Math 103 HW 9 Solutions to Selected Problems

4. Show that U (8) is not isomorphic to U (10).

Solution: Unfortunately, the two groups have the same order: the elements are U (n)
are just the coprime elements of Zn , so U (8) = {1, 3, 5, 7} while U (10) = {1.3.7.9}.
Thus, we must examine the elements further. We claim that U (10) is cyclic. This is
easy to calculate:

32 ≡ 9
33 = 27
≡7
4
3 ≡3·7
≡ 1 (mod 10)

which means 3 generates U (10).

Now if U (10) and U (8) were isomorphic, we have seen that this would mean U (8) was
cyclic as well. In particular, it would have a generator of order 4. However, we can see
that

32 = 9
≡1
2
5 = 25
≡1
2
7 = 49
≡ 1 (mod 8)

so every element of U (8) has order dividing 2. Therefore, U (8) is not cyclic, hence is not
isomorphic to U (10).

12. Let G be a group. Prove that the mapping α(g) = g −1 for all g in G is an
automorphism if and only if G is Abelian.

1
 Solution: α is clearly its own inverse, so it is always a bijective map. The only question
is whether it is a morphism of groups, so it is enough to show this is true if and only if
G is Abelian. If G is Abelian, then certainly

α(gh) = (gh)−1
= h−1 g −1
= g −1 h−1
= α(g)α(h)

since we can commute elements, so α is a morphism. On the other hand, by definition

α being a morphism is equivalent to (gh)−1 = g −1 h−1 for every g, h ∈ G. By problem
25 from Homework 4, this implies that G is Abelian. Putting the two together, we have
our result.

17. If G is a group, prove that Aut(G) and Inn(G) are groups.

Solution: We first show that each has an identity. The operation is function composi-
tion, so the identity here is just the identity function idG on G. This is certainly bijective,
and it is a morphism simply because G is a group. This shows idG ∈ Aut(G). Function
composition is also associative (see Theorem 0.8.1), we know that the composition of
bijective functions is bijective, and it easy to check that the composition of morphisms is
again a morphism 1 . Thus, Aut(G) is closed under multiplication. It remains to show it
is closed under inversion. We know at least that the function α−1 exists for α ∈ Aut(G)
(since bijective is equivalent to invertible). If we let g, h be in G, then

α−1 (gh) = α−1 (α(α−1 (g))α(α−1 (h))

= α−1 (α(α−1 (g)α−1 (h))) (since α is a morphism)
= α−1 (g)α−1 (h)

meaning α−1 is actually in Aut(G). This shows that Aut(G) is a group.

Inn(G) is defined as a subset of Aut(G), so we need not show associativity again. For
any g ∈ G, ege = g, so idG = φe , which is certainly an element of Inn(G). Furthermore,

φg φh (x) = φg (hxh−1 )
= ghxh−1 g −1
= φgh (x)

for each x ∈ G, so φg φh = φgh is in Inn(G). In particular, this show thats φg φg−1 =

φg−1 φg = φe , the identity, so (φg )−1 = φg−1 . Thus Inn(G) is closed under multiplication
and taking inverses, and contains the identity, so it is indeed a subgroup of Aut(G).

1
that is, if α and β are two morphisms from G to G, α(β(gh)) = α(β(gh)) = α(β(g)β(h)) =
α(β(g))α(β(h))

Page 2
24. Let φ be an automorphism of a group G. Prove that H = {x ∈ G|φ(x) = x} is
a subgroup of G.

Solution: For any morphism G → G, φ(e) = e, meaning e ∈ H. Since φ is a morphism,

if x, y ∈ H, φ(xy) = φ(x)φ(y) = xy, so xy ∈ H as well. We also know that φ(g)−1 =
φ(g −1 ) for all g ∈ G, so x ∈ H implies φ(x−1 ) = φ(x)−1 = x−1 ; ie, x−1 ∈ H. Thus, H is
a subgroup.

26. Suppose that φ : Z20 → Z20 is an automorphism and φ(5) = 5. What are the
possiblities for φ(x)?

Solution: Note that since Z20 is cyclic, generated by 1, φ is completely determined by

φ(1): φ(x) = φ(x · 1) = x · φ(1) since φ is a morphism. This shows that the morphisms
from Z20 to itself are precisely given by φm (x) = mx for m ∈ Z20 (this is a morphism
because φ(x + y) = m(x + y) = mx + my). To be an automorphism, it is enough for
φm (1) = m to generate Z20 , since for finite sets, surjective implies bijective. This means
that m must be coprime to 20. Let our φ be one of these φm . The only other constraint
we have is that φ(5) = 5 in Z20 ; that is, 5m ≡ 5 (mod 20). But we know this is true if
and only if 20 divides 5m − 5 = 5(m − 1), or in other words 4 divides m − 1. Checking
×
all the members of Z20 , we see that the only m satisfying this condition are m = 1, 9, 13
and 17, so these are the only possibilities for φ(x) = mx.
  
1 a 
30. The group a ∈ Z is isomorphic to what familiar group? What if Z
0 1 
is replaced by R?

Solution: Let G be this group (implicit here is that the operation is matrix multiplica-
tion). We claim that G is isomorphic
 to Z. To this end, we try to use the easiest map
1 a
φ : G → Z possible, given by φ( ) = a. This is a morphism because
0 1
     
1 a 1 b 1 a+b
φ( ) ) = φ( )
0 1 0 1 0 1
=a+b
   
1 a 1 b
= φ( ) + φ( )
0 1 0 1
 
1 a
On the other hand, we can see that φ is invertible: if we let ψ : Z → G, a 7→ ,
    0 1
1 a 1 a
then certainly ψ ◦ φ( )= and φ ◦ ψ(a) = a for all a ∈ Z, so ψ = φ−1 .
0 1 0 1
Thus, φ is an isomorphism. Nothing about our proof relied on any properties of Z,
besides that it had an additive structure, so, , replacing Z with R everywhere, it would
work for R as well.

Page 3
38. Let √
G = {a + b 2| a, b are rational}
and   
a 2b 
H= a, b are rational .
b a 
Show that G and H are isomorphic under addition. Prove that G and H are
closed under multiplication. Does your isomorphism preserve multiplication
as well as addition?

√
 
a 2b
Solution: Define φ : H → G by φ( ) = a + b 2 (which is in G since a, b ∈
b a
Q). This is definitely surjective, so we must show it is an injective morphism. Given
a, b, c, d ∈ Q,
     
a 2b c 2d a + c 2(b + d)
φ( + ) = φ(
b a d c b+d a+c
√
= (a + c) + (b + d) 2
√ √
=a+b 2+c+d 2
   
a 2b c 2d
= φ( ) + φ( )
b a d c

as desired, hence φ is a morphism. As we  provedin section 2 , φ being injective is equiv-

0 0
alent to saying that φ(h) = 0 implies h = (the identity in H) for any h ∈ H. In
0 0
√

a 2b
other words, we must show that (letting h = ) if a, b ∈ Q, a + b 2 = 0 implies
b a
√ √
a = b = 0. Suppose√ not; then a = −b 2,
√ so b must not be 0 or else a = 0 2 = 0.
a
But then − b = 2, meaning we can write 2 as the quotient of two rational numbers.
√
This forces 2 itself to be rational, as Q is a field (so closed under division by nonzero√
elements). However, it is well known (for example, often proved in Math 109) that 2
is irrational, so we must have a = b = 0 after all.

G is closed under multiplication, as

√ √ √
(a + b 2)(c + d 2) = ac + 2bd + (ad + bc) 2

which is in G since the rationals are closed under multiplication and addition. What’s
more,     
a 2b c 2d ac + 2bd 2(bc + ad)
=
b a d c (bc + ad) ac + 2bd
which is in H for the same reasons. This shows that H is closed under multiplication,
and also that φ preserves multiplication.
2
and is useful to prove yourself if you didn’t go to section

Page 4
44. Suppose that G is a finite Abelian group and G has no element of order 2.
Show that the mapping g 7→ g 2 is an automorphism of G. Show, by example,
that there is an infinite Abelian group for which the mapping g 7→ g 2 is
one-to-one and operation preserving but not an automorphism.

Solution: Call this map α. Since G is Abelian, α(gh) = ghgh = g 2 h2 ∀g, h ∈ G, hence
α is a morphism. Suppose α(g) = e. Then either g = e or g has order 2, so by assump-
tion we must have g = e. By the fact mentioned in the previous problem, this is enough
to show α is injective. As G is finite, injective implies bijective, so α is an automorphism.

Now consider the infinite Abelian group Z. In additive notation, α : Z → Z is de-

fined by α(x) = 2x. We know that every element of Z has infinite order except the
identity, so the proof above still works to show that α is an injective morphism (or we
can just divide the equation 2x = 2y by 2). However, injective does not imply bijective
in this case: the image of α is the even integers, which definitely isn’t all of Z. Therefore,
α is not surjective, and hence cannot be an automorphism.

55. Let φ be an automorphism of C∗ , the group of nonzero complex numbers

under multiplication. Determine φ(−1). Determine the possibilities for φ(i).

Solution: We have seen that isomorphisms preserve orders, and so (−1)2 = 1 implies
that φ(−1) has order 2. What are the elements of order 2 in C∗ ? Such an element—call
it x—must be a solution to x2 − 1 = 0, which factorizes as (x − 1)(x + 1) = 0. We cannot
have x = 1 since 1 has order 1, so we can divide by (the nonzero) x − 1 to get x + 1 = 0;
ie, x = −1. Thus, the only option is φ(−1) = −1.

Similarly, i2 = −1, i3 = −i, and i4 = 1, so φ(i) must have the same order, 4, as i.
The elements of order 4 are solutions in C∗ to x4 − 1 = 0, which we can factorize as

0 = x4 − 1
= (x2 − 1)(x2 + 1)
= (x − 1)(x + 1)(x − i)(x + i)

By the same reasoning as above, this means that the elements of order 4 must be ±1 or
±i. The former do not have order 4, and −i has order 4 too (for example, because it’s
i−1 ), so φ(i) must be ±i.

64. Prove that Q, the group of rational numbers under addition, is not isomorphic
to a proper subgroup of itself.

Solution: Let φ : Q → H be an isomorphism with a subgroup H of Q. We want to

show that H = Q. However, if x, y are ∈ Z, φ(x) = x · φ(1), while φ( xy ) (if y 6= 0)
is equal to xφ( y1 ), which—since φ(1) = φ(y · y1 ) = yφ( y1 )—must be xy φ(1). This shows
that φ is completely determined by φ(1), and in fact that φ is just multiplication by φ(1).

Page 5
φ is an isomorphism, so in particular it must be injective, with φ(1) 6= 0. But then
g
for any g ∈ Q, φ( φ(1) ) = g, so the image of φ is Q. Since im(φ) is contained in H almost
by definition, this forces H = Q, so H cannot be a proper subgroup.

Page 6