Solutions to Assignment 6 #2 a) Let g ∈ G. Since HN = G, there is at least one h ∈ H and n ∈ N such that g = hn. We need to show uniqueness. If hn = h0 n0 for h, h0 ∈ H and n, n0 ∈ N , then hn = h0 n0 (h0 )−1 hn = n0 (h0 )−1 h = n0 n−1 . The left side of the equation is in H and the right side is in N , so (h0 )−1 h = n0 n−1 ∈ H ∩N = {e}. Therefore, h = h0 and n = n0 , as desired. b) Let h ∈ H and n ∈ N . Let g = hnh−1 n−1 . Since H is normal, nhn−1 ∈ H, so g = h(nh−1 n−1 ) is a product of elements of H, so it is an element of H. Similarly, hnh−1 ∈ N , so g = (hnh−1 )n−1 ∈ N . Therefore, g ∈ H ∩ N , so g = e, and hnh−1 n−1 = e hn = nh. #3 Let a generate Z3 and let b generate Z4 . We write the operations on Z3 and Z4 multiplicatively and denote the identities in both groups by e. Let g = (a, b) ∈ Z3 × Z4 , so that g n = (an , bn ). If g n = (e, e), then an = e, so n is a multiple of 3, and bn = e, so n is a multiple of 4. The smallest positive n like this is 12, and g 12 = (e, e), so o(g) = 12. Since Z3 × Z4 only has 12 elements, Z3 × Z4 = hgi, which means that it is isomorphic to the cyclic group of order 12 by the map f : Z12 → Z3 × Z4 , f (k) = g k . On the other hand, Z9 is not isomorphic to Z3 ×Z3 . Suppose that f : Z9 → Z3 ×Z3 is a homomorphism and that f (1) = (am , an ). Then f (3) = (am , an )3 = (a3m , b3n ) = (e, e). That means that 3 ∈ ker(f ), so f is not one-to-one, and f can’t be an isomorphism. #4 If f : Z/6Z → Z/9Z is a homomorphism, then it is determined by f (6Z+1) – that is, if f (6Z+1) = 9Z+a, then f (6Z + n) = 9Z + na. First, we note that the map f (6Z + n) = 9Z + na is well-defined when a is divisible by 3. That is, if 6Z + n = 6Z + n0 , then n − n0 is a multiple of 6, so an − an0 is a multiple of 9 and 9Z + na = 9Z + n0 a. If it’s well-defined, then it’s a homomorphism: we have f (6Z + n + 6Z + n0 ) = f (6Z + n + n0 ) = 9Z + an + an0 f (6Z + n) + f (6Z + n0 ) = (9Z + an) + (9Z + an0 ) = 9Z + an + an0 There are three multiples of 3 between 0 and 8, so there are three homomorphisms: f0 (6Z + n) = 9Z + 0n = 9Z f3 (6Z + n) = 9Z + 3n f6 (6Z + n) = 9Z + 6n (f9 , f12 , etc. are also homomorphisms, but f9 = f0 , f12 = f3 , etc.) These are related to multiplication because fa (n) ≡ an (mod 9). (Other answers are possible). The map f (6Z + n) = 9Z + 2n isn’t a homomorphism because it’s not well-defined — for instance, 6Z + 2 = 6Z + 8, but 9Z + 2 · 2 6= 9Z + 2 · 8. 1 #5 a) Suppose that f (N1 ) ⊂ N2 and suppose that N1 g = N1 g 0 . To show that f¯ is well-defined, we have to show that N2 f (g) = N2 f (g 0 ). But g 0 g −1 ∈ N1 , so f (g 0 g −1 ) = f (g 0 )f (g −1 ) ∈ N2 , which implies that N2 f (g) = N2 f (g 0 ), so f¯ is well-defined. On the other hand, suppose that f¯ is well-defined. Then f¯(N1 ) = f¯(N1 e) = N2 f (e) = N2 . When g ∈ N1 , then on one hand we have f¯(N1 g) = N2 f (g). On the other hand, N1 g = N1 , so f¯(N1 g) = N2 , so N2 = N2 f (g) and f (g) ∈ N2 . This implies f (N1 ) ⊂ N2 . So f¯ is well-defined if and only if f (N1 ) ⊂ N2 . b) By the previous part, ft induces a homomorphism from R/Z to R/Z if and only if ft (Z) ⊂ Z. This happens exactly when t ∈ Z, and the resulting map is f¯t (Z + x) = Z + tx. We have Z + tx = Z if and only if tx ∈ Z, so ker(f¯t ) = {Z + k/t | k ∈ Z}. The graph of f¯2 (with a few points labeled) is: (Z + .4, Z + .8) Z + .8 Z + .6 (Z + .75, Z + .5) (Z + .2, Z + .4) Z + .4 Z + .2 Z Z Z + .2 Z + .4 Z + .6 Z + .8 c) Suppose that f is surjective and let h ∈ G2 . We claim that there is some g ∈ G1 such that f¯(N1 g) = N2 h. Let g ∈ f −1 (h); since f is surjective, this inverse image is nonempty. By the definition of f¯, we have f¯(N1 g) = N2 h. On the other hand, suppose that f −1 (N2 ) = N1 . We claim that if f¯(N1 g) = f¯(N1 g 0 ), then N1 g = N1 g 0 . We have: f¯(N1 g) = f¯(N1 g 0 ) N2 f (g) = N2 f (g 0 ) f (g)f (g 0 )−1 ∈ N2 f (g(g 0 )−1 ) ∈ N2 g(g 0 )−1 ∈ f −1 (N2 ) = N1 N1 g = N1 g 0 . as desired. 2 d) Let π : G → G/K be the canonical homomorphism. Then N ⊂ G is a normal subgroup and N/K ⊂ G/K is a normal subgroup (by the Correspondence Theorem). So we apply part (a) with G1 = G, N1 = N , G2 = G/K, N2 = N/K. Since π(N ) = N/K, π induces a map π̄ : G/N → (G/K)/(N/K). By part (c), this map is injective and surjective, so it is an isomorphism. For example, suppose that G = Z, N = 2Z, and K = 6Z. Then N / G, K / G, and K ⊂ N . We have G/N = Z/2Z ∼ = Z2 = {2Z, 2Z + 1} G/K = Z/6Z ∼ = Z6 = {6Z, 6Z + 1, . . . , 6Z + 5} ∼ Z3 = {6Z, 6Z + 2, 6Z + 4}. N/K = 2Z/6Z = The quotient (G/K)/(N/K) consists of two cosets, N/K and N/K + 1. The map π̄ : G/N → (G/K)/(N/K) sends 2Z to N/K = {6Z, 6Z + 2, 6Z + 4} and sends 2Z + 1 to N/K + 1 = {6Z + 1, 6Z + 3, 6Z + 5}. p. 70, #6 Note that for every x ∈ G, x2 = e and that the product of any two non-identity elements is the third non-identity element. These two properties and the fact that e is the identity completely determine the multiplication table of G. So any map that sends e to e and permutes a, b, and ab is an automorphism of G. There are six such maps, corresponding to the six permutations of {a, b, ab}. (We haven’t discussed this in class yet, but the automorphism group is S3 , the symmetric group on three elements.) p. 70, #10 G a finite group, T an automorphism such that T (x) = x iff x = e. Show that every g ∈ G can be written as x−1 T (x). Consider the map (not necessarily a homomorphism) f : G → G given by f (x) = x−1 T (x). Since G is a finite group, f is a bijection if and only if it is one-to-one if and only if it is onto, so it is enough to show that f is one-to-one. Suppose that x, y ∈ G and f (x) = f (y). Then x−1 T (x) = y −1 T (y) yx−1 = T (y)T (x)−1 yx−1 = T (yx−1 ). By our assumption on T , this implies that yx−1 = e, so y = x, and the map is one-to-one. 3