 3/3 = 20 elements of order 3 and 10 subgroups of order 3;
65. Using Sylow’s third theorem and counting the number of elements of orders 2, 3, and 5, we have that the number of Sylow p-subgroups in each respective cases is 15, 10, and 6.
66. By the previous exercise and Corollary 2 of Theorem 23.5, we know that |N (H)| = 10. Also, by the classification of groups of order 2p, where p is a prime (Theorem 7.3), we know that N (H) is isomorphic to Z10 or D5 . However, A5 has no element of order 10.
67. One of many is (12)(45).
CHAPTER 24
Finite Simple Groups
1. This follows directly from the “2·odd” Theorem (Theorem 24.2).
2. We may assume that n5 = 56 and n7 = 8. Then counting elements forces n2 = 1. Alternate proof. From n5 = 56 and n7 = 8, we have that for any Sylow 7-subgroup L7, N |(L7)| = 35. Because N (L7) is a cyclic group, it has a Sylow 5-subgroup L5 whose normalize N (L5) contains N (L7). But n5 = 56 implies that |N (L5)| = 5.
3. By the Sylow Theorems, if there were a simple group of order 216 the number of Sylow 3-subgroups would be 4. Then the normalizer of a Sylow 3-subgroup would have index 4. The Index Theorem (corollary of Theorem 24.3) then gives a contradiction.
4. Observe that n5 = 6 and use the Index Theorem.
5. Suppose G is a simple group of order 525. Let L7 be a Sylow 7-subgroup of G. It follows from Sylow’s theorems that |N (L7)| = 35. Let L be a subgroup of N (L7) of order 5. Since N (L7) is cyclic (Theorem 24.6), N (L) ≥ N (L7), so that 35 divides |N (L)|. But L is contained in a Sylow 5-subgroup (Theorem 23.4), which is Abelian (see the Corollary to Theorem 23.2). Thus, 25 divides |N (L)| as well. It follows that 175 divides |N (L)|. The Index Theorem now yields a contradiction.
6. The Index Theorem rules out n5 = 6. So we may assume that n3 = 7 and n5 = 21. Let L5 be a Sylow 5-subgroup. Then n5 = 21 implies that |N (L5)| = 15. By Theorem 23.6, N (L5) has an element of order 15 and by the Embedding Theorem (Corollary 2 of Theorem 24.3), G is isomorphic to a subgroup of A7. But A7 does not have an element of order 15 because any such element written in disjoint cycle form requires at least one 3-cycle and one 5-cycle.
7. Suppose that there is a simple group G of order 528 and L11 is a Sylow 11-subgroup. Then n11 = 12, |N (L11)| = 44, and G is isomorphic to a subgroup of A12. Then by the N/C Theorem in Example 17 of Chapter 10, |N (L11)/C(L11)| divides |Aut(Z11)| = 10, |C(L11)| = 22 or 44. In either case, C(L11) has elements of order 2 and 11 that commute. But then C(L11) has an element of order 22 whereas A12 does not.
8. We may assume the n3 = 10 and n5 = 6 or 36. The Index Theorem rules of 6. Let L5 be a Sylow 5-subgroup. This implies that |N (L5 )| = 15. By Theorem 23.6, we know that N (L5) is cyclic. Let K be the subgroup of N (L5) of order 3 and let L3 be a Sylow subgroup of G that contains K (see Sylow’s Second Theorem). Then L5K is a cyclic subgroup of N (K) of order 15. So, 15 divides N (K) . Moreover, by Exercise 45 of Chapter 23 and Sylow’s First Theorem, N (K) contains a subgroup K′ of L3 of order 9. Thus L5K′ is a group of order 45. Because groups of order 45 are Abelian, (see Example 6 in Chapter 23), N (L5) has order at least 45. But n5 = 36 means |N (L5 )| = 15.
9. Suppose that there is a simple group G of order 396 and L11 is a Sylow 11-subgroup. Then n11 = 12, |N (L11)| = 33, and G is isomorphic to a subgroup of A12. Since
|N (L11)/C(L11)| divides |Aut(Z11)| = 10, |C(L11)| = 33. Then C(L11) has elements of order 3 and 11 that commute. But then C(L11) has an element of order 33 whereas A12 does not.
10. 211, 223, 227, 229 and 233 are prime. The 2 odd test rules out 202, 206, 210, 214, 218, 222, 226, 230 and 234. The Index Theorem rules out 216 and 224. The Sylow test rules out the remaining cases.
11. If we can find a pair of distinct Sylow 2-subgroups A and B such that |A ∩ B| = 8, then |N (A ∩ B) ≥ AB|, so that N (A ∩ B) = G. Now let H and K be any pair of distinct Sylow 2-subgroups. Then 16 · 16/|H ∩ K| = |HK| ≤ 112 (Theorem 7.2), so that |H ∩ K| is at least 4. If |H ∩ K| = 8, we are done. So, assume |H ∩ K| = 4. Then N (H ∩ K) picks up at least 8 elements from H and at least 8 from K (see Exercise 45 of Chapter 23). Thus, |N (H ∩ K)| ≥ 16 and is divisible by 8. So, |N (H ∩ K)| = 16, 56, or 112. Since the latter two cases imply that G has a normal subgroup, we may assume |N (H ∩ K)| = 16. If N (H ∩ K) = H, then |H ∩ K| = 8, since N (H ∩ K) contains at least 8 elements from K. So, we may assume that N (H ∩ K) /= H. Then, we may take A = N (H ∩ K) and B = H.
12. Suppose that H is a subgroup of Sn of order n!/2 other than An. Then AnH = Sn and it follows from Theorem 7.2 that An H is a subgroup of An of index 2. But a subgroup of index 2 is normal (see Exercise 9 in Chapter 9).
13. If H is a proper subgroup of An+1 of order greater than n!/2, then [An+1 : H] < [An+1 : An] = n + 1 and it follows from the Embedding Theorem that An+1 is isomorphic to a subgroup of An, which is impossible.
14. Theorem 23.2 handles the case where p = q = r The Sylow Test for nonsimplicity (Theorem 24.1) shows that a group of order p2 q with p > q must have a normal Sylow p-subgroup. The same theorem implies that a simple group of order p2 q with p < q would have p2 subgroups of order q and more than one subgroup of order p2 Counting elements then yields a contradiction.
Finally consider a simple group of order pqr where p < q < r Then there are pq subgroups of order r, at least r subgroups of order q and at least q subgroups of order p. But pq(r 1) + r(q 1) + q(p 1) > pqr.
15. If A5 had a subgroup of order 30, 20, or 15, then there would be a subgroup of index 2, 3 or 4. But then the Index Theorem gives us a contradiction to the fact that G is simple.
16. The argument for Exercise 15 applies here as well.
17. (Solution by Gurmeet Singh) By Sylow’s Third Theorem we know that number of Sylow 5-subgroups is 6. This means that 6 is the index of the normalizer of a Sylow 5-subgroup. But then, by the Embedding Theorem, G is isomorphic to a subgroup of order 120 in A6 This contradicts Exercise 16.
18. Use Exercise 51 of Chapter 9.
19. Let α be as in the proof of the Generalized Cayley Theorem (Theorem 24.3). Then, if g Ker α we have gH = Tg(H) = H so that Ker α H. Since α(G) consists of a group of permutations of the left cosets of H in G, we know by the First Isomorphism Theorem (Theorem 10.3) that G/Ker α is isomorphic to a subgroup of S|G:H|. Thus, |G/Ker α| divides |G : H|!. Since Ker α ⊆ H, we have that
G : H H : Ker α = G : Ker α must divide G : H ! = G : H ( G : H 1)!. Thus, H : Ker α divides ( G : H 1)!. Since H and ( G : H 1)! are relatively prime, we have H : Ker α = 1 and therefore H = Ker α. So, by the Corollary of Theorem 10.2, H is normal. In the case that a subgroup H has index 2, we conclude that H is normal.
20. If φ is such a homomorphism then by Theorem 10.3 G/Ker φ ≈ Sn Since |G| > n! we must have |Ker φ| > 1. So, Ker φ is a proper, non-trivial normal subgroup of G.
21. If H is a proper normal subgroup of S5, then H ∩ A5 = A5 or {ε} since A5 is simple and H ∩ A5 is normal. But H ∩ A5 = A5 implies H = A5, whereas H ∩ A5 = {ε} implies H = {ε} or |H| = 2 because in a subgroup of Sn every element is even or half of its elements are even. (See Exercise 27 of Chapter 5.) If |H| = 2, then it has an element of the form (ab). But for any c ∈ {1, 2, . . . , 5} other than a or b, (ac)(ab)(ac) /= (ab). So this case is ruled out.
22. The Sylow Test for Nonsimplicity yields n5 = 6 and n3 = 4 or 10. The Index Theorem rules out n3 = 4. Let L3 be the Sylow 3-subgroup of the group and L5 be the Sylow 5-subgroup of the group. Then, by Corollary 1 of Theorem 24.3, we have |N (L3 | = 6 and |N (L5 )| = 10.
23. From Table 5.1 we see that the Sylow 2-subgroup of A4 is unique and therefore normal in A4 .
24. Let H = (12), (12345) First note that (12345)(12)(12345) 1 = (23) implies that (12)(12345)(12)(12345) 1 = (12)(23) = (123) so H contains an element of order 3. Moreover, (12345)(12) = (1345) H This means that H is divisible by 3, 4, and 5 and therefore H = 60 or 120. But H cannot be 60 for if so, then the subset of even permutations in H would be a subgroup of order 30 (see Exercise 27 in Chapter 5). This means that A5 would have a subgroup of index 2, which would a normal subgroup. This contradicts the simplicity of A5.
25. Suppose that S5 has a subgroup H that contains a 5-cycle α and a 2-cycle β. Say β = (a1a2). Then there is some integer k such that αk = (a1a2a3a4a5). Note that (a1a2a
) 1(a
)(
= (a1a2a5), so H contains an element of order 3. Moreover, since α 2βα2 = (a4
), H contains the subgroup (1), (a1a2), (a4a5), (a1a2)(a4a5) . This means that H is divisible by 60. But H cannot be 60 for if so, then the subset of even permutations in H would be a subgroup of order 30 (see Exercise 27 in Chapter 5). This means that A5 would have a subgroup of index 2, which would be a normal subgroup. This contradicts the simplicity of A5
26. Let p = G : H and q = G : K It suffices to show that p = q But if p < q, say, then q does not divide p!. This contradicts the Index Theorem.
27. Suppose there is a simple group of order 60 that is not isomorphic to A5 The Index Theorem implies n2 /= 1 or 3, and the Embedding Theorem implies n2 /= 5. Thus, n2 = 15. If every pair of Sylow 2-subgroups has only the identity element in common, then the union of the 15 Sylow 2-subgroups has 46 elements. But n5 = 6, so there are also 24 elements of order 5. This gives more than 60. As was the case in showing that there is no simple group of order 144, the normalizer of this intersection has index 5, 3, or 1. But the Embedding Theorem and the Index Theorem rule these out.
28. Say G is a simple group with a subgroup of index 4. By the Embedding Theorem G is isomorphic to a subgroup of A4 . Since |G| is divisible by 4, we must have that |G| = 4 or 12. The subgroup of A4 of order 4 is isomorphic to Z2 × Z2 so it is not simple. Since the Sylow 2-subgroup of A4 is the unique subgroup of order 4 it is normal in A4, A4 is not simple.
29. Suppose there is a simple group G of order p2q where p and q are odd primes and q > p. Since the number of Sylow q-subgroups is 1 mod q and divides p2 , it must be p2 Thus there are p2 (q 1) elements of order q in G These elements, together with the p2 elements in one Sylow p-subgroup, account for all p2 q elements in G. Thus there cannot be another Sylow p-subgroup. But then the Sylow p-subgroup is normal in G.
30. Let L and M be distinct maximal subgroups of G. Since N (K) contains both L and M it properly contains them. Since L is maximal, N (K) = G Because G is simple, we have K = {e}.
31. Consider the right regular representation of G Let g be a generator of the Sylow 2-subgroup and suppose that G = 2k n where n is odd. Then every cycle of the permutation Tg in the right regular representation of G has length 2k . This means that there are exactly n such cycles. Since each cycle is odd and there is an odd number of them, Tg is odd. This means that the set of even permutations in the regular representation has index 2 and is therefore normal. (See Exercise 27 in Chapter 5 and Exercise 9 in Chapter 9.)
32. If S5 had a subgroup H of order 40 or 30, then H S5 would be a subgroup of A5 of order 20, 30 or 15 since every element of H is even or half of them are even (40 is excluded by Lagrange’s Theorem). By Exercise 15, this is impossible.
33. If φ is such a homomorphism, then by Theorem 10.3, G/Ker φ ≈ Sn Since |G| > n! we must have |Ker φ| > 1. So, Ker φ is a proper, non-trivial normal subgroup of G.
34. See Exercise 6.
CHAPTER 25
Generators and Relations
1. u u because u is obtained from itself by no insertions; if v can be obtained from u by inserting or deleting words of the form xx 1 or x 1x, then u can be obtained from v by reversing the procedure; if u can be obtained from v and v can be obtained from w, then u can be obtained from w by first obtaining v from w then u from v
2. Let a be any reflection in Dn and let b = aR360/n. Then aZ(Dn) and bZ(Dn) have order 2 and generate Dn /Z(Dn ). Now use Theorem 25.5 and the fact that |Dn /Z(Dn )| = n = |Dn/2|.
4. Since b = b 1, we have bab = a2 . Then a = a6 = (bab)3 = ba3b so that ba = a3b. Thus, a3b = a2b and a = e. Finally, note that Z2 satisfies the relations with a = 0 and b = 1.
5. Let F be the free group on {a1, a2, , an} Let N be the smallest normal group containing {w1, w2, . . . , wt} and let M be the smallest normal subgroup containing w1, w2, , wt, wt+1, , wt+k Then F/N G and F/M G The homomorphism from F/N to F/M given by aN aM induces a homomorphism from G onto G ¯ .
To prove the corollary, observe that the theorem shows that K is a homomorphic image of G, so that |K| ≤ |G|.
6. Use the Corollary to Dyck’s Theorem.
7. Clearly, a and ab belong to ⟨a, b⟩, so ⟨a, ab⟩ ⊆ ⟨a, b⟩. Also, a and a 1(ab) = b belong to ⟨a, ab⟩
8. Use Theorem 25.5.
9. By Exercise 7, x, y = x, xy Also, (xy)2 = (xy)(xy) = (xyx)y = y 1y = e, so by Theorem 25.5, G is isomorphic to a dihedral group and from the proof of Theorem 25.5, |x(xy)| = |y| = n implies that G ≈ Dn.
10. 3. ⟨x, y, z | x2 = y2 = z2 = e, xy = yx, xz = zx, yz = zy⟩
11. Since x2 = y2 = e, we have (xy) 1 = y 1x 1 = yx. Also, xy = z 1yz, so that (xy) 1 = (z 1yz) 1 = z 1y 1z = z 1yz = xy.
12. a. b0a = a b. ba
13. First note that b2 = abab implies that b = aba.
a. So, b2abab3 = b2(aba)b3 = b2bb3 = b6 .
b. Also, b3abab3a = b3(aba)b3a = b3bb3a = b7a.
14. Observe that G is generated by 6 and 4 where 6 = 2, 4 = 2 and 6 4 = 3 = n Now use Theorem 26.5.
15. First observe that since xy = (xy)3(xy)4 = (xy)7 = (xy)4(xy)3 = yx, x and y commute. Also, since y = (xy)4 = (xy)3xy = x(xy) = x2y we know that x2 = e. Then y = (xy)4 = x4y4 = y4 and therefore, y3 = e. This shows that |G| ≤ 6. But Z6 satisfies the defining relations with x = 3 and y = 2. So, G ≈ Z6
16. In H, |xy| = 6.
17. Note that yxyx3 = e implies that yxy 1 = x5 and therefore ⟨x⟩ is normal. So, G = ⟨x⟩ ∪ y⟨x⟩ and |G| ≤ 16. From y2 = e and yxyx3 = e, we obtain yxy 1 = x 3 . So, yx2y 1 = yxy 1yxy 1 = x 6 = x2 . Thus, x2 ∈ Z(G). On the other hand, G is not Abelian for if so we would have e = yxyx3 = x4 and then |G| ≤ 8. It now follows from the “G/Z” Theorem (Theorem 9.3) that |Z(G)| /= 8. Thus, Z(G) = ⟨x2⟩. Finally, (xy)2 = xyxy = x(yxy) = xx 3 = x 2 , so that |xy| = 8.
18. Because a is a subgroup of Q6 of index 2 it is normal. Thus every element of Q6 can written in the form ai or aibj where 0 i < 6. The relation b2 = a3 reduces the aibj terms to just aib. To see that the only element of order 2 is a3 = b2 first observe that the relation b 1ab = a 1 implies that b 1aib = (b 1ab)i = ai and therefore aib = ba i . Finally, we note that e = (aib)2 = aibaib = ba iaib = b2 .
19. Since the mapping from G onto G/N given by x xN is a homomorphism, G/N satisfies the relations defining G.
20. If G were Abelian then the relation st = ts could be derived from sts = tst. But this same derivation would hold when s = (23) and t = (13). However, (23)(13) /= (13)(23).
21. For H to be a normal subgroup, we must have yxy 1 ∈ H = {e, y3 , y6 , y9 , x, xy3 , xy6 , xy9}. But yxy 1 = yxy11 = (yxy)y10 = xy10 .
22. Every element has the form xi or xiy where 0 ≤ i < 2n Let 0 < i < 2n Then xi ∈ Z(G) if and only if y 1xiy = xi . But y 1 xiy = (y 1 xy)i = (x 1 )i = x 1
So x2i = e This implies i = n A similar argument shows xiy Z(G) implies i = n But xny Z(G) and xn Z(G) imply y Z(G), which is false. So xny Z(G). To prove the second portion, observe that G/Z(G) has order 2n and is generated by a pair of elements of order 2.
23. First note that b 1 a2 b = (b 1 ab)(b 1 ab) = a3 a3 = a6 = e. So, a2 = e. Also, b 1ab = a3 = a implies that a and b commute. Thus, G is generated by an element of order 2 and an element of order 3 that commute. It follows that G is Abelian and has order at most 6. But the defining relations for G are satisfied by Z6 with a = 3 and b = 2. So, G ≈ Z6.
24. Since yx = x3y, the set S = {⟨y⟩, x⟨y⟩, x2⟨y⟩, x3⟨y⟩} is closed under multiplication on the left by x and y. Thus every element of G has the form xiyj with 0 ≤ i < 4 and 0 ≤ j < 4.
To compute the center observe that xyx = y and xy = yx3 . So x2 y = x(xy) = xyx3 = (xyx)x2 = yx2
Thus x2 ∈ Z(G). Also, xy2 = (xy)y =
2 = y(xyx)x = y2x
so that y2 ∈ Z(G). It follows that Z(G) = {e, x2 , y2 , x2y2}. (Theorem 9.3 shows |Z(G)| /= 8.)
Finally, observe that G/ y2 has order 8 and is generated by y y2 and xy y2 each of which has order 2.
25. In the notation given in the proof of Theorem 25.5 we have that |e| = 1, |a| = |b| = 2, ab = ba = . Next observe that since every element of D∞ can be expressed as a string of alternating a’s and b’s or alternating b’s and a’s, every element can be expressed in one of four forms: (ab)n , (ba)n , (ab)n a, or (ba)n b for some n Since ab = ba = , we have (ab)n = (ba)n = (excluding n = 0). And, since ((ab)na)2 = (ab)na(ab)na = (ab)(ab) (ab)a(ab)(ab) (ab)a, we can start at the middle and successively cancel the adjacent a’s, then adjacent b’s, then adjacent a’s, and so on to obtain the identity. Thus, |(ab)na| = 2. Similarly, |(ba)nb| = 2.
26. Use Theorem 25.4.
27. First we show that d = b 1 , a = b2 and c = b3 so that G = b . To this end, observe that ab = c and cd = a together imply that cdc = c and therefore d = b 1 Then da = b and d = b 1 together imply that a = b2 . Finally, cd = a and d = b 1 together imply c = b3 . Thus G = b . Now observe that bc = d, c = b3 , and d = b 1 yield b5 = e So G = 1 or 5. But Z5 satisfies the defining relations with a = 1, b = 3, c = 4, and d = 2.
28. From Theorem 25.5 and its proof, the group is dihedral and has order 2|ab|. To compute |ab|, note that (ab)3 = (aba)(bab) = (aba(aba) = aba2ba = e. So, the group is D3.
29. Since aba 1b 1 = e, G is an Abelian group of order at most 6. Then because Z6 satisfies the given relations, we have that G is isomorphic to Z6
30. F ⊕ Z3 where F is the free group on two letters.
32. There are only five groups of order 8: Z8 and the quaternions have only one element of order 2; Z4 ⊕ Z2 has 3; Z2 ⊕ Z2 ⊕ Z2 has 7; and D4 has 5.
CHAPTER 26
Symmetry Groups
1. If T is a distance-preserving function and the distance between points a and b is positive, then the distance between T (a) and T (b) is positive.
2. For any fixed v ′ in Rn define
Then Tv ′ v ′ Rn is the set of translations of Rn Closure and associativity follow from the observation
v
◦ Tw ′ = Tw
+v ′ ; T0 is the identity; (Tv ′ ) 1 = T v ′ .
3. See Figure 1.5.
4. Use Theorem 7.2.
5. There are rotations of 0◦ , 120◦ and 240◦ about an axis through the centers of the triangles and a 180◦ rotation through an axis perpendicular to a rectangular base and passing through the center of the rectangular base. This gives 6 rotations. Each of these can be combined with the reflection plane perpendicular to the base and bisecting the base. So, the order is 12.
6. 16
7. There are n rotations about an axis through the centers of the n-gons and a 180◦ rotation through an axis perpendicular to a rectangular base and passing through the center of the rectangular base. This gives 2n rotations. Each of these can be combined with the reflection plane perpendicular to a rectangular base and bisecting the base. So, the order is 4n
8. A drawing or model reveals the group consists of the identity, three 180◦ rotations and 4 reflections and is Abelian.
9. In R1, there is the identity and an inversion through the center of the segment. In R2 , there are rotations of 0◦ and 180◦ , a reflection across the horizontal line containing the segment, and a reflection across the perpendicular bisector of the segment. In R3, the symmetry group is G Z2, where G is the plane symmetry group of a circle. (Think of a sphere with the line segment as a diameter. Then G includes any rotation of that sphere about the diameter and any plane containing the diameter of the sphere is a symmetry in G. The Z2 must be included because there is also an inversion.)
10. No symmetry; symmetry across a horizontal axis only; symmetry across a vertical axis only; symmetry across a horizontal axis and a vertical axis.
11. There are 6 elements of order 4 since for each of the three pairs of opposite squares there are rotations of 90◦ and 270◦
12. It is the same as a 180◦ rotation.
13. An inversion in R3 leaves only a single point fixed, while a rotation leaves a line fixed.
14. A rotation of 180◦ about the line L
15. In R4 , a plane is fixed. In Rn , a hyperplane of dimension n 2 is fixed.
16. Consider a triangle whose sides have lengths a, b, c. The image of this triangle is also a triangle whose sides have lengths a, b, c Thus the two triangles are congruent (side-side-side).
17. Let T be an isometry, let p, q, and r be the three noncollinear points, and let s be any other point in the plane. Then the quadrilateral determined by T (p), T (q), T (r), and T (s) is congruent to the one formed by p, q, r, and s. Thus, T (s) is uniquely determined by T (p), T (q), and T (r).
18. Use Exercise 17.
19. The only isometry of a plane that fixes exactly one point is a rotation.
20. A translation a distance twice that between a and b along the line joining a and b
CHAPTER 27
Symmetry and Counting
1. The symmetry group is D4 . Since we have two choices for each vertex, the identity fixes 16 colorings. For R90 and R270 to fix a coloring, all four corners must have the same color so each of these fixes 2 colorings. For R180 to fix a coloring, diagonally opposite vertices must have the same color. So, we have 2 independent choices for coloring the vertices and we can choose 2 colors for each. This gives 4 fixed colorings for R180 . For H and V , we can color each of the two vertices on one side of the axis of reflection in 2 ways, giving us 4 fixed points for each of these rotations. For D and D′ , we can color each of the two fixed vertices with 2 colors and then we are forced to color the remaining two the same. So, this gives us 8 choices for each of these two reflections. Thus, the total number of colorings is
2. 21
3. The symmetry group is D3 . There are 53 5 = 120 colorings without regard to equivalence. The rotations of 120◦ and 240◦ can fix a coloring only if all three vertices of the triangle are colored the same so they each fix 0 colorings. A particular reflection will fix a coloring provided that fixed vertex is any of the 5 colors and the other two vertices have matching colors. This gives 5 4 = 20 for each of the three reflections. So, the number of colorings is
(120 + 0 + 0 + 3 20) = 30. 6
4. 92
5. The symmetry group is D6 The identity fixes all 26 = 64 arrangements. For R60 and R300 , once we make a choice of a radical for one vertex, all others must use the same radical. So, these two fix 2 arrangements each. For R120 and R240 to fix an arrangement, every other vertex must have the same radical. So, once we select a radical for one vertex and a radical for an adjacent vertex, we then have no other choices. So we have 22 choices for each of these rotations. For R180 to fix an arrangement, each vertex must have the same radical as the vertex diagonally opposite it. Thus, there are 23 choices for this case. For the 3 reflections whose axes of symmetry join two vertices, we have 2 choices for each fixed vertex and 2 choices for each of the two vertices on the same side of the reflection axis. This gives us 16 choices for each of these 3 reflections. For the 3 reflections whose axes of reflection bisect opposite sides of the hexagon, we have 2 choices for each of the 3 vertices on the same side of the reflection axis. This gives us 8 choices for each of these 3 reflections. So, the total number of arrangements is 1
7. The symmetry group is D4 . The identity fixes 6 5 4 3 = 360 colorings. All other symmetries fix 0 colorings because of the restriction that no color be used more than once. So, the number of colorings is 360/8 = 45.
8. 231
9. The symmetry group is D11. The identity fixes 211 colorings. Each of the other 10 rotations fixes only the two colorings in which the beads are all the same color. (Here we use the fact that 11 is prime. For example, if the rotation R2·360/11 fixes a coloring, then once we choose a color for one vertex, the rotation forces all other vertices to have that same color because the rotation moves 2 vertices at a time and 2 is a generator of Z11.) For each reflection, we may color the vertex containing the axis of reflection 2 ways and each vertex on the same side of the axis of reflection 2 ways. This gives us 26 colorings for each reflection. So, the number of different colorings is
10. 57
11. The symmetry group is Z6 . The identity fixes all n6 possible colorings. Since the rotations of 60◦ and 300◦ fix only the cases where each section is the same color, they each fix n colorings. Rotations of 120◦ and 240◦ each fix n2 colorings since every other section must have the same color. The 180◦ rotation fixes n3 colorings, since once we choose colors for three adjacent sections, the colors for the remaining three sections are determined. So, the number is
(n6 + 2 · n + 2 · n2 + n3).
12. 51
13. The first part is Exercise 13 in Chapter 6. For the second part, observe that in D4 we have φR0 = φR180 .
14. γg1 g 2 (x) = (g1g2)xH
15. R0, R180, H, V act as the identity and R90, R270, D, D′ interchange L1 and L2. Then the mapping g → γg from D4 to sym(S) is a group homomorphism with kernel {R0 , R180 , H, V }.
6. Say we proceed from x to y via the generators a1 , a2 , . . . , am and via the generators b1, b2, . . . , bn. Then y = xa1a2 · · · am = xb1b2 · · · bn so that a1a2 · · · am = b1b2 · ·
7. Both yield paths from e to a3b.
8. Cay{{(1, 0), (0, 1)} : Z4 ⊕ Z2}
10.
11. Say we start at x Then we know the vertices x, xs1, xs1s2, , xs1s2 · · · sn 1 are distinct and x = xs1s2 · · · sn. So if we apply the same sequence beginning at y, then cancellation shows that y, ys1, ys1s2, . . . , ys1s2 sn 1 are distinct and y = ys1s2 sn.
12. Trace the sequence b, b, b, a, b, b, b. The digraph could be called undirected because whenever x is connected to y, y is connected to x. Such a digraph (that is, one in which all arrows go both directions) is called a graph
13. If there were a Hamiltonian path from (0, 0) to (2, 0), there would be a Hamiltonian circuit in the digraph, since (2, 0) + (1, 0) = (0, 0). This contradicts Theorem 28.1.
14. Cay({2, 3} : Z6) does not have a Hamiltonian circuit.
15. a. If s1, s2, . . . , sn 1 traces a Hamiltonian path and sisi+1 sj = e, then the vertex s1s2 · · · si 1 appears twice. Conversely, if sisi+1 · · · sj /= e, then the sequence e, s1, s1s2, . . . , s1s2 sn 1 yields the n vertices (otherwise, cancellation gives a contradiction).
· · ·
b. This is immediate from part a.
16. The digraph is the same as those shown in Example 3 except all arrows go in both directions.
17. The sequence traces the digraph in a clockwise fashion.
18. A circuit is 4 ∗ ((3 ∗ a), b).
19. Abbreviate (a, 0), (b, 0), and (e, 1) by a, b, and 1, respectively. A circuit is 4 ∗ (4 ∗ 1, a), 3 ∗ a, b, 7 ∗ a, 1, b, 3 ∗
, 4 ∗ a, 1, 3 ∗ a, b, 3 ∗ a, b, 3 ∗ a, b.
20. Notice that the digraph has four triangles. Start somewhere and call that triangle 1. Now once you enter any of the other three triangles, you must cover all three points before leaving it. The digraph does have a Hamiltonian path, starting at vertex (124) and ending at vertex (1).
21. Abbreviate (R90, 0), (H, 0), and (R0, 1) by R, H, and 1, respectively. A circuit is 3 ∗ (R, 1, 1), H, 2 ∗ (1, R, R), R, 1, R, R, 1, H, 1, 1.
22. Abbreviate (a, 0), (b, 0) and (e, 1) by a, b and 1 respectively. A circuit is 2 ∗ (3 ∗ (a, b), a, 1, 3 ∗ a, b, 3 ∗ a, 1).
23. Abbreviate (a, 0), (b, 0), and (e, 1) by a, b, and 1, respectively. A circuit is 2 ∗ (1, 1, a), a, b, 3 ∗ a, 1, b, b, a, b, b, 1, 3 ∗ a, b, a, a.
24. Abbreviate (a, 0), (b, 0) and (e, 1) by a, b and 1 respectively. A circuit is (m 1) ∗ 1, a, 2 ∗ 1, ((m 3)/2) ∗ [2 ∗ a, b, 3 ∗ a, 1, b, b, a, 3 ∗ b, 1], 2 ∗ a, b, 3 ∗ a, 1, b, b, a, 2 ∗ b, 1, 3 ∗ a, b, a, a.
25. Abbreviate (r, 0), (f, 0), and (e, 1) by r, f , and 1, respectively. Then the sequence is r, r, f , r, r, 1, f , r, r, f , r, 1, r, f , r, r, f , 1, r, r, f , r, r, 1, f , r, r, f , r, 1, r, f , r, r, f , 1.
26. Abbreviate (r, 0), (f, 0) and (e, 1) by r, f and 1 respectively. A circuit in Dn Zn+1 is (n 1) ∗ (n ∗ 1, r), n ∗ 1, f, n ∗ ((n 1) ∗ r, 1), (n 1) ∗ r, f.
27. m ∗ ((n 1) ∗ (0, 1), (1, 1))
28. Adapt the argument given in the proof of Theorem 28.1.
29. Abbreviate (r, 0), (f, 0), and (e, 1) by r, f , and 1, respectively. A circuit is 1, r, 1, 1, f, r, 1, r, 1, r, f, 1.
30. Abbreviate (a, 0), (b, 0) and (e, 1) by a, b and 1 respectively. Then a circuit is 1, a, 1, 1, b, a, 1, a, 1, a, b, 1, 1, 2 ∗ (a, 1, 1, b, a, 1, a, 1, a, b, 1, 1), a, 1, 1, b, a, 1, a, 1, a, b, 1.
31. 5 ∗ [3 ∗ (1, 0), (0, 1)], (0, 1)]
32. 12 ∗ ((1, 0), (0, 1)).
33. 12 ∗ ((1, 0), (0, 1))
34.
35. Letting V denote a vertical move and H a horizontal move and starting at (1,0), a circuit is V, V, H, 6 ∗ (V, V, V, H).
36. It suffices to show that x travels by a implies xab 1 travels by a (for we may successively replace x by xab 1). If xab 1 traveled by b, then the vertex xa would appear twice in the circuit.
37. In the proof of Theorem 28.3, we used the hypothesis that G is Abelian in two places: We needed H to satisfy the induction hypothesis, and we needed to form the factor group G/H. Now, if we assume only that G is Hamiltonian, then H also is Hamiltonian and G/H exists.
7. By using t = 1/2 in the second part of the proof of Theorem 29.2 we have that all single errors can be detected.
8. C′ can detect any 3 errors, whereas C can only detect any 2 errors.
9. Observe that a vector has even weight if and only if it can be written as a sum of an even number of vectors of weight 1. So, if u can be written as the sum of 2m vectors of even weight and v can be written as the sum of 2n vectors of even weight, then u + v can be written as the sum of 2m + 2n vectors of even weight and therefore the set of code words of even weight is closed. (We need not check that the inverse of a code word is a code word since every binary code word is its own inverse.)
10. Since the minimum weight of any nonzero member of C is 4, we see by Theorem 29.2 that C will correct any single error and detect any triple error. (To verify this, use t = 3/2 in the last paragraph of the proof for Theorem 29.2.)
11. No, by Theorem 29.3.
12. H = 2 2 1 0 0 1
2
The code is 0000, 1011, 2022, 0121, 0212, 1102, 2201, 2110, 1220 . It will correct any single error and detect any double error. 2201.
Yes, the code will detect any single error because it has weight 3.
14. Observe that the subset of code words that end with 0 is a subgroup H If H is a proper subgroup, note that it has index 2. The same is true for every component.
15. Suppose u is decoded as v, and x is the coset leader of the row containing u. Coset decoding means v is at the head of the column containing u So, x + v = u and x = u v. Now suppose u v is a coset leader and u is decoded as y. Then y is at the head of the column containing u. Since v is a code word, u = u v + v is in the row containing u v Thus u v + y = u and y = v
16. For 11101 we get 11100 or 11001. For 01100 we get 11100. No, because the code word could have been 11100 or 11001. Yes, only the code word 11100 differs in one position from the received word.
001001 is decoded as 001101 by all four methods. 011000 is decoded as 111000 by all four methods. 000110 is decoded as 100110 by all four methods. Since there are no code words whose distance from 100001 is 1 and three whose distance is 2, the nearest-neighbor method will not decode or will arbitrarily choose a code word; parity-check matrix decoding does not decode 100001; the standard-array and syndrome methods decode 100001 as 000000, 110101, or 101011, depending on which of 100001, 010100, or 001010 is a coset leader.
18. Here 2t + s + 1 = 6. For t = 0 and s = 5, we can detect any 5 or fewer errors; for t = 1 and s = 3, we can correct any one error and detect any 2, 3 or 4 errors; for t = 2 and s = 1, we can correct any 1 or 2 errors and detect any 3 errors.
19. For any received word w, there are only eight possibilities for wH. But each of these eight possibilities satisfies condition 2 or the first portion of condition 3′ of the decoding procedure, so decoding assumes that no error was made or one error was made.
20. The last row is obtained by adding 10000 to each code word. So the code words can be obtained by subtracting 10000 from each member of the last row. (Since the code is binary, this is the same as adding 10000 to each member of the last row.)
21. There are 34 code words and 36 possible received words.
22. Yes, because the rows are nonzero and distinct.
23. No; row 3 is twice row 1.
24. Suppose that we can use the nearest-neighbor method to correct any t or fewer errors and the weight of the code is k < 2t + 1. Let u be a code word of weight k. Let u ′ be the vector obtained from u by changing ⌈k/2⌉ ≤ t components of u to 0. If
k is even, we have d(u, u ′) = k = d(0, u ′) so that the nearest neighbor of u ′ is not unique. If k is odd, then d(0, u ′) < d(u, u ′) and u ′ is not decoded as u
Now suppose that the nearest-neighbor method will detect any 2t or fewer errors and that the weight of the code is at most 2t. Let u be a code word whose weight is the weight of the code. Then the error made by changing all the components of u to 0 is not detected.
25. No. For if so, nonzero code words would be all words with weight at least 5. But this set is not closed under addition.
26. Say G = 1 0 a1 a2 a3 0 1 b1 b2 b3 . To detect 3 errors the minimum weight of nonzero code words must be 4. Thus any nonzero code word has at most one zero component. Since (10)G = 10a1a2a3 and (01)G = 01b1b2b3 we have ai /= 0 and bi /= 0 for i = 1, 2, 3. Because (21)G = 2, 1, 2a1 + b1, 2a2 + b2, 2a3 + b3 we must have ai /= bi. Thus the last three columns for G are 1 or 2 But then (11)G = 11000, 1 a contradiction.
27. By Exercise 24, for a linear code to correct every error the minimum weight must be at least 3. Since a (4,2) binary linear code only has three nonzero code words, if each must have weight at least 3, then the only possibilities are (1,1,1,0), (1,1,0,1), (1,0,1,1),(0,1,1,1) and (1,1,1,1). But each pair of these has at least two components that agree. So, the sum of any distinct two of them is a nonzero word of weight at most 2. This contradicts the closure property.
31. By Exercise 14 and the assumption, for each component exactly n/2 of the code words have the entry 1. So, determining the sum of the weights of all code words by summing over the contributions made by each component, we obtain n(n/2). Thus, the average weight of a code word is n/2.
32. Suppose every vector of weight t + 1 is a coset leader. Let v be a code word of weight 2t + 1 and w the vector obtained from v by changing the first t + 1 nonzero component to 0. Then wt(w v) = t + 1 so that w v is a coset leader. But w + C = w v + C and w has weight t. This contradicts the definition of coset leader.
33. Let c, c ′ ∈ C. Then, c + (v + c ′) = v + c + c ′ ∈ v + C and (v + c) + (v + c ′) = c + c ′ ∈ C, so the set C ∪ (v + C) is closed under addition.
34. Let v be any vector. If u is a vector of weight 1, then wt(v) and wt(v + u) have opposite parity. Since any vector u of odd weight is the sum of an odd number of vectors of weight 1, it follows that wt(v) and wt(v + u) have opposite parity. Now, mimic the proof of Exercise 23.
35. If the ith component of both u and v is 0, then so is the ith component of u v and au, where a is a scalar.
CHAPTER 30
Introduction to Galois Theory
1. Note that φ(1) = 1. Thus φ(n) = n. Also, for n = 0, 1 = φ(1) = φ(nn 1) = φ(n)φ(n 1) = nφ(n 1), so that 1/n = φ(n 1). So, by properties of automorphisms, φ(m/n) = φ(mn 1 ) = φ(m)φ(n 1 ) = φ(m)φ(n) 1 = mn 1 = m/n
2. Z2
3. If α and β are automorphisms that fix F , then αβ is an automorphism and, for any x in F , we have (αβ)(x) = α(β(x)) = α(x) = x. Also, α(x) = x implies, by definition of an inverse function, that α 1(x) = x So, by the Two-Step Subgroup Test, the set is a group.
4. Instead, observe that Z2 ⊕ Z2 ⊕ Z2 has 7 subgroups of order 2.
5. Suppose that a and b are fixed by every element of H. By Exercise 29 in Chapter 13, it suffices to show that a b and ab 1 are fixed by every element of H By properties of automorphisms we have for any element φ of H, φ(a b) = φ(a) + φ( b) = φ(a) φ(b) = a b. Also, φ(ab 1) = φ(a)φ(b 1) = φ(a)φ(b) 1 = ab 1 .
6. By √ E x e r c is e 4 of Chapter 19, the splitting field is Q( √ 2, i) Since [Q( 2, i) : Q] = 4, | √Gal(E/Q)| = 4. It follows that Gal(E/Q) = {ǫ, α, β , √α β } where β has fixed field Q( 2), α has fixed field Q(i), and αβ has fixed field Q( 2).
7. It suffices to show that each member of Gal(K/F ) defines a permutation on the ai’s. Let α ∈ Gal(K/F ) and write f (x) = cnxn + cn 1xn 1 + · · · + c0. Then 0 = f (ai) = cnan + cn 1an 1 + · · · + c0 . So, 0 = α(0) = α(cn)(α(ai)n + α(cn 1)α(ai)n 1 + + α(c0) = cn(α(ai)n + cn 1α(ai)n 1 + · · · + c0 = f (α(ai)). So, α(ai) = aj for some j, and therefore α permutes the ai’s.
i i
8. Use Corollary 3 of Theorem 16.2 and Exercise 7 of this chapter.
11. a. Z20 ⊕ Z2 has three subgroups of order 10. b. 25 does not divide 40 so there is none. c. Z20 ⊕ Z2 has one subgroup of order 5.
12. See Example 4 in this chapter.
13. The splitting fi√ e l d over R is√R( √ 3). The Galois group is the identity and the mapping a + b 3 → a b 3.
14. Observe that D6 has exactly three subgroups of order 6.
15. Use Theorem 21.3.
16. Use the Corollary to Theorem 23.2 and Theorem 11.1.
17. If there were a subfield K of E such that [K : F ] = 2 then, by the Fundamental Theorem of Galois Theory (Theorem 30.1), A4 would have a subgroup of index 2. But by Example 5 in Chapter 7, A4 has no such subgroup.
18. Let ω = ( 1 + i √ 3)/2. The splitting field of x3 1 over Q is Q(ω). Since [Q(ω) : Q] = 2, the Galois group of x3 1 over Q is Z2 The splitting field of x3 2 over Q is Q( 3 2, ω). Since [Q( 3 2, ω) : Q] = 6, the Galois group has order 6 and is generated by α and β where α( 3 2) = ω 3 2, α(ω) = ω and β( 3 2) = 3 2, β(ω) = ω2 . Since αβ = βα, the group must be S3 (see Theorem 7.2 and the remark at the end of the proof).
19. This follows directly from the Fundamental Theorem of Galois Theory (Theorem 30.1) and Sylow’s First Theorem (Theorem 23.3).
20. Use the subgroup lattice for D5
21. Let ω be a primitive cube root of 1. Then Q Q ( √ 3 2) Q(ω, √ 3 2) and Q ( √ 3 2) is not the splitting field of a polynomial in Q[x].
22. Use the Fundamental Theorem and the fact that Gal(E/F ) is finite.
23. By the Fundamental Theorem of Finite Abelian Groups (Theorem 11.1), the only Abelian group of order 10 is Z10. By the Fundamental Theorem of Cyclic Groups (Theorem 4.3), the only proper, nontrivial subgroups of Z10 are one of index 2 and one of index 5. So, the lattice of subgroups of Z10 is a diamond with Z10 at the top, 0 at the bottom, and the subgroups of indexes 2 and 5 in the middle layer. Then, by the Fundamental Theorem of Galois Theory, the lattice of subfields between E and F is a diamond with subfields of indexes 2 and 5 in the middle layer.
24. Q(ω + ω4)
25. By Example 7, the group is Z6.
26. Z3
27. This follows directly from Exercise 21 in Chapter 24.
28. Let K be the subgroup of rotations in Dn. The desired series is {R0} ⊂ K ⊂ Dn.
29. This follows directly from Exercise 43 in Chapter 23.
30. Note that A4 has a normal Sylow 2-subgroup.
31. This follows directly from Exercise 50 in Chapter 10.
32. Let {e} = H0 ⊂ H1 ⊂ · · · ⊂ Hn = G be the series that shows that G is solvable. Then H0 ∩ H ⊂ H1 ∩ H ⊂ · · ⊂ Hn ∩ H = H shows that H is solvable.
33. Since K/N ⊳ G/N , for any x G and k K, there is a k′ K such that k′N = (xN )(kN )(xN ) 1 = xNkNx 1 N = xkx 1 N So, xkx 1 = k′ n for some n ∈ N . And since N ⊆ K, we have k′ n ∈ K.
34. Use parts 7, 6 and 1 of Theorem 15.1.
35. Since G is solvable there is a series {e} = K0 ⊂ K1 ⊂ · · · ⊂ Km = G
such that K i+1/K i is Abelian. Now there is a series
= K i
=
i Ki where |(Lj+1/Ki)/(Lj/Ki)| is prime. Then
, Ki Ki Ki
= Ki+1 and each Lj+1/Lj is prime (see Exercise 42 of Chapter 10). We may repeat this process for each i.
36. Mimic the analysis carried out for 3x5 15x + 5 at the end of Chapter 30.
CHAPTER 31
Cyclotomic Extensions
1. Since ω = cos π + i sin π = cos 2π + i sin 2π , ω is a zero of 3 3 6 6 x6 1 = Φ1(x)Φ2(x)Φ3(x)Φ6(x) = (x 1)(x + 1))(x2 + x + 1)(x2 x + 1), it follows that the minimal polynomial for ω over Q is x2 x + 1.
2. Use Theorem 31.1
3. Over Z, x8 1 = (x 1)(x + 1)(x2 + 1)(x4 + 1). Over Z2 , x2 + 1 = (x + 1)2 and x4 + 1 = (x + 1)4 . So, over Z2, x8 1 = (x + 1)8 . Over Z3, x2 + 1 is irreducible, but x4 + 1 factors into irreducibles as (x2 + x + 2)(x2 x 1). So, x8 1 = (x 1)(x + 1)(x2 + 1)(x2 + x + 2)(x2 x 1). Over Z5, x2 + 1 = (x 2)(x + 2), x4 + 1 = (x2 + 2)(x2 2), and these last two factors are irreducible. So, x8 1 = (x 1)(x + 1)(x 2)(x + 2)(x2 + 2)(x2 2).
4. Use xn 1 = (x 1)(xn 1 + xn 2 + + x + 1) and that fact that nth roots of unity form a cyclic group of order n
5. Let ω be a primitive nth root of unity. We must prove ωω2 ωn = ( 1)n+1 . Observe that ωω2 ωn = ωn(n+1)/2 When n is odd, ωn (n +1)/ 2 = (ωn )(n +1)/ 2 = 1(n +1)/ 2 = 1. When n is even, (ωn/2)n+1 = ( 1)n+1 = 1.
6. Φ3(x).
7. If [F : Q] = n and F has infinitely many roots of unity, then there is no finite bound on their multiplicative orders. Let ω be a primitive mth root of unity in F such that φ(m) > n. Then [Q(ω) : Q] = φ(m). But F ⊇ Q(ω) ⊇ Q implies [Q(ω) : Q] ≤ n.
8. Observe that xn 1 = (x 1)(xn 1 + xn 2 + + x + 1) and use Theorem 31.1.
9. Let 2n + 1 = q Then 2 ∈ U (q) and 2n = q 1 = 1 in U (q) implies that |2| = 2n So, by Lagrange’s Theorem, 2n divides |U (q)| = q 1 = 2n .
10. We know Φ2(0) = 1. Now observe that and use induction.
xn 1 = (x 1)
d|n 1<d<n Φd(x)Φn(x)
11. Let ω be a primitive nth root of unity. Then 2nth roots of unity are ±1, ±ω, , ±ωn 1 These are distinct, since 1 = ( ωi)n , whereas 1 = (ωi)n
12. Let α be a primitive mnth root of unity. Then αn is a primitive mth root of unity and αm is a primitive nth root of unity. This shows that (xm 1)(xn 1) splits in the splitting field of xmn 1. Conversely, let β be a primitive mth root of unity and γ be a primitive nth root of unity. It suffices to show that βγ = mn. Let H = βγ . Since (βγ)mn = (βm )n (γn )m = 1 1 we know H mn Since (βγ)m = βm γn = γn and, by Theorem 4.2, γn = γ , we know that n divides H . By symmetry, m divides H . Thus H mn. This proves that the splitting field of (xm 1)(xn 1) contains a primitive mnth root of unity.
13. First observe that deg Φ2n(x) = φ(2n) = φ(n) and deg Φn( x) = deg Φn(x) = φ(n). Thus, it suffices to show that every zero of Φn( x) is a zero of Φ2n(x). But ω is a zero of Φn( x) means that ω = n, which in turn implies that ω = 2n. (Here ω means the order of the group element ω.)
14. Since the two sides are monic and have the same degree it suffices to prove that k 1 every zero of Φp k (x) is a zero of Φ (xp ). Let ω be a zero of Φ (x) and note that ω = pk implies that ωp k 1 = p. Φ8 (x) = x4 + 1, Φ27 (x) = (x9 )2 + x9 + 1 = x18 + x9 + 1.
15. Let G = Gal(Q(ω)/Q) and H1 be the subgroup of G of order 2 that fixes cos( 2π ). Then, by induction, G/H1 has a series of subgroups H1/H1 ⊂ H2/H1 ⊂ ⊂ Ht/H1 = G/H1, so that |Hi+1/H1 : Hi/H1| = 2. Now observe that |Hi+1/H1 : Hi/H1| = |Hi+1/Hi|.
16. Use Theorem 31.4.
17. Instead, we prove that Φn(x)Φpn(x) = Φn(xp). Since both sides are monic and have degree pφ(n), it suffices to show that every zero of Φn(x)Φpn(x) is a zero of Φn(xp). If ω is a zero of Φn(x), then |ω| = n. By Theorem 4.2, |ωp| = n also. Thus ω is a zero of Φn(xp). If ω is a zero of Φnp(x), then |ω| = np and therefore |ωp| = n
18. Use Theorem 31.4.
19. Let ω be a primitive 5th root of unity. Then the splitting field for x5 1 over Q is Q(ω). By Theorem 31.4, Gal(Q(ω)/Q) ≈ U (5) ≈ Z4 Since ⟨2⟩ is the unique subgroup strictly between 0 and Z4, we know by Theorem 32.1 that there is a unique subfield strictly between Q and E
}
20. Use Theorem 31.4 and Theorem 30.1.
21. Suppose that a prime p = 2m + 1 and m is not a power of 2. Then m = st where s is an odd integer greater than 1 (the case where m = 1 is trivial). Let n = 2t + 1. Then 1 < n < p and 2t mod n = 1. Now looking at p mod n and replacing 2t with 1, we have (2t)s + 1 = ( 1)s + 1 = 0. This means that n divides the prime p, which is a contradiction.
22. The three automorphisms that take ω → ω4 , ω → ω 1 , ω → ω 4 have order 2.
