Algebra Qual

Algebra Qualifying Exam
John Dusel
To MBM: we suer together.
Thanks to Paul Oesser, Nathan Manning (Algebra Qualmasters for
2010,2009), Adam Katz, and Mei-Chu Chang.
Contents
2011 Qual 13
Part A 13
2011 A1 13
2011 A2 13
2011 A3 14
2011 A5 14
2011 A6 15
2011 A7 15
2011 A8 15
Part B 16
Part C 16
2010 Qual 17
Part A 17
2010 A1 17
2010 A2 18
2010 A3 19
2010 A4 21
Part B 22
2010 B1 22
2010 B3 22
2009 Qual 25
Part A 25
2009 A1 25
2009 A2 unnished 26
2009 A3 26
2009 A4 27
2009 A5 unnished 28
2009 A6 28
2009 A7 unnished 29
2009 A8 29
Part B 30
2009 B1 30
3
4 CONTENTS
2009 B2 32
2009 B3 33
2009 B4 34
2009 B5 34
Part C 37
2009 C1 37
2009 C3 37
2009 C5 38
2008 Qual 41
Part A 41
2008 A1 41
2008 A2 42
2008 A3 43
2008 A4 46
2008 A5 48
Part B 50
2008 B1 50
2008 B2 50
2008 B3 52
2008 B4 54
2008 B5 54
Part C 55
2008 C1 55
2008 C2 55
2008 C3 58
2008 C4 58
2007 Qual 59
Part A 59
2007 A1 59
2007 A2 59
2007 A3 60
2007 A4 60
Part B 60
2007 B1 60
An example where Hom(, M) is not exact 61
2007 B2 62
2007 B3 unnished 63
2007 B4 63
2007 B5 63
Part C 65
CONTENTS 5
2007 C2 65
2007 C3 65
2007 C4 67
2007 C5 67
2007 C6 68
2006 Qual 69
Part A 69
2006 A1 unnished 69
2006 A2 69
2006 A3 (maybe incorrect) 69
2006 A4 70
2006 A5 71
Part B 71
2006 B1 71
2006 B2 72
2006 B3 74
2006 B4 74
2006 B5 75
Part C 76
2006 C1 76
2006 C2 78
2005 Qual 81
Part A 81
2005 A1 unnished 81
2005 A2 81
2005 A3 82
2005 A4 83
2005 A5 83
Part B 84
2005 B1 84
2005 B2 unnished 85
2005 B3 85
2005 B4 85
Part C 86
2005 C1 86
2005 C2 87
2005 C3 88
2005 C4 89
2004 Qual 91
Part A 91
6 CONTENTS
2004 A1 unnished 91
2004 A2 91
2004 A3 91
Part B 92
2004 B1 92
2004 B2 93
2004 B3 93
2004 B4 93
2004 B5 94
2004 B6 95
Part C 96
2004 C1 96
2004 C2 96
2004 C4 97
2004 C6 98
2004 Qual, version 2 (October) 101
Part A 101
2004v2 A1 unnished 101
2004v2 A2 101
2004v2 A3 102
2004v2 A4 102
2004v2 A5 102
Part B 103
Part C 104
2004 C1 104
2004v2 C3 105
2004v2 C4 106
2003 Qual 109
Part A 109
2003 A1 109
2003 A2 unnished 109
2003 A3 109
2003 A4 110
2003 A5 110
2003 A6 111
Part B 111
2003 B1 unnished 111
2003 B3 111
CONTENTS 7
2003 B5 112
2003 B6 112
Part C 113
2003 C2 113
2003 C3 114
2002 Qual 117
Part A 117
2002 A1 117
2002 A2 117
2002 A3 117
2002 A4 119
2002 A6 119
Part B 119
2002 B4 119
2002 B5 120
2002 B6 121
Part C 121
2002 C1 121
2002 C3 122
2002 C5 123
2001 Qual 125
Part A 125
2001 A1 125
2001 A2 125
2001 A3 125
2001 A4 126
2001 A5 126
Part B 126
2001 B1 126
2001 B2 127
2001 B5 127
Part C 127
2001 C2 127
2001 C3 128
2001 C5 129
2001 C6 129
2000 Qual 133
Part A 133
8 CONTENTS
2000 A4 133
Part B 133
2000 B1 133
2000 B2 133
2000 B3 134
2000 B4 134
2000 B5 134
Why is not a free Z-module 135
2000 B6 135
2000 B8 136
1999 Qual 137
Parts A and B 137
1999 A3 137
1999 AB5 137
1999 AB7 137
1999 AB8 138
1999 AB10 139
Part C 139
1999 C3 139
1999 C4 140
1999 C6 141
1998 Qual 143
Part A 143
1998 A2 143
1998 A4 143
1997 Qual 145
Part B 145
1997 B2 145
1997 B3 146
1997 B4 146
Part C 146
1997 C1 146
1997 C2 147
1996 Qual 149
Part B 149
CONTENTS 9
1996 B1 149
1996 B2 149
1996 B3 150
Part C 151
1996 C1 151
1996 C3 151
1996 C4 152
1996 C5 152
1995 Qual 155
Part A 155
1995 A1 155
1995 A2 155
1995 A3 155
1995 A4 156
1995 A5 156
Part B 156
1995 B1 156
1995 B4 158
Part C 158
1995 C3 (incomplete) 158
1994 Qual 161
Part A 161
1994 A1 161
1994 A2 improvement needed 161
1994 A3 161
Part B 162
1994 B2 162
1994 B4 163
1994 Version 2 Qual 165
Part A 165
1994v2 A4 165
Part B 165
1994v2 B3 165
1994v2 B4 166
1994v2 B5 166
10 CONTENTS
1993 Qual 169
Part A 169
1993 A6 169
Part B 169
1993 B2 169
1993 B3 170
1993 B4 171
1993 B7 171
Part C 172
1993 C1 172
1993 C3 173
1993 C5 173
Qual Seminar Homework 175
06/12/10 175
Classify all groups of order 45 175
Exam Questions from Academic Year 20092010 177
Part B 177
201B Midterm 177
201B Midterm Question 1 177
201B Final Exam 180
Final Exam Question 1 180
Part C 188
Assignment 1 188
III.3 #3 188
III.3 #4 190
III.3 #6 190
III.3 #7 192
III.6 #1 192
III.6 #2 193
III.6 #3 193
CONTENTS 11
III.6 #10 194
Additional Exercise 1 195
Exercise 2 195
Exercise 3 196
Assignment 2 196
VI.1 #1 196
V.1 #3 197
V.1 #7 197
V.1 #14a 199
V.1 #17 199
V.1 #19 200
Exercise 2 201
Exercise 3 202
Exercise 4 203
Exercise 5 203
Assignment 3 203
V.2 #3 203
V.2 #4 204
Final Exam 208
201 C Final Exam Problem 1 208
Bibliography 217
12 CONTENTS
Any mistakes in this document are due to JMD. Comments and
corrections should be sent to the e-mail address at the end of this
document.
2011 Qual
Part A
2011 A1.
(1) Suppose that G is a group and that H
1
,= H
2
are subgroups
of G. State and prove a sucient condition for H
1
H
2
to be
a group.
(2) Give an example of a group G and two subgroups K
1
, K
2
such
that K
1
K
2
is strictly contained in the subgroup generated
by K
1
, K
2
.
Answer (1). This is a special case of I.2 exercise 18. We want a
condition (*) under which H
1
H
2
= H
1
H
2
= H
1
H
2
(join). The
denitions are
H
1
H
2
= h : h H
1
or h H
2
,
H
1
H
2
= h
n
1
1
h
nt
t
: h
i
H
1
H
2
, n
i
Z.
To get () H
1
H
2
= H
1
H
2
explore the contrapositive
_
H
1
H
2
,= H
1
H
2
().
Certainly H
1
H
2
H
1
H
2
, but the converse need not hold if there
exists h
n
1
1
h
nt
t
/ H
1
H
2
. In other words, the converse fails if H
1
H
2
is not closed under multiplication.
Condition. If H
1
H
2
is closed under multiplication then H
1
H
2
=
H
1
H
2
.
The proof is immediate.
Answer (2). Use G = Z
6
, K
1
= 2, K
2
= 3. In this case K
1

K
2
= 0, 2, 3, 4 and I.2.17 gives K
1
K
2
= K
1
+ K
2
= a + b : a
K
1
, b K
2
= Z
6
.
2011 A2. Suppose that G is a group and H, K are subgroups of
G such that G is the internal direct product of H and K. Prove that
H, K G and that G/H

= K.
Deduce that there is no subgroup H of S
5
such that S
5
is isomorphic
to the direct product of H and A
5
.
13
14 2011 QUAL
Proof. From the denition I.8.8 of internal direct product we have
(1) H, K G; (2) G = KH = kh : k K, h H; (3) K H = e.
Dene a function K G/H by k

kH. (1) implies is a map of
groups, (2) inplies is surjective, and (3) implies is injective.
(Sketchy) Suppose H S
5
and S
5

= H A
5
. The fact implies
H S
5
and [H[ = [S
5
: A
5
] = 2. This means H = (ij) for some
distinct i, j 1, . . . , 5. It is well-known that (in particular) for all
S
5
we have (ij)
1
= ((i)(j)), and for all such choices of i, j
there exists a with ((i)(j)) ,= (ij). There is no way H can be
normal, so there can be no such H.
2011 A3. Prove that a free abelian group is a free group if and
only if it is cyclic. Give an example to show that a cyclic group may
not be free.
Proof. See 2003 A4 and p.65 of Hungerford.
Let A be a free abelian group.
If A is cyclic then A = x for some element x A. Each element
of A has the form x
n
for some x Z. Dene A Z by x
n
Z and
verify that it is an isomorphism of groups (Z is the free group on one
generator).
For the converse we contradict the contrapositive: suppose that A
is not cyclic and A is a free group. Then A is free on a set x, y, . . .
and A contains the reduced word xyx
1
y
1
,= 1. This implies A is not
abelian, contradiction.
Example. Consider Z
2
, which is cyclic on 1. A fancy reason why
Z
2
is not free is that Z has no nite subgroups: We have 1 Z
2
and 1 Z but no group homomorphism from Z
2
into Z.
2011 A5. Suppose that (the group) H acts on a set S. Given
s S dene the H-orbit of s. Prove that if s
S is another element
then either s
belongs to the orbit of s or that the orbit sets of s and

s
are disjoint.
Definition. The H-orbit of s is Hs = h.s : h H.
Proof. Dene a relation on S by
s s
if and only if s
Hs.
It is enough to prove that is an equivalence relation on S.
Certainly is reexive: s = e
H
.s.
Suppose s s
, meaning s
Hs: there exists h H such that

s
= h.s. The denition of group-action gives

h
1
.s
= h
1
.(h.s) = (h
1
h).s = e.s = s,
PART A 15
which means s Hs
or equivalently s
s.
Suppose s s
: s
= h.s and s
: s
= g.s
. Then the denition

of group-action again shows s s
easily.
2011 A6. Prove that a group of order 200 cannot be simple.
Proof. Let [G[ = 200 = 2
3
5
2
. Sylow 3 indicates that n
5
(G) = 1,
and Sylow 2 implies that this unique Sylow 5-subgroup is normal in
G.
2011 A7. Let t be an indeterminate, and consider the ring of poly-
nomials 1[t]. Using the fact that 1[t] is a principal ideal domain prove
that the ideal generated by t
2
+ 2 is maximal. Suppose now that we
work with the ring C[t]. Find a polynomial f such that the ideal gener-
ated by f is proper and strictly contains the ideal generated by t
2
+2.
Proof. The ideal (t
2
+ 2) 1[t] comprises all f(t
2
+ 2) with
f 1[t]. Suppose I 1[t] is an ideal and (t
2
+ 2) I 1[t]. Claim
I = (t
2
+ 2) or I = 1[t]. Since 1[t] is a principal ideal domain there
exists a (monic WLOG) polynomial p 1[t] such that I = (p). The
containment (t
2
+ 2) (p) implies that there exists g 1[t] such that
t
2
+ 2 = pg. Considering degrees, there are three cases:
If deg p = 2, deg f = 0 then p is just a multiple of t
2
+ 2 and
so (p) (t
2
+ 2). So I = (t
2
+ 2).
If deg p = 0, deg f = 2 then p is a unit and so (p) = 1[t].
If deg p = 1 = deg f then we have a contradiction: t
2
+ 2 is
irreducible over the reals. (Write down a factorization into
linear polynomials, FOIL, and go forward until an equation is
found which does not hold in the reals.)
Over the complex numbers, t
2
+2 = (t i
2)(t +i
2) is reducible.
This shows that (t
2
+ 2) _ (t + i
2). Any element of the latter has

degree at least 1, which precludes membership of any units. Thus we
have found f = t +i
2 as desired.
2011 A8. Suppose that R = Z
6
and S = 2, 4. Prove that S
1
R
is a nite eld and identify the eld.
Proof. Doing something like
RS :=
Union[
Map[{#, 2} &, Range[0, 5]],
Map[{#, 4} &, Range[0, 5]]
]
16 2011 QUAL
eqclass[x_List] :=
Select[
RS,
Or[
2 (x[[1]] #[[2]] - x[[2]] #[[1]]) == 0,
4 (x[[1]] #[[2]] - x[[2]] #[[1]]) == 0
] &
]
Map[#[[1]] &, Union[Map[eqclass[#] &, RS]]]
will show that S
1
R = 0/2, 2/2, 1/2, 1/4, 3/2, 3/4, 4/2, 5/2, 5/4, or-
der 9. Once we show that every element not equal to 0/2 has a multi-
plicative inverse, it will follow that S
1
F

= F
9
. Tedious.
Part B
Part C
2010 Qual
Part A
2010 A1. Let G be a group and let A be an abelian group. Let
: G Aut(A) be a group homomorphism. Let A
G be the set
A G with the binary operations
(a, g)(a
, g
) = (a +(g)(a
), gg
).
(1) Prove that A
G is a group.
(2) Find : Z
2
Aut(Z
m
) such that the dihedral group D
m
is
isomorphic to Z
m

Z
2
. Do not forget to prove the isomor-
phism!
1. Must prove
(1) A
G is closed under the binary operation

(2) The binary operation is associative
(3) A
G has an identity
(4) all element of A
G have an inverse.
1 is clear. 2 is straightforward and tedious. 3. Let e = (e
A
, e
G
). 4.
The inverse of (a, g) is ((g
1
)(a), g
1
).
2. The dihedral group of order 2m is D
2m
= r, s : [r[ = 1, [s[ =
m, rs = s
1
r. Accordingly, there exist group isomorphisms : s

=
Z
m
, : r

= Z
2
.
The plan is to nd

: r Aut(s) and show D
2m

= s
r.
Use the isomorphisms , to get : Z
2
AutZ
m
and D
2m

= Z
m
Z
2
.
Dene

: r Aut(s) by

(r) = rsr (= rsr
1
). The relations in
D
2m
give

(r)(s) = rsr = s
1
r
2
= s
1
and g g
1
is an automorphism
for any group G. All elements of D
2m
can be expressed as s
i
r
j
where
i, j Z. Dene a map
D
2m
s
i
r
j

(s
i
, r
j
) s
r.
Verify that is an isomorphism.
The isomorphism induces an isomorphism

: Auts

AutZ
m
whereby

: f

f = f
1
. This yields an isomorphism : Z
2
17
18 2010 QUAL
AutZ
m
, where =

1
as shown.
r

Auts
Z
2
/ _ _ _
AutZ
m
.
Now let

: s
r

Z
m

Z
2
be

(s
i
, r
j
) = ((s
i
), (r
j
)). In
conclusion
D
2m
= s
= Z
m
Z
2
.
2010 A2.
(1) Let G be a group of order np where p is a prime. Find a
sucient condition for a subgroup of order p to be normal.
(2) List all isomorphism classes of abelian groups of order 120.
(3) Is there a simple group of order 120?
(4) What is the maximal possible number of elements of order 5
in a group or oder 120?
Answer (1). Pending.
Answer (2). Recall
Theorem (FTFAG). Every nite abelian group is isomorphic to a
group of the form
Z
p
n
1
1
Z
p
n
k
k
where the p
i
are primes whose powers p
n
i
i
are uniquely determined by
the group.
Given a prime p and a positive integer k there is in general one
abelian group of order p
k
for each partition of k. In other words if
k = r
1
+ + r
t
is a partition of k then
t
r=1
Z
r
i
is an abelian group
of order p
k
. The uniqueness portion of the Theorem indicates that
distinct partitions of k yield distinct isoclasses.
Accordingly, if G is abelian of order 120 = 2
3
3 5 then G is
isomorphic to one of the following:
Z
8
Z
3
Z
5
, Z
2
Z
4
Z
3
Z
5
, Z
2
Z
2
Z
2
Z
3
Z
5
.
Answer (3). We will need the following:
Lemma. If [G[ = p
n
m with gcd(m, p) = 1 then n
p
(G)[m.
PART A 19
This is proved below.
Assume G is simple of order 120. Sylow 3 implies n
5
1 (mod 5)
and n
5
[24. Accordingly n
5
1, 6, and simplicity forces n
5
= 6
because a unique Sylow 5-subgroup would be normal by Sylow 2.
Dene an action of G on Syl
5
(G) = P
1
, . . . , P
6
by conjugation.
This induces a homomorphism G S
6
whereby g(i) = j i gP
i
g
1
=
P
j
. Obviously this is a nontrivial homomorphism, and since its kernel
is normal in G simplicity forces it to be a monomorphism G S
6
.
Identify G with the subgroup of S
6
to which it is isomorphic.
If G ,< A
6
then G A
6
is a proper subgroup of G. Then
[S
6
[ = [GA
6
[ =
[G[[A
6
[
[G A
6
[
=
120 360
[G A
6
[
and Lagranges theorem implies [G : GA
6
] = 2, meaing GA
6
G.
This cannot be true unless G A
6
= G, so G < A
6
.
Now [A
6
: G] = 3 by Lagranges theorem, so the action of A
6
on
G by conjugation yields another homomorphism A
6
S
3
which is
nontrivial because [A
6
: G] = 3. But A
6
is simple, so A
6
S
3
must be
a monomorphism. This is a contradiction since [A
6
[ = 360 > 6 = [S
3
[.
In conclusion, G cannot be simple.
Proof of lemma. Let P be a Sylow p-subgroup of G. Sylow 2
says that every other Sylow p-sobgroup is conjugate to P; so there are
as many Sylow p-subgroups of G as there are conjugates to P. Thus
n
p
(G) = [G : N
G
(P)], and since [G[ = [G : N
G
(P)][N
G
(P)[ we see
that n
q
(G) divides p
n
m. The conclusion of Sylow 3 is that n
p
(G) 1
(mod p), which is to say that n
p
(G) is coprime to p Therefore, for the
relation n
p
(G)[p
n
m to hold it must be true that n
p
(G)[m.
Answer (4). Since [G[ = 120 < a subgroup P is a Sylow 5-
subgroup if and only if [P[ = 5. But in this case P is cyclic, P = a
for some a G. On the other hand, a G has order 5 i [a[ = 5
i a Syl
5
(G). Now, if P and Q are distinct Sylow 5-subgroups of
G, then a = P ,= Q = b implies P Q = e. (Any non-identity
elements generates the whole cyclic group.) So when n
5
= 6 (largest
possible by part 3) G contains 6 (5 1) = 24 (exclude the identity)
elements of order 25.
2010 A3. Let G be a group acting on a set X. We say that the
action of G on X is transitive if & only if for all x, x
X there
exists g G such that gx = x
. Prove the following statements.

(1) For all x X, Gx = X. In particular, [G : Stab
G
(x)] = [X[
and if [G[ < then [X[ divides [G[.
20 2010 QUAL
(2) The subgroups Stab
G
(x) are conjugate for all x X.
(3) Suppose that [X[ = k and let : G S
k
= Bij(X) be the
group homomorphism given by the action. Then k divides
n = [G : ker ] and n divides k!.
(4) Suppose that G is a nite simple group, [G[ 3 and H is a
proper subgroup of G with index k. Then G is isomorphic to
a subgroup of A
k
. In particular, [G[ divides
k!
2
.
Hint. Consider the action of G on the set G
_
H.
1. Fix x X. That Gx X is trivial. The reverse containment
Gx X is immediate from transitivity: if x
X then there exists

g G with gx = x
, so x
Gx. In particular, by [Hungerford II-4.2]

the index [G : Stab
G
(x)] equals the cardinal number of xs orbit
[G : Stab
G
(x)] =
gx : g G
Gx
.
If Gis a nite group, then by Lagranges theorem[G[ = [G : Stab
G
(x)][X[
shows that [X[
[G[.
2. Let x, y X; use transitivity to nd g G such that y = gx. If
h Stab
G
(x) then
(ghg
1
)x = (gh)(g
1
x) = (gh)x = g(hx) = gx = y
whence Stab
G
(x) g
_
Stab
G
(y)
_
g
1
. Rephrasing, under the inner
automorphism of G given by g g
1
we have established that
Stab
G
(x) g
_
Stab
G
(y)
_
g
1
. Taking the inverse of this automorphism
and applying it to Stab
G
(y) provides the reverse inclusion. Therefore,
equality holds.
Remark: As there is an action of G on the set of its subgroups by
conjugation, this result says that if G acts transitively on X then it also
acts transitively on the set of stabilizers of elements of X.
3. The homomorphic image of a subgroup is a subgroup, so Im <
S
k
. As S
k
is a nite group of order k!, Lagranges theorem provides
[Im [
k!. Since kernels are normal, G

_
ker is a group and, by the
rst isomorphism theorem [Hungerford I-5.7] G
_
ker Im ; thus
[G : ker ] = [Im [ which implies n[k!.
An element g of G belongs to the kernel of only if gx = x for all
x X, i.e. ker

xX
Stab
G
(x). In particular, if x X is given
then ker < Stab
G
(x) and by [Hungerford I-4.5] we have
n = [G : Stab
G
(x)][Stab
G
(x) : ker ].
By (??), [G : Stab
G
(x)] = [X[ shows that k[n.
PART A 21
4. Let X = G
_
H be the set of left cosets of H in G (so that
[X[ = k) and assume the action of G on X is transitive. Let : G S
k
be as in (??). As ker G, the simplicity of the latter forces either
ker = e or G; since ker < H _ G it must be that ker = e,
whence is a monomorphism. Consequently G Im < S
k
, and we
may identify G with Im , regarding G as a subgroup of S
k
(in the
sense as done for S
k1
S
k
).
By way of contradiction, suppose that G is not a subgroup of A
k
.
Then G ,= G A
k
implies
G
_
G A
k
, e
which in turn implies 1 < [G : G A
k
]. By [Hungerford I-4.8], 1 <
[G : G A
k
] [S
k
, A
k
] = 2 forces [G : G A
k
] = 2 which implies
G A
k
G. The simplicity of G renders this impossible; we must
conclude that G = G A
k
, which means that G < A
k
. In particular,
[G[ divides
k!
2
.
2010 A4. An element e in a ring R is said to be an idempotent if
e
2
= e. The center Z(R) of a ring R is the set of all elements x R
such that xr = rx for all r R. An element of Z(R) is called central.
Two central idempotents f, g R are called orthogonal if fg = 0.
Suppose that R is a unital ring.
(1) If e is a central idempotent, then so is 1
R
e and e, 1
R
e are
orthogonal.
(2) eR and (1
R
e)R are ideals and R = eR (1
R
e)R.
(3) If R
1
, . . . , R
n
are rings with identity then the following state-
ments are equivalent
(a) R

= R
1
R
n
(b) R contains a set of orthogonal central idempotents e
1
, . . . , e
n
such that e
1
+ +e
n
= 1
R
and e
i
R

= R
i
for 1 i n.
(c) R = I
1
I
n
where K
k
is an ideal of R and R
k

= I
k
.
1. Easy.
2. It is straightforward to show that eR is an ideal. Since e, 1
R
e
have the same properties it follows that (1
R
e)R is an ideal. To show
the second part use the following theorem.
Theorem. If I, J R are ideals such that I+J = R and IJ = 0,
then R

= I J.
3. (a) implies (b): Identify the rings under the given isomorphism
for simplicity. The identity element in R
1
R
n
is (1
R
1
, . . . , 1
Rn
).
For 1 i n dene e
i
= (0, . . . , 0, 1
R
i
, 0, . . . , 0). Then for all i, e
2
i
= e
i
22 2010 QUAL
and if (r
1
, . . . , r
n
) R
1
R
n
corresponds to r R under the given
isomorphism then show re
i
= e
i
r, so the e
i
are central idempotents.
By denition, if i ,= j then e
i
e
j
= 0, so the e
i
are orthogonal. Now
e
1
+ +e
n
= 1
R
. Also, e
i
R

= e
i
(R
1
R
n
)

= 0 0 R
i
0 0 = R
i
.
(b) implies (c): R contains a set of orthogonal central idempotents
e
1
, . . . , e
n
such that e
1
+ + e
n
= 1
R
and e
i
R

= R
i
for 1 i n.
As in (2), e
i
R is an ideal of R for 1 i n. If r e
i
R (e
1
R +
+e
i1
R+e
i+1
R+ +e
n
R) then, as in (2), r = 0 by orthogonality.
Since e
1
+ + e
n
R, we have e
1
R + + e
n
R = R as a set. Put
I
k
= e
k
R

= R
k
(by hypothesis) to get (c).
(c) implies (a): R = I
1
I
n
where I
k
is an ideal of R and
R
k

= I
k
. Using the isomorphisms provided by the hypothesis, R =
I
1
I
n

= R
1
R
n
, which is (a).
Part B
2010 B1. Let V be a r-dimensional vector space over a eld K
and suppose f Hom
K
(V, K) is nonzero. Prove that f is surjective
and determine dimker f.
Proof. Assume r > 0 and et v
1
, . . . , v
r
be a basis of V over
K. There exists v =
n
i=1
a
i
v
i
V 0 (with a
i
K) such that
f(v) =
n
i=1
a
i
f(v
i
) ,= 0. This means f(v
i
) ,= 0 for some i 1, . . . , r.
Given b K
, f maps the element

b
f(v
i
)

b
f(v
i
)
b = b, so f is surjective.
It follows that dimf(V ) = 1 = dimK, now Sylvesters law of nullity
states that r = dimf(V ) + dimker f, which means dimker f = r
1.
2010 B3. V and W are vector spaces of a eld K.
(1) Dene a vector space structure on Hom
K
(V, W).
(2) Given a basis X = x
1
, . . . , x)n of V , produce a natural basis
of V
= Hom
K
(V, K).
(3) If dimV = r, dimW = s and X = x
1
, . . . , x
r
, Y = y
1
, . . . , y
s
are bases for V, W respectively nd a natural basis for Hom

K
(V, W).
Answer (1). As a set, Hom
K
(V, W) = K-linear f : V W.
Dene addition and scalar multiplication pointwise via
(f +g)(v) = f(v) + g(v), (af)(v) = af(v)
where the operators on the RHSs are the usual ones in W.
Answer (2). Let X = x
i
: i I be a basis for V over K. All v
V have the form v =
iI
a
i
x
i
where almost all a
i
K are zero. For
PART B 23
any f V
we have f(v) =
iI
a
i
f(x
i
). Dene x
i
V
by x
i
(x
j
) =
ij
. Since X is linearly independent, so is X
= x
i
: i I. To see
that X
spans V
just observe that f(v) =
iI
[a
i
f(x
i
)]x
i
(v).
Answer (3). For each i 1, . . . , r express f(x
i
) = a
i1
y
1
+ +
a
is
y
s
. Now V

= K
r
, W

= K
s
and f has matrix (relative to X, Y )
[f] = [a
ij
]
1js
1ir
Mat
rs
(K). So [f] =
j
a
ij
E
ij
where E
ij
is the
r s matrix with 1
K
in position (i, j) and 0
K
elsewhere. There is an
isomorphism Mat
rs
(K)

= Hom
K
(V, W), so E
ij
: 1 i r, 1 j
s (or their images) is a basis over K.
2009 Qual
Part A
2009 A1. Let D
8
be the dihedral group of order 8, list all the
conjugacy classes.
Solution. More clearly,
D
8
= a, b : [a[ = 4, [b[ = 2, ba = a
1
b.
The elements are
e, a, a
2
, a
3
, b, ab, a
2
b, a
3
b.
Obviously the conjugacy class of e is e, since for any x D
8
xex
1
= xx
1
= e.
To obtain the conjugacy classes for the elements of the form a
i
, two
types of expressions must be considered:
(a
j
)a
i
(a
j
)
1
= a
j
a
i
a
j
= a
i
a
j
a
j
associativity
= a
i
and
(a
j
b)a
i
(a
j
b)
1
= a
j
ba
i
ba
j
= a
j
a
i
bba
j
= a
j
a
i
a
j
= a
i
.
Thus, the conjugacy class of a
i
is a
i
, (a
i
)
1
. Using actual numbers
gives a, a
3
and a
2
.
To obtain the conjugacy classes for the elements of the form a
i
b two
similar types of expressions must be considered:
(a
j
)a
i
b(a
j
)
1
= a
j
a
i
ba
j
= a
j
a
i
a
j
b
= a
i+2j
b
25
26 2009 QUAL
and
(a
j
b)a
i
b(a
j
b)
1
= a
j
ba
i
bba
j
= a
j
ba
i
a
j
= a
j
ba
ij
= a
2ji
b.
Thus, the conjugacy class of a
i
b is a
i+2j
b, a
2ji
b. Using actual num-
bers gives ab, a
3
b and b, a
2
b.
The answer is therefore
e, a, a
3
, a
2
, ab, a
3
b, b, a
2
b.
2009 A2 unnished.
2009 A3. Find all the Sylow 3-subgroups of S
6
Definition. A 3-subgroup is a subgroup in which every element
has order equal to a power of 3. A Sylow 3-subgroup is a -maximal
3-subgroup.
Results [2] II.5.3, II.5.8 indicate that P < S
6
is a Sylow 3-subgroup
if and only if [P[ = 9. For general interest, note that there are exactly
_
720
9
_
, which is to say roughly
1.36 10
20
9-subsets of S
6
.
Solution. Sylows second theorem implies that all Sylow 3-subgroups
are conjugate to one another, so nding a single subgroup of S
6
with
order 9 and taking all its conjugates will provide the answer.
A particular subgroup of order 9 is
H =
(1 2 3), (4 5 6)
_
since all elements have the form (1 2 3)
i
(4 5 6)
j
and both generators
have order 3. Since, for all S
6
,
(1 2 3)
1
=
_
(1) (2) (3)
_
(4 5 6)
1
=
_
(4) (5) (6)
_
it follows that
Syl
3
(S
6
) =
_
_
_
(1) (2) (3)
_
,
_
(4) (5) (6)
_
: S
6
_
_
.
PART A 27
2009 A4. Let
_
P,
i
_
and
_
Q,
i
_
be coproducts of the family
A
i
: i I of objects in a category G. Prove that P and Q are
equivalent.
Definition. P is a coproduct means
i
: A
i
P and for any
object B of G and maps
i
: A
i
B there is a unique : P B such
that
i
=
i
for all i.
P
!
/
B
A
i
i
O
i
>
~
~
~
~
~
~
~
Similar for Q.
P and A are equivalent means there exists f : P Q and g :
Q P such that g f = 1
P
and f g = 1
Q
.
Proof. Since P and Q are coproducts, there are unique f : P Q
and g : Q P to close the following diagrams
P
!f
/
Q
A
i
i
O
i
?

(f
i
=
i
for all i) and
Q
!g
/
P
A
i
i
O
i
?

(g
i
=
i
for all i). Compose these diagrams in two dierent ways:
(1)
P
gf
/
P
A
i
i
O
i
>
~
~
~
~
~
~
~
P a coproduct implies that gf is the unique morphism which
closes this diagram. The map 1
P
also closes the diagram, so
1
P
= g f.
28 2009 QUAL
(2)
Q
fg
/
Q
A
i
i
O
i
?

Q a coproduct implies that f g is the unique morphism which
closes this diagram. The map 1
Q
also closes the diagram, so
1
Q
= f g.
2009 A5 unnished.
Qualmaster Comment. A quick note about something I said yester-
day in response to Drews question about Z
96
and S =
_
2
k
[i = 1, 2, ...
_
.
I said that if T is another multiplicative set and T S, then R R
T

R
S
. This is only true, though, if S contains no zero-divisors, which is
certainly not true in this case.
I have an excerpt from Eisenbuds Commutative Algebra, with a
View Toward Algebraic Geometry that might help to give short answers
for similar problems. There is a mild complication coming from the
fact that if fg=0 in R and we adjoin an inverse of f, then we had better
make g=0. (p59)
What this means in this case is that since 3(2
5
) = 96 = 0 and we
have adjoined an inverse to 2
5
, 3 must be equal to 0. We can show
pretty quickly that
n
2
k
n+3
2
k
. There are suddenly a lot less fractions to
work with. Few enough to work with on a qual. We see quickly that
any two fractions
p
q
and
r
s
such that ps qr = 3 will be equivalent, so
the re(For this particular problem, next you would show that
1
2
k
1
2
k+2
and
1
4
2
2
. By transitivity, the only fractions you should have left at this
point are 0,
1
2
and
2
2
, so once you show that these are in fact distinct,
you are good.)
2009 A6. Find the center of the ring of all n n matrices over
Z
6
.
Definition. The center of M
n
(Z
6
) is
C (M
n
(Z
6
)) = A M
n
(Z
6
) : AB = BA for all B M
n
(Z
6
) .
Solution. For all A M
n
(Z
6
),
A =
n
i=1
n
j=1
a
ij
E
ij
where E
ij
is the matrix with entry (i, j) equal to 1 and all other entries
equal to zero (standard basis element of M
n
(Z
6
) as a Z
6
-module). Since
PART A 29
matrix multiplication is linear, it is clear that A C (M
n
(Z
6
)) if and
only if AE
ij
= E
ij
A for all i, j = 1, . . . , n.
By matrix multiplication, AE
ij
is a matrix consisting of a single
column: its jth column is the ith column of A, and the rest of its
entries are zero. Also, E
ij
A is a matrix consisting of a single row: its
ith row is the jth row of A, and the rest of its entries are zero.
Examining the equations
AE
ii
= E
ii
A, i = 1, . . . , n
shows A to be diagonal: The ith column of AE
ii
is the ith column of
A and all other entries are zero; the ith row of E
ii
A is the ith row of
A and all other entries are zero.
_
_
a
1i
.
.
.
a
ni
_
_
=
_
_
a
i1
a
in
_
_
shows that a
ki
= 0 when k ,= i (in other words all entries of A o the
main diagonal are zero). Thus A = diag(a
11
, . . . , a
nn
).
Examining the equations
AE
i(i+1)
= E
i(i+1)
A, i = 1, . . . , n 1
shows that A is a scalar matrix: The (i + 1)th column of AE
i(i+1)
equals the ith column of A, and all other entries are zero; the ith row
of E
i(i+1)
A equals the (i +1)th row of A, and all other entries are zero.
_
_
a
1i
.
.
.
a
ni
_
_
=
_
_
a
(i+1)1
a
(i+1)n
_
_
These matrices agree at position (i, i + 1), and this shows that a
ii
=
a
(i+1)(i+1)
for all i = 1, . . . , n 1. Thus a
11
= a
22
= = a
nn
, which
means
C (M
n
(Z
6
)) = aI
n
: a Z
6
.
2009 A7 unnished.
2009 A8. Prove that a UFD is integrally closed.
Definition. Let R, S be commutative unitary rings with R S.
An element s S is integral over R i it is a root of a monic
polynomial in R[x].
S is an integral extension of R i all elements of S are
integral over R.
30 2009 QUAL
The integral closure of R in S is the set of all elements of
S which are integral over R. The integral closure of a domain
D over its eld of fractions is called the normalization of D.
R is integrally closed in S i R is equal to its integral closure
in S. A domain is integrally closed i it is integrally closed
in its eld of fractions.
Proof. Let D be a unique factorization domain and F be Ds eld
of fractions. The goal is to prove that if u F is integral over D, then
u D.
By denition, u = a/b, where a, b D, b ,= 0 and (a, b) = 1
(U.F.D.). Suppose (U.F.D.) a = s
e
1
1
. . . s
em
m
, where s
i
D are irre-
ducible and e
i
are positive integers. Let
f(x) = x
n
+ r
n1
x
n1
+ + r
1
x +r
0
R[x]
be a monic polynomial of which a/b is a root
f(a/b) = (a/b)
n
+r
n1
(a/b)
n1
+. . . r
1
(a/b) + r
0
= 0.
Re-arrange this expression
a
n
= b
_
r
n1
a
n1
rab
n2
r
0
b
n1
_
.
If t is an irreducible divisor of b with b = tq, then
a
n
= s
ne
1
1
. . . s
nem
m
= tq
_
r
n1
a
n1
rab
n2
r
0
b
n1
_
.
Accordingly, t = s
i
for some i 1, . . . , m. This means that each
irreducible divisor b is an irreducible divisor of a, and it follows that
each such t divides (a, b) = 1. In other words each irreducible divisor
of b is a unit, so b is merely a product of units and a/b = ab
1
D.
Part B
2009 B1. Let U and W be subspaces of a nite dimensional vector
space V . Prove that
dim(U +W) + dim(U W) = dimU + dimW.
Theorem ([2] IV 2.4, p.183). Every vector space V over a division
ring D has a basis and is therefore a free D-module. More generally
every linearly independent subset of V is contained in a basis of V .
Lemma. The (proper) class of vector spaces over a division ring D
is closed under intersection.
PART B 31
Proof sketch (main result). Choose a basis X of U W (jus-
tied by Lemma and the Theorem). Choose bases Y, Z of U, W re-
spectively that both contain X (justied by the Lemma). Prove that
X(Y X) (Z X) (disjoint unions) is a basis of U +V . Now take
the dimension.
For the proof that X (Y X) (Z X) is a basis:
Proposition. If A, B V are linearly independent and AB = 0,
then A B is linearly independent.
Proof. Suppose
1
, . . . ,
n
D and v
1
, . . . , v
n
A B satisfy
n
i=1
i
v
i
= 0. If v
1
, . . . , v
n
A xor v
1
, . . . , v
n
B then
i
= 0
for all i by the hypothesis of linear independence. Otherwise, re-label
the vectors so that v
1
, . . . , v
k
A and v
k+1
, . . . , v
n
B. Now
A
k
l=1
l
v
l
=
n
j=k+1
(
j
)v
j
B
shows that
k
l=1
l
v
l
,
n
j=k+1
(
j
)v
j
A B, whence
k
l=1
l
v
l
= 0 =
n
j=k+1
(
j
)v
j
.
Using the linear independence of A and B, we infer that
i
= 0 for all
i.
The linear independence of X (Y X) (Z X) follows by
induction. As for spanning, observe that
U +W = u +w : u U, w W
=
_
yY
a
y
y +
zZ
b
z
z : a
y
, b
z
D
_
=
__
xX
a
x
x +
yY X
a
y
y
_
+
_
xX
b
x
x +
zZX
b
z
z
_
: a
x
, a
y
, b
x
, b
z
D
_
=
_
xX
(a
x
+b
x
)x +
yY X
a
y
y +
zZX
b
z
z : a
x
, a
y
, b
x
, b
z
D
_
= span (X (Y X) (Z X))
because as a
x
, b
x
run through all elements of D so does their sum.
Comments. This is [2], IV 2.15.
32 2009 QUAL
2009 B2. Prove that if m and n are coprime integers, then
Z
m
Z
Z
n
0.
Remark. We can do better and show that Z
m
Z
Z
n
Z
d
, where
d = (m, n). Then the result can just be read o (actually only part of
the big proof is needed).
Proof. Suppose m = dq
1
and n = dq
2
, and from now on write
=
Z
. The generators of the tensor product do this:
a b = a (b1) = (ab) 1
and since Z
m
is an R-R bimodule
(ab) 1 = ab(1 1)
so that Z
m
Z
n
= 1 1 is cyclic with generator 1 1. Observe
m(1 1) = m1 = 0 1
and
n(1 1) = n 1 = 1 n = 1 0 = 0
which shows that the order of 1 1 (and hence the order of the tensor
product) divides d. In particular,
Z
m
Z
n
d.
Dene : Z
m
Z
n
Z
d
by
(a, b) = ab (mod d).
To see that is well-dened suppose (a, b) = (c, d). This means a c
(mod d) and b e (mod d), suppose
a = dq
3
+c, b = dq
4
+e.
Then
ab = mnq
5
+mq
6
+nq
7
+ce
where q
5
= q
3
q
4
, q
6
= eq
3
, and q
7
= cq
4
. So
ab ce = mnq
5
+mq
6
+ nq
7
= dq
1
dq
2
q
5
+ dq
1
q
6
+dq
2
q
7
= dq
8
0 (mod d)
where q
8
= dq
1
q
2
q
5
+ q
1
q
6
+ q
2
q
7
. Similarly tedious is the verication
that is middle-linear using the ring axioms of Z
d
.
The universal property of the tensor product provides an abelian
group homomorphism

: Z
m
Z
n
Z
d
such that

(a b) = (a, b).
In particular

(1 1) = 1 (mod d).
PART B 33
If the order of 1 1 equals k < d, then k(1 1) = 0. But then
0 =

(0) =

_
k(1 1)
_
= k
(1 1) = k1 = k ,= 0
which is crazy. Therefore the order of 1 1 is d.
2009 B3. Let R be a ring. Let M be a nitely generated R-module,
and let N be a projective R-module. Prove that if f : M N is a
surjective homomorphism, then the kernel of f is nitely generated.
Lemma. If P is a projective R-module that is a quotient of an R-
module M, then P is isomorphic to a direct summand of M.
Proof. Let : M P denote the canonical epimorphism from
the hypothesis that P is a quotient of M. Since P is projective there
exists a unique R-module homomorphism f : P M making the
following diagram
P
1
P
!f
.~}
}
}
}
M

/
P
/
0
commute: f = 1
P
. But this is one of the (several) equivalent criteria
for the short exact sequence
0 ker

M

P 0
to be split exact. In other words, there exist R-modules J, K and
R-module isomorphisms ker J, P K such that the following
diagram
0
/
ker

/
O
M
f
/
O
P
/
O
0
0
/
J
/
J K
/
K
/
0
commutes. Certainly then, P K J K M.
Lemma. If M is an R-module that is generated by n elements, then
every quotient of M is generated by at most n elements.
Proof sketch. Suppose that M is generated by m
1
, . . . , m
n
.
Suppose Q is a quotient of M with canonical R-module epimorphism
: M Q. Verify that Q is generated by (m
i
) : 1 i n.
Proof of 2009 B3. Using the rst lemma with M
= ker f, there
are R-modules J, K and R-module isomorphisms ker f J, N K
34 2009 QUAL
such that
0
/
ker f

/
O
M
f
/
O
N
/
O
0
0
/
J
/
J K
/
K
/
0
commutes. Most importantly M ker f N; in particular the latter is
nitely generated. Using the rst isomorphism theorem for R-modules
on the canonical projection epimorphism : ker f N ker f gives
(ker f N)
_
ker ker f.
By the second lemma ker f, which is isomorphic to a quotient of a
nitely generated R-module, must also be nitely generated.
2009 B4. Let R = 1[x] and A = Rz
1
Rz
2
Rz
3
, where
ann(z
1
) =
_
(x+1)
2
(x
2
+1)
_
, ann(z
2
) =
_
(x
2
+1)
2
_
, ann(z
3
) =
_
x
4
1
_
.
Find the elementary divisors and invariant factors of A.
Solution. Note x
4
1 = (x 1)(x + 1)(x
2
+ 1). Since R is a
principal ideal domain and each Rz
i
is cyclic, a standard theorem (IV
6.4(iii)) indicates that Rz
i
R
_
ann(z
i
). So
A
R
_
(x + 1)
2
(x
2
+ 1)
_
R
_
(x
2
+ 1)
2
_
R
_
(x 1)(x + 1)(x
2
+ 1)
_.
The elementary divisors are the prime-power factors appearing in the
denominators, i.e. the prime-power factors of the polynomials mak-
ing up the annihilators
(x + 1)
2
, (x
2
+ 1), (x
2
+ 1)
2
, (x 1), (x + 1), (x
2
+ 1).
The power grid gives the invariant factors
(x
2
+ 1)
(x
2
+ 1)(x + 1)
(x
2
+ 1)(x + 1)
2
(x 1)
2009 B5. If A M
n
(C), then A is conjugate to its transpose.
Remark. Assuming conjugate means similar.
Lemma. If A
i
M
r
i
(C) for i = 1, . . . , n and A =
n
i=1
A
i
, then
A
T
=
n
i=1
A
T
i
.
PART B 35
Proof. Write
A =
_
_
A
1
.
.
.
A
n
_
_
=
_
_
A
1
_
_
+ +
_
_
A
i
_
_
+ +
_
_
A
n
_
_
.
For each i = 1, . . . , n
_
_
A
i
_
_
T
=
_
_
a
11
a
1r
i
.
.
.
.
.
.
.
.
.
a
r
i
1
a
r
i
r
i
_
_
T
=
_
_
a
11
a
r
i
1
.
.
.
.
.
.
.
.
.
a
1r
i
a
r
i
r
i
_
_
=
_
_
A
T
i
_
_
36 2009 QUAL
whence
A
T
=
_
_
_
_
A
1
_
_
+ +
_
_
A
i
_
_
+ +
_
_
A
n
_
_
_
_
T
=
_
_
A
1
_
_
T
+ +
_
_
A
i
_
_
T
+ +
_
_
A
n
_
_
T
=
_
_
A
T
1
_
_
+ +
_
_
A
T
i
_
_
+ +
_
_
A
T
n
_
_
=
_
_
A
T
1
.
.
.
A
T
n
_
_
=
n
i=1
A
T
i
.
Proof of the main result. Ais similar to a matrix J in Jordan
form,
A J =
n
i=1
J
k
i
(
i
)
where
i
C are As eigenvalues. By elementary properties of the
similarity relation and the Lemma,
A
T
J
T
=
n
i=1
J
k
i
(
i
)
T
.
For all i = 1, . . . , n
J
k
i
(
i
)
T
=
_
i
1 0 0
0
i
1 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0
i
_
_
T
=
_
i
0 0
1
i
0
0 1 0
.
.
.
.
.
.
.
.
.
.
.
.
0 0
i
_
_
.
The latter is similar to J
k
i
(
i
) by using elementary row/column op-
erations (EROs/ECOs) to move the 1 entries back above the main
diagonal.
As J is a direct sum of Jordan blocks, each ERO/ECO aects
exactly one direct summand. Hence, if J
k
i
(
i
)
T
J
k
i
(
i
) by m
i
PART C 37
EROs/ECOs, then J
T
J by (
i
m
i
) EROs/ECOs. Now
A
T
J
T
J A.
Part C
Do four out of the ve problems. Inducate which of the 4 that I am
to grade.
2009 C1.
Let K be a eld of characteristic zero. Let f K[x] be a cubic
whose discriminant is a square in K. Show that either f is irreducible
over K or factors completely over K.
Proof. Since char K = 0, f has three distinct roots u
i
in some
splitting eld F K,
f(x) = (x u
1
)(x u
2
)(x u
3
) F[x].
Now, f is either irreducible or reducible; it will suce to show that
when f is reducible, it actually splits in K[x].
Suppose f = hk with h, k K[x]. Then (without loss of generality:
re-label h, k) deg h = 1, deg k = 2; say h(x) = x u
1
(again without
loss of generality: re-label the roots). Now k is either reducible or
irreducible over K. If k is irreducible over K, then u
2
, u
3
F K by
virtue of the setup. Consequently there exists Aut
K
F such that
(u
2
) = u
3
.
Regarding Aut
K
F < S
3
the K-automorphism identies with the
permutation (23), which is clearly odd. But since D is a square in K, a
standard theorem indicates that Aut
K
F < A
3
, which is a contradiction
because the former has been shown to contain an odd permutation.
In conclusion, k is irreducible over K. But this just means that k
splits in K; therefore f also splits in K.
2009 C3.
Let g
n
(x) be the nth cyclotomic polynomial over , let p be a prime
integer not dividing n, let F = Z
p
. Suppose the canonical image g
n
(x)
of g
n
(x) in F[x] remains irreducible in F[x] and let E = F[x]/( g
n
(x)).
Show that the Galois group G(E/F) is isomorphic to the group of units
(Z
n
)
of Z
n
.
Proof. Note that F[x] is a PID and ( g
n
) is a maximal ideal, so
E really is a eld. By design E contains a root u = x + ( g
n
) of g
n
;
38 2009 QUAL
actually E F(u). The denition of g
n
is
g
n
=
(n)
j=1
(x
j
)
where
j
are the distinct primitive nth roots of unity in a xed algebraic
closure of .
Since p ,[n, g
n
is the nth cyclotomic polynomial over F and u, as
a root of g
n
, is a primitive nth root of unity over F. Now E = F(u)
contains all nth roots of unity and is hence a splitting eld of x
n
1 over
F, in other words E is a cyclotomic extension of F of order n. Using
p ,[n again, (x
n
1)
= nx
n1
is coprime to x
n
1 so its irreducible
factors are separable. This means that the extension E F is Galois,
and by the fundamental theorem
[Aut
F
E[ = [E : F] = [F(u) : F] = (n).
Each element Aut
F
E is completely and uniquely determined
by () =
i
(some i 0, . . . , n 1). But
1
() =
j
similarly,
and since
1
= 1
E
it follows that i (Z
n
)
. Accordingly, there is a
function
Aut
F
E

i
_
(Z
n
)
,
_
which is easily seen to be a monomorphism of groups. This means that
the Galois group is abelian and isomorphic to a subgroup of (Z
n
)
.
But the order of the two groups coincides, so it must be the case that
Aut
F
E (Z
n
)
.
The steps in this proof are rst show that E is a cyclotomic exten-
sion of F (algebraic, Galois, spliting eld of g over F), and then using
this to derive the desired conclusion. Most of the work goes into the
rst part.
2009 C5.
(a) Dene what it means for f [x] to be solvable by radicals.
Definition. The polynomial f [x] is solvable by radicals if and
only if there exists a radical extension F of and a splitting eld E
of f over such that F E .
Equivalently, since char = 0, f is solvable by radicals i its Galois
group (over ) is solvable.
(b) Show that f(x) = x
5
6x + 3 is not solvable by radicals.
Theorem. If f [x] is irreducible and deg f = p and f has ex-
actly two complex roots, then the Galois group of f over is isomorphic
to S
p
.
PART C 39
Eisensteins criterion (p = 3) shows that f is irreducible in [x].
Using calculus, the graph of f can be produced and indicates that f
has exactly three real (and hence exactly two complex) roots. By the
theorem, the Galois group of f over is S
5
, which is not solvable.
2008 Qual
Part A
Do four out of the ve problems. All answers must be justied.
2008 A1. Let G be a group and let A be an abelian group. Let
: G Aut(A) be a group homomorphism. Let A
G be the set
A G with the binary operation
(a, g)(a
, g
) = (a + (g)(a
), gg
) .
(i) Prove that A
G is a group.
Proof. Associativity is straightforward and not worth writing down;
compare both sides of the equation to be proved
(a, g) ((a
, g
)(a
, g
)) vs. ((a, g)(a
, g
)) (a
, g
).
The identity element is
e
A
G
= (e
A
, e
G
)
and the inverse of (a, g) A
G is
(a, g)
1
=
_
(g
1
)(a
1
), g
1
_
.
Both statements should be veried, which is also easy.
(ii) Prove that the dihedral group D
m
is isomorphic to Z
m
Z
2
for
some : Z
2
Aut(Z
m
).
Definition ([2], I 6.13).
D
m
= x, y : [x[ = m, [y[ = 2, yx = x
1
y;
note [D
m
[ = 2m.
Lemma. If G is a group and H, K < G such that K < N
G
(H),
then there exists a group homomorphism : K Aut(H).
Proof. Use the map
K k (k) Aut(A)
whereby
(k) : a kak
1
41
42 2008 QUAL
and verify that its operation preserving.
Lemma. If N, K < G satisfy N G, N K = e, and G = NK,
then G N
K with as in the previous Lemma.

Proof. Dene f : N
K G by
f(n, k) = nk.
Obviously f is surjective; verify that its operation preserving. Use the
equivalence of injectiveness with ker f = e via the hypothesis that
N K = e.
Proof of 2008 A1(ii). Let N = x Z
m
and K = y Z
2
.
Since [D
m
: N] = 2 it follows that N D
m
, hence N
Dm
(N) = D
m
.
It is clear that the hypotheses of Lemma 1 are satised, so there is a
group homomorphism : K Aut(N). Verify that N K = e;
D
m
= NK is true by the denition of D
m
. By Lemma 2, D
m

N
K Z
m
Z
2
.
2008 A2. Let p, q be primes with p > q, and suppose G is a group
of order p
n
q for some n > 0. Prove that G is not simple.
Lemma. Under the stated hypotheses, n
p
(G) divides q.
A more general statement is true: replace q with m such that
(p, m) = 1.
Proof. Sylows second theorem says that any two Sylow p-subgroups
P, Q < G are conjugate, i.e. there exists g G satisfying gQg
1
= P.
It follows that n
p
equals the size of the conjugacy class of Q under
conjugation
n
p
=
gQg
1
: g G
.
It is a theorem that the latter quantity is exactly [G : N
G
(Q)]. By
Lagranges theorem, p
n
m = n
p
[N
G
(Q)[ and so
n
p
[p
n
m.
Sylows third theorem says that n
p
1 (mod p); in particular
p ,[n
p
= (n
p
, p) = 1
since p is prime. Accordingly, it must be true that n
p
[m.
Proof of 2008 A2. From the Lemma it follows in particular that
n
p
(G) q < p.
By Sylows third theorem, n
p
(G) 1 (mod p), i.e.
n
p
(G) = 1 +kp
PART A 43
for some k N. These facts necessitate k = 0, otherwise
n
p
(G) 1 + p > p,
contradictory to the Lemmas implication. Thus n
p
(G) = 1, which
means that G has a unique Sylow p-subgroup P _ G. I know the
containment is proper since the order of P is by denition a power of
p.
Sylows second theorem indicates that all Sylow p-subgroups are
conjugate to one another. As P is unique, it follows that any conjugate
of P is equal to P. But this is what it means for P G; hence
there exists a non-trivial normal subgroup of G. Accordingly G is not
simple.
2008 A3. Let G be a group with [G[ = 120 = 2
3
3 5.
Lemma. Let , S
6
and suppose has the (unique) disjoint cycle
decomposition
= (a
1
. . . a
k
1
)(b
1
. . . b
k
2
) . . .
Then
1
has the disjoint cycle decomposition
1
=
_
(a
1
) . . . (a
k
1
)
__
(b
1
) . . . (b
k
2
)
_
. . .
Proof. If (i) = j, then
1
_
(i)
_
=
_
(i)
_
= (j).
Accordingly, if the ordered pair i j appears in the disjoint cycle decom-
position of , then the ordered pair (i) (j) appears in the disjoint
cycle decomposition of
1
.
The lemma will be used frequently in the sequel without reference.
(i) List all isomorphism classes of abelian groups of order 120.
Theorem (Fundamental Theorem of Finite Abelian Groups). Ev-
ery nite abelian group is the direct product of cyclic groups whose
orders have a certain property.
Solution. The isoclasses are
G A
_
Z
120
,
Z
2
Z
4
Z
3
Z
5
Z
2
Z
2
Z
2
Z
3
Z
5
_
.
44 2008 QUAL
(ii) Can G be simple?
Answer. Suppose G is simple. Playing around with Sylows the-
orems shows that n
5
= 6. Let G act by conjugation on the set
Syl
5
= F
1
, F
2
, F
3
, F
4
, F
5
, F
6
of Sylow 5-sybgroups of G. This provides a group homomorphism

: G S
6
whereby verify homoM
(g)(i) = j if and only if gF
i
g
1
= F
j
.
The kernel of is a normal subgroup of G; certainly ker ,= G, so the
simplicity of G necessitates ker = e. Thus is a monomorphism;
identify G and im , from now on regarding G < S
6
.
Ask whether or not G < A
6
. If not, let Gbe an odd permutation
and observe that for any S
6
,
GA
6
(
1
) = S
6
so that GA
6
= S
6
. Since every group involved is nite, the product
theorem ([2], I.4 Theorem 7) gives
[S
6
[ = [GA
6
[ =
[G[[A
6
[
[G A
6
[
= [G A
6
[ =
[G[[A
6
[
[S
6
[
= 60.
By Lagranges theorem, [G[ = [G : G A
6
][G A
6
[, whence [G :
GA
6
] = 2. But this means GA
6
G which is forbidden. Conclude
G < A
6
.
Choose a Sylow 5-subgroup S of G. Since n
5
(G) = 6, the index
[G : N
G
(S)] = 6 as well. By Lagranges theorem,
120 = [G[ = [G : N
G
(S)] [N
G
(S)[ = 6[N
G
(S)[
which implies [N
G
(S)[ = 20. Noting the obvious fact that S < G < A
6
,
The next step is to show that A
6
contains no subgroup of order 20.
If K is a subgroup of A
6
with order 20 = 2
2
5, then Sylows
theorems imply that n
5
(K) 1 (mod 5) and n
5
(K)[4, in other words
n
5
(K) = 1. The unique Sylow 5-subgroup K
1
of K is thus normal.
As a group of order 5, K
1
is cyclic with generator equal to some 5-
cycle in A
6
. Any 5-cycle xes one element of 1, . . . , 6; it may be
assumed without loss of generality that 6 is xed by K
1
s generator.
For simplicity it may also be assumed, again without loss of generality,
that
K
1
=
(1 2 3 4 5)
_
.
Since any conjugate of (1 2 3 4 5) is again a 5-cycle, for simplicity the
ambient group may be taken to be S
5
rather than S
6
.
PART A 45
Any permutation N
S
5
(K) satises
(1 2 3 4 5)
1
=
_
(1) (2) (3) (4) (5)
_
= (1 2 3 4 5)
k
for some k 1, 2, 3, 4. Certainly there are exactly 5 4 = 20 such
, which means that [N
S
5
(K)[ = 20. Since K
1
K, K is contained in
the normalizer of K
1
and hence K = N
S
5
(K
1
). But (1 2 4 3) N
S
5
(K
1
)
since it is an odd permutation satisfying
(1 2 4 3)(1 2 3 4 5)(1 2 4 3)
1
= (2 4 1 3 5) = (1 2 3 4 5)
2
.
In conclusion, A
6
contains no subgroup of order 20. At this point it is
clear that G cannot be simple, so there are no simple groups of order
120.
(iii) What is the maximal possible number of elements of order 5 in
G?
Solution. A Sylow 5-subgroup of G has order 5, so it is cyclic.
Sylows second theorem states that if g G has order 5 then g F
i
for some i. Furthermore, each F
i
is realized through conjugation: for
each i, Syl
5
= xF
i
x
1
: x G.
There can be either 1 or 6 Sylow 5-subgroups (cf. part (ii)). Distinct
Sylow 5-subgroups have trivial intersection: If F
i
,= F
j
then there exists
x F
i
F
j
. Each such x generates F
i
, thus x, x
2
, x
3
, x
4
/ F
j
, so
F
i
F
j
= e.
Summarily, each Sylow 5-subgroup contains exactly 4 elements of
order 5, and each element of G having order 5 is contained in some
Sylow 5-subgroup. Since distinct Sylow 5-subgroups have distinct non-
identity elements, there can be at most 6 4 = 24 elements of order 5
in G.
(iv) How many conjugacy classes are there in S
6
? The answer pro-
vided here works for any n, but I am going to specialize to n = 6.
Definition. If S
6
is the product of disjoint cycles of lengths
n
1
, . . . , n
r
with n
1
n
r
(including 1-cycles) then the list n
1
, . . . , n
r
is called the cycle type of .
Definition. A partition of 6 is a non-decreasing sequence of pos-
itive integers whose sum is 6.
Proposition. , S
6
are conjugate if and only if they have the
same cycle type.
Proof. () is immediate from the Lemma.
() Suppose
1
and
2
have the same cycle type. Including 1-cycles,
order their cycles in non-decreasing length somehow. If 2 cycles in
46 2008 QUAL
1
/
2
have the same length this is not unique. Cover the parentheses
with your nger, this gives two lists of the numbers 1 through 6 where
in each number appears exactly once. Say
1
= (a
11
. . . a
1n
1
) . . . (a
r1
. . . a
rnr
)
a
11
, . . . , a
1n
1
, . . . , a
r1
, . . . , a
rnr
2
= (b
11
. . . b
1n
1
) . . . (b
r1
. . . b
rnr
)
b
11
, . . . , b
1n
1
, . . . , b
r1
, . . . , b
rnr
.
Let S
6
denote the permutation dened by
(a
ij
) = b
ij
.
Since the parentheses appear in the same positions in
1
and
2
s dis-
joint cycle decompositions it follows from the Lemma that
1
1
=
2
.
So there exists a bijection between conjugacy classes of S
6
and cycle
types. But cycle types are in one-to-one correspondence with partitions
of 6. There are 11 partitions of 6, so there are 11 conjugacy classes in
S
6
.
2008 A4. Let Z[i] = x+yi : x, y Z, i
2
= 1. This is a unital
ring and Z identies with a subring of Z[i].
(i) Is the ideal of Z[i] generated by 5 prime? To say that (5) is
prime means for all x, y Z[i], xy (5) implies x (5) or y (5).
This statement is false when x = 2 +i, y = 2 i:
(2 + i)(2 i) = 4 + 1 = 5 (5)
but 2 i / (5). So (5) is not prime.
(ii) Is Z[i] a domain? If so, describe its eld of fractions. Note that
Z[i] is a commutative unital ring, since Z is. Let a + ib, c + id Z[i]
and suppose
0 = (a + ib)(c + id) = (ac bd) + i(ad +bc).
Accordingly,
ac bd = 0
ad + bc = 0.
Multiplying the rst equation by c, the second by d and adding shows
that a(c
2
+d
2
) = 0. As Z is a domain, it follows that a = 0 or c
2
+d
2
= 0.
If a = 0, then what we have is
ib(c + id) = 0 = ibc = bd
PART A 47
a contradiction since ibc is purely imaginary while bd is an integer. It
must be that c
2
+d
2
= 0, and since c
2
, d
2
0 it follows that c
2
= 0 = d
2
whence c = 0 = d. Consequently, c + id = 0, which shows that Z[i] is
a domain.
Let F denote the eld of fractions of Z[i]. Formally, F consists of
equivalence classes of ordered pairs (x, y) Z[i] (Z[i] 0) under
the relation
(x, y) (x
, y
) xy
= yx
.
The equivalence class of (x, y) is denoted x/y.
The eld of fractions of Z is , so any eld containing Z and i must
contain
[i] = r +is : r, s .
In particular, F [i]. The multiplicative inverse of c + id/1 in F is
1
c + id
=
id
c
2
+d
2
,
so
a +ib
c +id
= (a + ib)
id
c
2
+d
2
=
bd
c
2
+d
2
+ i
ad
c
2
+ d
2
[i]
whence F [i]. In conclusion, F [i].
(iii) Choose a maximal ideal P in Z[i] and describe the localization
of Z[i] at P. An element x Z[i] is irreducible if and only if the ideal
(x) is maximal in the set of all proper principal ideals of Z[i]. As
the Gaussian integers form a euclidean domain, they are a principal
ideal domain. To answer this question it suces to nd an irreducible
element of Z[i], take its principal ideal, and write down the localization
of Z[i] at said ideal.
The element 2 Z[i] is not irreducible since 2 = (1 + i)(1 i)
and neither factor on the right belongs to Z[i]
. However, 3 Z[i] is
irreducible: it is irreducible over Z and any purported factorization
3 = (a +bi)(c +di)
must satisfy ad + bc = 0 and ac bd = 3. Tedious consideration
of the latter equations shows that such a factorization cannot occur.
Consequently, the ideal (3) is maximal.
The localization of Z[i] at (3) is dened as follows: Let S = Z[i]
(3), observe this set is multiplicative. Then the ring of fractions
S
1
Z[i] =
_
a +bi
c +di
: 3 ,[c +di
_
is the desired localization.
48 2008 QUAL
2008 A5.
Definition ([2], X 3.1). A morphism f : C D of a category ( is
epic if and only if for all objects E of ( and all morphisms g, h : D E
the following implication is true
g f = h f g = h.
(i) Give an example of a category in which a morphism between two
objects is epic if and only if it is surjective.
Example. The example is the category Set whose objects are sets
and whose morphisms are functions. Here is a proof of the equivalence
of the two concepts.
() Assume f : C D and D = f(C). Given an object E of
( choose g, h : D E to satisfy g f = h f. Given y D, by
hypothesis there exists x C such that y = f(x). But now
g(y) = g f(x) = h f(x) = h(y)
and since y was chosen arbitrarily it follows that g = h.
() Assume f is epic and take E = D. The identity map 1
D
: D
D is available, and by virtue of the setup we also have the inclusion
map : f(C) D. Note that Hom
Set
_
f(C), D
_
Hom
Set
(D, D) and,
for all x C
f : x f(x) f(x)
1
D
f : x f(x) f(x).
Accordingly f = 1
D
f; by hypothesis = 1
F
. In particular it is
true that
f(C) = dom = dom 1
D
= D
which means f is surjective.
(ii) Give an example of a category ( and of an epic morphism be-
tween two objects in ( which is not surjective.
Lemma. If R is a ring and f, g are ring homomorphisms R
with f[
Z
= g[
Z
, then f = g.
Proof (A. Katz). Observe
f(1) = f(1 1) = f(1)f(1) = f(1)g(1).
PART A 49
When n Z 0,
f(1) = f
_
n
n
_
= f
_
n
n
_
g
_
n
n
_
= f(n)f
_
1
n
_
g(n)g
_
1
n
_
.
Here is a long chain of implications with the hypothesis invoked at just
the right time. Im still using n Z 0.
f(n)f
_
n
n
_
= f(1) = f(n)f
_
1
n
_
g(n)g
_
1
n
_
f
_
1
n
_
f(n)f
_
n
n
_
= f
_
1
n
_
f(n)f
_
1
n
_
g(n)g
_
1
n
_
f
_
n
n
_
f
_
1
n
_
= f
_
n
n
_
f
_
1
n
_
g(n)g
_
1
n
_
f(1)f
_
1
n
_
= f(1)f
_
1
n
_
g(n)g
_
1
n
_
f
_
1
1
n
_
= f
_
1
1
n
_
g(n)g
_
1
n
_
f
_
1
1
n
_
= f
_
1
n
_
f(n)g
_
1
n
_
f
_
1
n
_
= f
_
1
n
n
_
g
_
1
n
_
f
_
1
n
_
= f(1)g
_
1
n
_
f
_
1
n
_
= g(1)g
_
1
n
_
f
_
1
n
_
= g
_
1
1
n
_
f
_
1
n
_
= g
_
1
n
_
.
50 2008 QUAL
Now for any
m
n
,
f
_
m
n
_
= f(m)f
_
1
n
_
= g(m)g
_
1
n
_
= g
_
m
n
_
whence f = g.
Example. Let : Z denote inclusion; it is evidently a ring ho-
momorphism. This map is the candidate for an epic and non-surjective
morphism (in the category Ring of course). The latter is obviously true,
for example
1
2
im , so we need only verify that is epic. For any
ring R choose ring homomorphisms f, g : R to satisfy f = g .
But this just means f[
Z
= g[
Z
, whence f = g by the lemma. Thus
satises the denition of epic.
Part B
All rings are assumed to be unital, and all modules are assumed to
be unitary unless otherwise stated.
2008 B1. Prove that
[x]/(x
5
4x + 2)
is a eld. Show, on the other hand, that
Z[x]/(x
5
4x + 2)
is not a eld.
Proof. Denote f(x) = x
5
4x+2. Eisensteins criterion (Hunger-
ford III.6.15) with p = 2 Z shows f to be irreducible in [x]. But
then (f) is maximal and so [x]/(f) is a eld.
It is true that f is irreducible in Z[x], but false that (f) is maximal
in Z[x], because the latter is not a PID. (See Gallians () proof of
Theorem 17.5). In particular (f) (x, 2). Appeal to Hungerford
III.2.20(ii).
2008 B2. Let R be a ring, and let A, B, C be three R-modules
such that B is a submodule of A, and C A/B. Prove that if C is a
projective R-module, then A B C.
PART B 51
Definition. An R-module C is projective if and only if given
any diagram of R-modules and R-module maps
C
f
X
g
/
Y
/
0
with the bottom row exact there exists a map of R-modules h : C X
such that g h = f.
Lemma (Short Five Lemma). Let R be a ring and
0
/
X
f
/
Y
g
/
Z
/
0
0
/
X
/
Y
/
Z
/
0
be a commutative diagram of R-modules and R-module maps with short
exact rows. Then
(1) If , are monomorphisms, then is a monomorphism.
(2) If , are epimorphisms, then is a epimorphism.
(3) If , are isomorphisms, then is a isomorphism.
Proof that A B C. Let j : C A/B denote the isomor-
phism given in the hypothesis. From the setup there is a short exact
sequence
0 B

A

C 0,
where , denote inclusion and projection, respectively. Consider the
diagram
C
1
C
A

/
C
/
0
(the bottom row is exact). Since C is projective, there exists a map of
R-modules h : C A such that h = 1
C
.
Dene a map of R-modules : B C A by verify
(b, c) = (b) + h(c),
52 2008 QUAL
and consider the following diagram:
0
/
B

/
1
B
B C

/
C
/
j
0
0
/
B
/
A
/
A/B
/
0
Since = 0 and h = 1
C
, it follows that this diagram is commu- verify
tative. Because j, 1
B
are isomorphisms, the Short Five Lemma implies
that is also an isomorphism.
2008 B3. Let R be a commutative ring and I an ideal of R. Let
A be an R-module and denote by IA the submodule of A generated by
all elements of the form ra with r I and a A. Prove that there is
an isomorphism of R-modules
(R/I)
R
A A/IA.
The rst proof is more elementary than the second.
Proof. Since R/I is an R-bimodule it is permissible to view the
(R/I)
R
A as a left R-module
r(s +I a) = (rs) + I a.
Note that I acts trivially on the latter. Running this process backwards
r +I a = r(1 + I a) = (R/I)
R
A = R(1 + I A)
shows that 1 + I a : a A is a set of generators for (R/I)
R
A
over R.
Dene : A (R/I)
R
A by
(a) = 1 +I a.
Verify that is an R-module homomorphism, the last paragraph makes
it clear that is in fact an epimorphism. Observe that if
r
i
a
i
IA
(nite sum), then
r
i
a
i
_
=
(r
i
a
i
) =
1+Ir
i
a
i
=
r
i
+Ia
i
=
0a
i
= 0
so that IA ker .
Passing to the quotient yields an R-module epimorphism
: A/IA (R/I)
R
A,

(a +IA) = 1 +I a.
PART B 53
To show that

is an isomorphism it suces to exhibit a R-module
homomorphism whose composite with the former is the identity on
A/IA. Dene a map
(R/I) A (r +I, a)

ra +IA A/IA.
Verify that is well-dened and middle-linear. The universal property
of the tensor product provides a unique (R-module) homomorphism
: (R/I)
R
A A/IA,

(r + I a) = (r + I, a).
For all a + IA A/IA,

(a + IA) =

(1 + I a) = a +IA.
Therefore

is an inverse for

, both of which are consequently R-
module isomorphisms.
The second proof is more elegant than the rst.
Proof. The following sequence is exact
0 I

R

R/I 0
where , denote the canonical injection, projection R-module homo-
morphisms respectively. Proposition IV 5.4 says that
R
A is right-
exact, so the resulting sequence
I
R
A
1
A
R
R
A
1
A
R/I
R
A 0
is exact. In particular, 1
A
is an R-module epimorphism. By the
rst isomorphism theorem
(R
R
A)
_
(ker 1
A
) R/I
R
A,
and by exactness ker 1
A
= im 1
A
. Observe that (all sums are
nite)
im 1
A
=
_
1
A
_
n
i
(r
i
a
i
)
_
: r
i
I, a
i
A
_
=
_
n
i
_
1
A
(r
i
a
i
)
_
: r
i
I, a
i
A
_
=
_
n
i
_
(r
i
) a
i
_
: r
i
I, a
i
A
_
I
R
A.
Theorem IV 5.7 indicates that there are isomorphisms
R
R
A A, I
R
A IA.
Accordingly
A/IA (R
R
A)
_
(im 1
A
) = (R
R
A)
_
(ker 1
A
) R/I
R
A.
54 2008 QUAL
2008 B4. Let V and W be two vector spaces over a eld k, and
f : V W be a linear map. Prove that f is surjective if and only if
its dual map f
is injective.
Definition. The dual of f is dened by
W
() V
where f
() = f. Its easy to verify that f
is a linear map.
The dual map provides an arrow between arrows.
Proof. (f surjective f
injective) Hypothesis: ker f
= 0.
Let w W be given, by way of contradiction suppose w W im f.
A non-zero element of W
is given by w
, where
w
(x) =
_
0 if x ,= w,
1 if x = w.
For all v V ,
(f
(w
)) (v) = w
f(v) = w
(f(v)) = 0
by the way in which w was chosen. Hence w
ker f
, contradictory
to the hypothesis. In conclusion f is surjective.
(f surjective f
injective) Hypothesis: im f = W. Assume

, W
satisfy f
() = f
(). For all w W there exists (by

hypothesis) v V such that w = f(v). Thus
(w) = f(v) = (f
()) (v) = (f
()) (v) = f(v) = (w),

and since this is valid for all w it follows that = . In conclusion f
is injective.
2008 B5. Find the Jordan normal form of the following matrix
over the eld of complex numbers:
_
_
0 1 0 1
0 0 0 1
0 0 0 1
0 0 0 0
_
_
Solution. Call that matrix M. Make me a sandwich. Cofactor
expansion along row 4 of I
4
M gives ch
A
() =
4
, so = 0 is the
PART C 55
only eigenvalue of M. Accordingly, 0 is the only entry appearing on
the diagonal of Ms Jordan form:
J =
_
_
0 ? 0 0
0 0 ? 0
0 0 0 ?
0 0 0 0
_
_
.
By denition the algebraic multiplicity of = 0 is 4. It follows that the
sum of the sizes of the blocks of J equals 4. The eigenspace of = 0
equals the kernel of M; a quick computation shows that
geom. mult.(0) = nullity(M) = 2.
Consequently, J contains exactly 2 blocks. Finally, observe that M
2
,=
0 but M
3
= 0; this means that Ms minimal polynomial is x
3
, hence
the size of the largest block in J is 3. Putting all this together gives
J =
_
_
0 1 0 0
0 0 1 0
0 0 0 0
0 0 0 0
_
_
which is the Jordan normal form of M.
Part C
Do any 3 problems.
2008 C1.
2008 C2. Let F be a splitting eld of f K[x] over K. Prove
that if an irreducible polynomial g K[x] contains a root in F, then g
splits into linear factors over F.
Corollary ([2], V 1.9, p.236). Let E and F be extension elds of
K and let u E and v F be algebraic over K. Then u and v are
roots of the same irreducible polynomial f K[x] if and only if there
is an isomorphism of elds K(u) K(v) which sends u onto v and is
the identity on K.
Theorem ([2], V 3.8, p.260). Let : K L be an isomorphism
of elds, S = f
i
a set of polynomials (of positive degree) in K[x],
and S
= f
i
the corresponding set of polynomials in L[x]. If F is a
splitting eld of S over K and M is a splitting eld of S
over L, then
is extendible to an isomorphism F M.
56 2008 QUAL
Proposition ([2], V 3 exercise 2, p.267). If F is a spliting eld of
S over K and E is an intermediate eld, then F is a splitting eld of
S over E.
Proof. First it will be shown that if

K is an algebraic closure of
K containing F, then for any K-monomorphism : F

K it is true
that im = F. The desired conclusion will then be deduced.
Write
f(x) =
n
i=1
k
i
x
i
, k
i
K
and let a K-monomorphism : F

K and a root u F of f be
given. Observe that 0 = f(u) implies
0 = f(u) =
n
i=1
k
i
u
i
=
n
i=1
(k
i
u
i
) =
n
i=1
(k
i
)(u
i
) =
n
i=1
k
i
(u)
i
which is nothing but f
_
(u)
_
; in other words (u)

K is also a root
of f.
The hypothesis that F is a splitting eld of f over K can be used
to express
f(x) = k
n
i=1
(x u
i
), k K, u
i
F.
Since

K is a eld,

K[x] is a euclidean domain and hence a unique fac-
torization domain. Thus the previous display is the unique factored
representation of f as an element of F[x]

K[x]. Since u is a root of f
if and only if (x u)
f, it follows that u
1
, . . . , u
n
is the exhaustive list
of fs roots. Combining this with the observation made in the previous
paragraph shows that (u
i
) u
1
, . . . , u
n
for all i = 1, . . . , n. As is
injective, it follows that the action of on the set of fs roots is merely
to permute its elements.
The facts that F = K(u
1
, . . . , u
n
), [K = 1
K
, and permutes
u
1
, . . . , u
n
clearly imply that (F) = F. Now we can prove the
desired result.
Let g K[x] be irreducible and suppose u F is a root of g. By
denition

K contains all roots of g. If v

K is a root of g, then by
the Corollary there exists a K-isomorphism of elds : K(u) K(v)
such that (u) = v. Here is why extends to a K-automorphism of
K: The algebraic closure

K is the splitting eld for K[x] over K; in
PART C 57
particular g splits in

K[x]. The setup is
K
F

K
K(u)

/
K(v)
K K
whereas the setup for the Theorem is
F M
K

/
L
In the present situation

K must be used instead of F, which is another
reason F

K is necessary; also S = g, S
= g, and we should
be able to use K(u) for K and K(v) for L. For the latter is must
be known that

K is a splitting eld for g over K(u) and for g over
K(v); this is guaranteed by the Proposition. Thus extends to a
K-isomorphism :

K

K.
Now [F : F

K is a K-monomorphism. It follows from the
preliminary result of this proof that im = F, so v = (u) F. As
v was an arbitrary root of g, it is clear that all roots of g are members
of F. Therefore g splits over F.
Summary. Using Hungerfords method
Show that if F

K and : F

K is a K-monomorphism,
then im = F.
Observe that sends roots to roots.
Use splitting eld to factor f, note uniqueness. This gives
the list of fs roots.
Use monomorphism to conclude that just permutes the
roots. Use K- to get the result.
Prove that if v

K is a root of g, then v F.
Pick u F (hypothesis), quote result giving a K-isomorphism
: K(u)toK(v) with (u) = v.
Quote the Theorem and Exercise to extend :

K

K.
Restrict [F : F

K to get a K-monomorphism and use
the rst part. v F.
58 2008 QUAL
2008 C3.
2008 C4. If Z
p
F is a eld extension of degree n then x x
p
is a Z
p
-automorphism of F of order exactly n whose xed eld is Z
p
.
Proof. Denote (x) = x
p
. By hypothesis, there exists a basis
u
i
: 1 i n of F over Z
p
F =
_
n
i=1
a
i
u
i
: a
i
Z
p
_
.
As there are p choices for each a
i
, it follows that [F[ = p
n
.
Consider F
= F 0, a nite abelian group under multiplication.

By the Fundamental Theorem of Finite Abelian Groups,
F
i=1
Z
m
i
, m
1
[m
2
[ . . . [m
M
.
Call the group on the right G. Lagranges Theorem implies that, for
all i, the order of each a Z
m
i
divides m
i
and hence divides m
M
. Thus
m
M
a = 0 for all a G, which means that u
m
M
= 1 for all u F
. So
all p
n
1 elements of F
are roots of x
m
M
1. The latter has at most
m
M
distinct roots in F
, it must be that G = Z
p
n
1
, whence
F
Z
p
n
1
.
By Lagranges Theorem [u[ divides p
n
1 for all u F
, hence the
equation u
p
n
1
= 1
F
holds for all u F
. Consequently, u
p
n
= u for
all u F
, which is to say that each u F
is a root of x
p
n
x F[x].
Trivially, 0 is also a root of this polynomial.
Observe that
i
(u) = u
p
i
for all u F and all i = 1, . . . n 1. For
i = n,
n
(u) = u = 1
F
(u); thus [[ divides n. If there exists i < n such
that
i
= 1
F
, then
u
p
i
= u u F
or, equivalently
u
p
i
u = 0 u F
whence all p
n
elements of F are roots of x
p
i
x F[x]. But the latter
has at most p
i
(< p
n
) roots in F, a contradiction.
2007 Qual
Part A
Do four out of the ve problems.
2007 A1. This question is similar to 2010 A1 and 2008 A1.
2007 A2. Let R be a principal ideal domain and dene : R
0 N
0
by (a) = n if a = up
1
p
2
. . . p
n
for a unit u and prime
elements p
i
R. Consider the condition
(*) a
1
, a
2
R d R : a
1
R +a
2
R = (a
1
+da
2
)R.
(1) Show that if R satises the condition (*), then R is a Euclidean
domain with Euclidean function .
(2) Show that (*) implies that |(R) |(R/A) is onto for each
ideal A or R, where |(T) denotes the group of units of the
commutative ring T.
(3) Show that Z does not satisfy (*).
Another way to write (*) is (a
1
) + (a
2
) = (a
1
+da
2
).
1. Let a, b R. If b ,= 0 then certainly (a) (ab). We must
nd q, r R such that a = qb + r with r = 0 or (r) < (b). There
exists d R such that (a) + (b) = (a +db); write
a = (d)b + (a + db).
Since (a + db)[b it is clear that (a + db) (b). If strict inequality
holds then use q = d, r = a + db. If equality holds then by the
denition of we have a + db = ub for some unit u. Then
a = (u d)b
so use q = u d, r = 0.
2. Let A = (a) and select a unit v + (a) R/A. Then there exists
t R such that vt + (a) = 1 + (a), in other words vt 1 A. Write
vt 1 = ma for some m R. Then 1 (vt) + (ma) = (v) + (a) and
so R = (v) + (a) = (v +da) for some d by (*). Thus 1 = s(v +da) for
some s R, in other words v +da is a unit. Now v +(a) = v +da+(a),
and so the canonical map is surjective.
59
60 2007 QUAL
3. If it were the case that Z satised (*), then (2) would be true.
For the ideal (5), there are 4 units in Z/(5) but only two units in Z.
The canonical map is not surjective.
2007 A3. Let S
5
operate on itself by conjugation. How many
orbits does S
5
have?
Proof. This proof references results from [2008 A3(iv)]. In this
context, orbit means conjugacy class. Permutations , S
5
are
conjugate if and only if they have the same cycle type. The cycle types
of S
5
are in one-to-one correspondence with the partitions of 5. There
are seven partitions of 5
5
4 + 1
3 + 2
3 + 1 + 1
2 + 2 + 1
2 + 1 + 1 + 1
1 + 1 + 1 + 1 + 1
so there are seven conjugacy classes in S
5
.
2007 A4. Duplicates 2008 A2.
Part B
Solve four problems out of ve. All rings are assumed to be unital
and all modules are assumed to be unitary and left unless specied
otherwise.
2007 B1. Give an example of a ring R and an R-module M such
that
(i)
R
M is not exact. Dummit and Foote [1] p.403 (4) says that
M is at if and only if
R
M is exact, so it would suce to nd a
non-at R-module. Item (5) says that the Z-module (abelian group)
/Z is injective and not at, so this will be the example.
Example. Let R = Z and M = /Z. The sequence
0 Z

Z
where (n) = 2n is exact, but the induced homomorphism
1
/Z
: Z
Z
(/Z) Z
Z
(/Z)
PART B 61
is not injective. It is closer to the problem statement, but uglier, to
say that the induced sequence
0 Z
Z
(/Z)
1
Q/Z
Z
Z
(/Z)
is not exact. At any rate, to see this just observe that
0 ,= 1 1/2 +Z 2 1/2 +Z = 1 0 +Z = 0
i.e. ker 1
/Z
,= 0.
(ii) Hom
R
(M, ) is not exact. [1] p.402 (1) says that Hom
R
(M, )
is exact if and only if M is projective. So it would suce to nd a
non-projective R-module. I know is one of those, but Ive used that
too many times. Lets go with Z
2
.
Example. Let R = Z and M = Z
2
, and write Hom for Hom
Z
. The
sequence given by canonical projection
Z

Z
2
0
is exact, but the induced sequence
Hom(Z
2
, Z)

Hom(Z
2
, Z
2
) 0
is not exact. Why?
The integers Z have no nonzero nite subgroups (all subgroups are
isomorphic to nZ for some n Z), and the homomorphic image of Z
2
in Z would be a nite subgroup, so it is clear that Hom(Z
2
, Z) 0. But
Hom(Z
2
, Z
2
) is comprised of the zero and the identity homomorphisms,
so it must be that is the zero map and therefore is not surjective.
An example where Hom(, M) is not exact. This is the equiv-
alent to showing that Z
2
is not an injective Z-module, which is true
because it is not a divisible abelian group.
Apply the functor (with R = Z and M = Z
2
) to
0 Z
2
Z
4
Z
2
0
and obtain
0 Hom(Z
2
, Z
2
) Hom(Z
4
, Z
2
) Hom(Z
2
, Z
2
) 0.
All three terms in the new sequence are isomorphic to Z
2
for various
(obvious) reasons, so it can be re-written
0 Z
2
Z
2
Z
2
0.
This sequence is not exact: the rst map is an injection between two
sets of the same (nite) cardinality and hence is an isomorphism. If the
sequence is assumed to be exact, the kernel of the second map equals Z
2
,
62 2007 QUAL
and hence the third term in the sequence is actually 0. Contradiction:
Z
2
,= 0.
2007 B2. Let R be a ring and M
i
, N
i
, i = 1, 2, 3 be R-modules.
Consider a diagram
0
/
M
1
f
1
/
M
2
f
2
/
M
3
/
0
0
/
N
1
g
1
/
N
2
g
2
/
N
3
/
0
with exact rows (all maps are homomorphisms of R-modules).
(i) Suppose that there exists
3
Hom
R
(M
3
, N
3
) such that g
2
2
=
3
f
2
. Prove that there exists
1
Hom
R
(M
1
, N
1
) making the diagram
commute. Which conditions should
2
and/or
3
satisfy to ensure that
1
is injective?
Proof. To begin, check that im
2
f
1
im g
1
. By exactness it
suces to check that the former is a subset of ker g
2
. Given m
1
M
1
g
2
2
f
1
(m
1
) =
3
f
2
f
1
(m
1
) commutativity
=
3
(0) exactness
= 0.
This ensures that the function
1
:= g
1
1

2
f
1
is well-dened. As a composite of R-module homomorphisms it is au-
tomatic that
1
Hom
R
(M
1
, N
1
). As for commutativity,
g
1
1
= g
1
g
1
1

2
f
1
=
2
f
1
.
To guarantee that
1
is injective it suces to require that
2
be injec-
tive; in this case
1
is a composite of three injective maps and must
itself be injective.
(ii) Suppose that there exists
1
Hom
R
(M
1
, N
1
) such that g
1
1
=
2
f
1
. Prove that there exists
1
Hom
R
(M
3
, N
3
) making the diagram
commute. Which conditions should
2
and/or
3
satisfy to ensure that
3
is surjective?
Proof. To begin, verify that g
2
2
is constant on ker f
2
. If m
2

ker f
2
, then
g
2
2
(m
2
) = g
2
2
f
1
(m
1
) for some m
1
M
1
by exactness
= g
2
g
1
1
(m
1
) commutativity
= 0. exactness
PART B 63
This ensures that the function
3
:= g
2
2
f
1
2
is well-dened. As a composite of R-module homomorphisms, it is clear
that
3
Hom
R
(M
3
, N
3
). As for commutativity, if m
2
M
2
then
3
f
2
(m
2
) = g
2
2
f
1
2
f
2
(m
2
) = g
2
2
(m
2
).
For
3
to be surjective it suces to assume that
2
is surjective.
2007 B3 unnished.
2007 B4. Let R be a domain, A be (sic) a n n matrix over R.
(i) Prove that if the system Ax = 0 has a non-trivial solution then
det A = 0.
Proof. Let v R
n
0 satisfy
Av = 0 = 0v.
Then 0 R is an eigenvalue of A with eigenvector v. One characteri-
zation (from [2] VII 3.5(vii) and VII 5.5) of the determinant is
det A = (1)
n
1
. . .
n
where
i
are the eigenvalues of A (with multiplicity). Since
i
= 0 for
some i, it follows that det A = 0.
(ii) Prove or provide a counterexample to the converse. The con-
verse is: if det A = 0 then Ax = 0 has a non-trivial solution.
Proof. If 0 = det A = (1)
n
1
. . .
n
, then since R is a domain
it follows that
i
= 0 for some i. But then there is an eigenvector
v R
n
0 of A for 0, whence the system Ax = 0 has a non-trivial
solution.
(iii) Which, if any, of these statements remain true if we drop the
assumption that R is a domain? Statement (i) remains true, since the
formula used therein for the determinant is valid over any commutative
ring. That said, commutativity is essential.
Statement (ii) may fail in case R contains a zero-divisor, for in this
situation the equation (1)
n
1
. . .
n
= 0 does not imply that
i
= 0
for some i.
2007 B5. Let R, S be commutative rings, : R S be a ring
homomorphism.
64 2007 QUAL
(i) Extend to a ring homomorphism

: Mat
n
(R) Mat
n
(S) and
show that det
_
(A)
_
=
_
det(A)
_
for all A Mat
n
(R).
Proof. Dene : Mat
n
(R) Mat
n
(S) by
_
[a
ij
]
_
=
_
(a
ij
)
.
Observe
_
[a
ij
] + [b
ij
]
_
=

_
[a
ij
+b
ij
]
_
=
_
(a
ij
+b
ij
)
=
_
(a
ij
) + (b
ij
)
=
_
(a
ij
)
+
_
(b
ij
)
_
[a
ij
]
_
+

_
[b
ij
]
_
and similarly,
_
[a
ij
][b
ij
]
_
=

_
[a
ij
]
_
_
[b
ij
]
_
,
which shows that is a ring homomorphism.
By denition,
det[a
ij
] =
Sn
(sgn )
n
i=1
a
i(i)
.
Of course,
_
det[a
ij
]
_
=
_
Sn
(sgn )
n
i=1
a
i(i)
_
=
Sn
_
(sgn )
n
i=1
a
i(i)
_
and since sgn = 1 implies (sgn ) = sgn
Sn
_
(sgn )
n
i=1
a
i(i)
_
=
Sn
(sgn )
n
i=1
(a
i(i)
)
= det
_
(a
ij
)
= det
_
_
[a
ij
]
_
_
.
PART C 65
(ii) Prove that the constant term of the characteristic polynomial of
a matrix A Mat
n
(R) equals (1)
n
det(A).
Proof. Recall the denition of characteristic polynomial
A
(x) = det(xI
n
A).
Now, xI
n
A Mat
n
_
R[x]
_
; consider the evaluation homomorphism
e
0
: R[x] R dened by
e
0
(f) = f(0).
By part (i), e
0
extends to a ring homomorphism e
0
: Mat
n
_
R[x]
_

Mat
n
(R) satisfying
det
_
e
0
(A)
_
= e
0
_
det(A)
_
for all A Mat
n
_
R[x]
_
. Certainly, e
0
_
det(xI
n
A)
_
is the constant
term of As characteristic polynomial. Observe
det
_
e
0
(xI
n
A)
_
= det(0I
n
A) = det(A) = (1)
n
det(A).
Part C
Attempt any ve. All are worth 10 points.
2007 C2.
Prove that (
2,
3) = (
2 +
3). I can do this with a, b 1

rather than 2, 3.
Proof. The inclusion (
a +
b) (
a,
b) is obvious.
Since (
a +
b) is a eld,
(
a +
b)
1
a +
b
=
1
a b
(
b)
so (
b) (
a +
b). Then
a =
(
a +
b) + (
b)
2
(
a +
b)
and similarly
b (
a +
b). Thus the reverse inclusion holds.

2007 C3.
66 2007 QUAL
Let F be a Galois extension of a eld K of degree 27. Prove that
there exist Galois extensions of K contained in F of degree 3 and 9.
Theorem (Fundamental Theorem of Galois Theory). If F is a
nite-dimensional Galois extension of K, then there is a one-to-one
correspondence between the set of all intermediate elds of the extension
and the set of all subgroups of the Galois group Aut
K
F (given by E
E
= Aut
E
F) such that:
(1) the relative dimension of two intermediate elds is equal to the
relative index of the corresponding subgroups; in particular,
Aut
K
F has order [F : K].
(2) F is Galois over every intermediate eld E, but E is Galois
over K if and only if the corresponding subgroup E
= Aut
E
F
is normal in G = Aut
K
F; in this case G/E
is (isomorphic to)
the Galois group Aut
K
E of E over K.
Lemma. Let G be a group of order p
n
, where p is prime. For each
k = 0, . . . , n, G has a normal subgroup of order p
k
.
Proof of 2007 C3. Denote G = Aut
K
F. As F K is a nite-
dimensional Galois extension, the Fundamental Theorem may be used
throughout this proof. Note rst that [G[ = [F : K] = 3
3
. Using
the lemma, there exist H
i
(i = 1, 2) such that [H
i
[ = 3
i
and H
i
G.
Lagranges theorem indicates that [G : H
1
] = 9 and [G : H
2
] = 3. The
Galois correspondence gives intermediate elds
K H
1
, H
2
F
such that (Fundamental Theorem) H
i
K is Galois. ALso,
[H
1
: K] = [G : H
1
] = 9
[H
2
: K] = [G : H
2
] = 3
as desired.
Proof of the Lemma. Induct on n. When n = 0, G (and the
problem) is trivial. Assume a group of order p
n1
has normal subgroups
of order p
i
for i = 0, . . . , n 1.
Let [G[ = p
n
, as a p-group G has non-trivial center. Lagranges
theorem says that [C(G)[ divides p
n
, so in particular p divides [C(G)[.
Cauchys theorem gives an element a C(G) of order p. Thus a <
C(G) has order p and is normal in G. By Lagranges theorem the
group G/a has order p
n1
. The induction hypothesis gives normal
subgroups N
i
G/a of order p
i
for each i = 0, . . . , n 1. Corollary
I 5.12 in [2] says that
N
i
G/a N
i
= K
i
/a
PART C 67
where a < K
i
G. By Lagranges theorem [K
i
[ = p
i+1
. As i runs
through 0, . . . , n1, i+1 runs through 1, . . . , n. As the trivial subgroup
is (trivially) normal in G, the result follows.
2007 C4.
Prove that if the Galois group of the splitting eld of a cubic over
the rational numbers is Z
3
then all roots of the cubic are real.
Proof. Let f [x] denote the irreducible cubic, and let u
i
de-
note its three roots (distinct) in its splitting eld F = (u
1
, u
2
, u
3
) over
. As the coecients of f are rational, f has either three real roots or
one real root and two complex roots which occur as a conjugate pair
(say u
1
1 and u
3
= u
2
C). By way of contradiction, suppose the
latter is true. Any Aut
F is completely determined by its action

on the roots u
i
. Complex conjugation is an automorphism of C and
induces an element Aut
F whereby
(u
1
) = u
1
(u
2
) = u
3
(u
3
) = u
2
.
Certainly is a non-identity element of order 2 in the Galois group.
However, this is a contradiction to the fact that the Galois group is Z
3
:
any non-zero element here must have order 3.
In conclusion, it must be the case that f has three real roots.
2007 C5.
Prove that a nite-dimensional extension of a nite eld is Galois.
Theorem. If K is a nite eld and F is a nite-dimensional ex-
tension of K, then F is algebraic and Galois over K if and only if F
is a splitting of a set T K[x] of separable polynomials.
Proof. Omitted.
Proof of 2007 C5. Suppose K = p
n
with the usual qualica-
tions, and denote m = [F : K] < . Note that F is algebraic over
K since m is nite (given u M the set 1, u, . . . , u
m
is linearly in-
dependent, etc.). Then [F[ = p
nm
and it suces to show that F is a
splitting eld of a separable polynomial over K. Consider
f(x) = x
p
nm
x K[x].
For all u F, u
p
nm
= u, so each element of F is a (distinct) root of
f. But f can have at most p
nm
roots in some splitting eld over K, so
since [F[ = p
nm
it follows that F must be a splitting eld for f over K.
In fact, f has no multiple roots in F, which means that f is separable
over K. In conclusion F, as a splitting eld of the set of separable
polynomials T = f over K, is Galois over K.
68 2007 QUAL
2007 C6.
Prove that in a nite eld of characteristic p, every element has a
unique pth root. Give an example to show that the condition that the
eld be nite is necessary.
Proof. Let [F[ = p
n
. It suces to prove that the function
whereby
F u

u
p
F
is an automorphism.
Obviously (uv) = (u)(v). The binomial theorem provides
(u + v) =
p
i=0
_
p
i
_
u
i
v
pi
= u
p
+ v
p
= (u) + (v)
since p divides
_
p
i
_
for each i = 1, . . . , p1. That is a monomorphism
follows from the fact that F is a domain. Finally, to see that is
surjective recall that there is a group isomorphism
_
F
,
_
_
Z
p
n
1
, +
_
so that u
p
n
1
= 1 for each u F
. But then u
p
n
= u (even for u = 0),
so that
_
u
p
n1
_
= u.
Counterexample. The eld of rational functions Z
2
(x) has char-
acteristic 2, but the element x has no square root. Indeed, if f, g Z
2
[x]
are coprime and
_
f(x)
g(x)
_
2
= x
then f(x)
2
= xg(x)
2
which is a contradiction unless g = 0 since the
LHS is of even degree while the RHS is of odd degree. But g cannot be
the zero polynomial, for in this case the quotient f/g is not dened.
2006 Qual
Part A
2006 A1 unnished.
2006 A2. Find the Sylow 2-subgroups of S
5
.
Solution. Since S
5
is a nite group of order 2
3
3 5, [2] II 5.8
indicates that P is a Sylow 2-subgroup of S
5
if and only if [P[ = 8.
By Sylows second theorem [2], II 5.9 all Sylow 2-subgroups of S
5
are
conjugate. To answer this question, then, it suces to nd a subgroup
of order 8 and compute all of its conjugates. Consider
(1 2 3 4), (1 2)(3 4)
_
.
This subgroup is isomorphic to D
8
and therefore has order 8. So
Syl
2
(S
5
) =
_
(1 2 3 4), (1 2)(3 4)
_
1
: S
5
_
.
After a signicant amount of tedious work, one sees that up to an
isomorphism the Sylow 2-subgroups of S
5
are D
8
and Q
8
, the quater-
nions.
2006 A3 (maybe incorrect). Consider the group
H =
_
_
_
_
_
1 a c
0 1 b
0 0 1
_
_
: a, b, c Z
1
2
_
_
_
under multiplication. Is it nilpotent?
Definition. The ascending central series of H is dened in-
ductively as follows: C
1
(H) = C(H). Project : H H/C(H) and
form the center C
_
H/C(H)
_
, dening
C
2
(H) =
1
_
C
_
H/C(H)
_
_
.
Iterate, dening
C
i+1
(H) =
1
_
C
_
H/C
i
(H)
_
_
.
69
70 2006 QUAL
where denotes the appropriate version of canonical projection in each
instance. The result is a series of subgroups
e < C
1
(H) < C
2
(H) < . . .
Definition. H is nilpotent if and only if C
n
(H) = H for some
n.
Answer in the negative. The only hope is try and form a few
terms of the ascending central series of H. From [2009 A6] we see that
C
1
(H) = C(H) =
_
_
_
_
_
1 0 0
0 1 0
0 0 1
_
_
_
_
_
= e.
Projecting gives H H/C
1
(H) = H/e H. Continuing, we get
C
2
(H) C(H/C
1
(H)) C(H) = e. At this point it is clear that
the ascending central series of H is simply the trivial subgroup; hence
there is no hope that H is nilpotent.
2006 A4.
(1) Determine the units in Z[
5].
(2) Is 7 irreducible? Prove your answer.
1. Dene a norm function by N(a +b
5) = a
2
+5b
2
. Claim that
x is a unit if and only if N(x) = 1.
Assume a
2
+ 5b
2
= 1. If b = 0 then there is nothing to prove.
Assume b ,= 0, then a
2
1 (mod 5) and elementary number theory
says that a 1 or 4 (mod 5). If a = 5x + 1 then a
2
= 25x
2
+
10x + 1, and so 25x
2
+ 10x = 5b
2
. The function f(x) = x(5x + 2)
has range [1/5, ) which in particular contains no negative integers.
Something similar happens in the cases a 4 (mod 5) and a
2
+ 5b
2
=
1. Therefore the units in Z[
5] are 1 + 0
5.
The proof of the claim is annoying and standard. See Gallian Ch.
18 Exercise 1. There is a sketch at the back of the book.
2. Yes. Suppose 7 = (a+b
5)(c+d
5); then 49 = (a
2
+5b
2
)(c
2
+
5d
2
). There are two cases, without loss of generality.
If (a
2
+ 5b
2
) = 1 and (c
2
+ 5d
2
) = 49: If a ,= 0 then 1 a
2
a
2
+ 5b
2
= 1, which means a = 1 and b = 0. So a + b
5 is a unit.
If a = 0 then b
2
= 1/5, impossible.
If (a
2
+ 5b
2
) = 7 = (c
2
+ 5d
2
): We get 0 5b
2
= 7 so 0 b
2
2, meaning b = 1. But then a

2
+ 5 = 7 which implies a
2
= 2,
impossible.
Thats kinda sketchy but I hate these problems.
PART B 71
2006 A5. Prove or disprove that the ring of Gaussian integers
Z[i] is a PID.
Remark. The plan is to prove that Z[i] is a domain, exhibit the
measure and state the properties (ED) that it satises, and use the
measure to prove PID.
Proof. The Gaussian integers are
Z[i] = a +bi : a, b Z, i
2
= 1.
Observe that Z[i] is a commutative unital ring. The proof that Z[i]
forms an integral domain is [2008 A4(ii)].
Let : Z[i] 0 N be dened by
(a +bi) = a
2
+b
2
.
This function satises the following:
(1) If x, y Z[i] and xy ,= 0, then (x) (xy).
(2) If x, y Z[i] and y ,= 0, then there exist q, r Z[i] such that
x = qy +r and either r = 0 or (r) < (y).
Let I be nonzero ideal of Z[i]. The well-ordering of N provides a
least element of the set (x) : x I, say this element is (), I.
Note that () I. Given x I, use (2) to obtain q, r Z[i] such that
x = q + r etc. Since I is an ideal, q I. Thus r = x q I, and
since () is minimal among all elements in I it follows that r = 0. In
other words, [x for all x I; hence I (), so that I = (). As I
was chosen arbitrarily from the nonzero ideals of Z[i], it follows that
the latter is a PID.
Part B
Attempt any four, all questions are worth 10 points.
2006 B1.
(a). Let R be a ring with identity and M a left module for R.
Recall that M is indecomposable if M cannot be written as a direct
sum of two non-zero submodules. Prove that if f : M M is a
homomorphism of modules then f
2
= f implies either f = 0 or f is
the identity.
Proof. Observe that a ker f im f if and only if f(a) = 0
and there exists b M such that a = f(b). Using the hypothesis,
f(a) = f
2
(b) = f(b), whence
0 = f(a) = f(b) = a.
72 2006 QUAL
Therefore ker f im f = 0. Note also that any a M is expressible as
a =
_
a f(a)
_
+ f(a).
By hypothesis af(a) ker f, and certainly f(a) im f. Accordingly,
M = ker f + im f as a set. The standard theorem on internal direct
sums gives
M = ker f im f,
and as M is indecomposable one of these submodules must be 0. If
im f = 0 then f = 0. Elseif ker f = 0, then we can show that f is the
identity: given a M (nonzero) there exists b M such that a = f(b)
since im f = M. By hypothesis
f(a) = f
2
(b) = f(b) = a,
so f is the identity.
(b). Suppose now that M is decomposable. Prove that there exists
f : M M such that f
2
= f and f is dierent from 0 and the
identity.
Constructive proof. Write
M = A
1
A
2
where A
i
are non-zero submodules. Inherent in this setup are canon-
ical projection
i
: M M
i
and canonical injection
i
: M
i
M
homomorphisms. By denition
1

1
: M M satises f
2
= f, but
if a A
2
then
1

1
(a) =
1
(0) = 0
shows that
1

1
is not the identity. Since A
1
is non-zero, it is guar-
anteed that this homomorphism is not 0.
2006 B2. Let R be a ring with identity and e R is such that
e
2
= e.
(a) Prove that 1 e has the same property.
Proof. Well,
(1 e)(1 e) = 1
2
e e +e
2
= 1 e e + e
= 1 e.
PART B 73
(b) Prove that ReR(1e) = 0 and hence that R = ReR(1e).
Proof. Note that
Re = re : r R, R(1 e) = r(1 e) : r R
are ideals of R. By denition, a Re R(1 e) if and only if there
exist r, s R such that
re = a = s(1 e).
Multiply by e on the right
re
2
= s(1 e)e = re = s(1 e)e
and observe that (1 e)e = e e
2
= e e = 0, so that
re = s0 = 0 = a = 0.
This proves Re R(1 e) = 0.
Any a R can be expressed as
ae +a ae = ae + a(1 e) Re +R(1 e),
this means R = Re + R(1 e) as a set. The standard theorem on
internal direct sums now implies that R = Re R(1 e).
(c) Prove that the R-module Re is projective.
Definition. An R-module P is projective if and only if for any
diagram of R-modules and R-module homomorphisms
A
f
/
B
/
0
P
O
with top row exact, there exists an R-module homomorphism : P
A such that f = .
Theorem. An R-module P is projective if and only if there exists
a free R-module F and an R-module M such that F = P M.
Proof of part 3. Observe that R, regarded as a module for it-
self, is free on 1. Part 2 shows that R = Re R(1 e), and since
the latter direct summand is an R-module it follows that Re is projec-
tive.
74 2006 QUAL
2006 B3. Let R be a ring with identity. Regard R as a right
R-module in the usual way and let M be a right R-module. Prove that
Hom
R
(R, M) M as abelian groups.
Proof. Elements of Hom
R
(R, M) are R-linear maps f : R M,
f(ar) = f(a)r, f(a + b) = f(a) + f(b); the group operation on these
elements is (f + g)(a) = f(a) + g(a).
Dene : M Hom
R
(R, M) by
m
(r) = mr.
It is unchallenging to verify that im really does lie within Hom
R
(R, M).
Equally uninteresting is the task of verifying that is an abelian group
homomorphism.
To see that is injective, observe that m ker if and only if
m
= 0, i.e. for all r R,
0 =
m
(r) = mr.
Certainly then, m = 0. As for being a surjection, given f Hom
R
(R, M)
put m = f(1) and observe that, for all r R,
f(r) = f(1r) = f(1)r = mr =
m
(r).
Therefore is an isomorphism of abelian groups.
2006 B4. Consider the ring R = C[x] of polynomials in an inde-
terminate x with coecients in C.
Remark. C is a eld, so C[x] is a euclidean domain and conse-
quently a principal ideal domain.
(a) Let M be a torsion-free module for R with two generators. Prove
that M is free of rank at most two.
Proof. The short proof is that [2] IV 6.5 implies that M a
nitely generated torsion-free module over a principal ideal domain
is free; checking the linear independence of Ms generators will provide
the rank of M. Following are the full details:
Suppose X is a generating set of M over R, [X[ = 2. Since M is
torsion free, for all x X it is true that rx = 0 if and only if r = 0.
There must exist a maximal linearly-independent subset S X. If
S = X then M = X is obviously free of rank 2. If S _ X then
S = a for some a M. The submodule a of R is certainly free of
rank 1. If b X S, (in fact b = X S) then there are r, s R
(not both of which are 0) such that
ra +sb = 0 = sb = ra.
PART B 75
It is guaranteed that s ,= 0, for in that case r = 0 as well: forbidden.
Thus sX a, which implies that sM < a.
Dene a homomorphism of R-modules f : M M by a sa. By
design, im f = sM, ker f = 0 (torsion-free). The rst isomorphism
theorem
M
f
/
M
M/0

/
sM
shows that M is isomorphic to a submodule of a. The standard
theorem on submodules of a free module shows that M is isomorphic
to a free module of rank 1.
(b) Prove that if M is a cyclic R-module and M ,= R then M is
torsion. Under what condition on the torsion ideal will M be simple?
Proof. Assume M is unitary so that M = Ra for some a M,
reference [2] IV 1.5(ii). The order ideal
O
a
= r R : ra = 0
is an ideal of R. If O
a
= 0 then ra = 0 if and only if r = 0. Conse-
quently M is free on a, i.e. M is a free R-module of rank 1. But R
is also a free R-module of rank 1 (since R is unital, it is free on 1).
The uniqueness of the free R-module of rank 1 shows that M R, in
contradiction to the hypothesis that M ,= R. Therefore M must be
torsion.
There is ([2] IV 6.4(iii)) an isomorphism of R-modules M = Ra
R/O
a
, where submodules of M are in one-to-one correspondence with
ideals of R containing O
a
. Since R is a PID, O
a
= (t) for some t R.
It is clear that M has no proper submodules (i.e. is simple) if and
only if there are no proper ideals of R containing (t). The latter is true
precisely when t is irreducible (equiv. prime), so if O
a
is a prime ideal,
then M is a simple R-module.
2006 B5.
(a) Prove that if A and B are invertible nn matrices with entries
in an integral domain R, then A+rB is invertible in the quotient eld
K of R for all but nitely many r.
Proof. Write A = (a
ij
), B = (b
ij
), A +rB = (c
ij
) where
c
ij
= a
ij
+rb
ij
.
76 2006 QUAL
Everyone knows ([2] VII 3.5) that
det(A +rB) =
Sn
c
1(1)
. . . c
n(n)
.
Treating r as an indeterminant, the summands c
1(1)
. . . c
n(n)
can be
regarded as polynomials f
in r of degree at most n. Denote f =
and observe that deg f n. By III 6.7, f has at most n distinct roots
in K.
Regarding A + rB as a matrix over K, VII 3.7 indicates that the
former is invertible if and only if its determinant is a unit. In a eld,
each nonzero element is a unit, so A + rB is invertible over K if and
only if its determinant is nonzero. The polynomial f(r) = 0 for at most
n elements r K.
(b) Prove that the minimal polynomial of a linear transformation
of an n-dimensional vector space has degree at most n.
Proof. Let F be a eld and V an n-dimensional vector space over
F. Choose a linear transformation : V V . Let p
and q
denote
the characteristic and minimal polynomials of , respectively. Denote
the invariant factors of by q
1
, . . . , q
t
= q
. Recall that the invariant

factors may be found by putting the matrix xI
n
A Mat
nn
_
F[x]
_
in RREF: the nonzero diagonal entries are the q
i
.
Corollary VII 4.7(i) states that, relative to some basis of V , has
matrix D
1
D
t
, where D
i
is the companion matrix of q
i
(so that
D
t
is the companion of q
). Then
p
= det(D
1
D
t
) = q
1
. . . q
shows that q
divides p
, say p
= fq
. Since K has no zero-divisors,

n = deg p
= deg(fq
) = deg f + deg q
.
Part C
Answers should be self-contained as much as possible.
2006 C1. Let K F be a nite-dimensional extension.
(i) Dene what it means for F to be separable over K. Note rst
that F K is algebraic.
Definition. An algebraic extension F is separable over K if and
only if the minimal polynomial of each element of F has no multiple
roots in some splitting eld (over K).
PART C 77
(ii) Prove from scratch that if K is nite then F is separable over
K.
Lemma. A polynomial f K[x] is separable if and only if f
,= 0.
Proof. This result is unrelated to nite elds and so the proof is
omitted.
Lemma. If K is a nite eld, then each element of K has a unique
pth root in K.
Proof. The Frobenius map
K a

a
p
K
is a Z
p
-automorphism of K.
Certainly ,= 0, thus as a map of elds it must be injective. It is
clear that (ab) = (a)(b); that preserves addition follows from the
Freshmans dream i.e. the characteristic p binomial theorem
(a +b) =
p
i=1
_
p
i
_
a
i
b
ip
= (a) + (b).
As an injective map from a set into itself, it is immediate that is a
surjection.
Proof of part (ii). Since K is nite it has characteristic p > 0
(prime). Choose u F and let f K[x] be its irreducible polynomial,
say f(x) =
n
i=0
a
i
x
i
with a
n
= 1.
By way of contradiction, suppose that f is not separable. Then the
rst lemma indicates that
f
(x) =
n
i=1
ia
i
x
i1
= 0
which implies that ia
i
= 0 for i = 1, . . . , n. Certainly then, each i
such that a
i
,= 0 must be divisible by p. But this means that f is a
polynomial in x
p
, say f(x) = g(x
p
) where g K[x]. Using the second
lemma, each of gs coecients has a unique pth root; now since K has
characteristic p, it follows that f(x) =
_
g(x)
_
p
. For this reason, f is
reducible in K[x]: a contradiction to the fact that f is an irreducible
polynomial.
(iii) Prove that if K is of characteristic zero then F is separable
over K.
78 2006 QUAL
Proof. Suppose u F has minimal polynomial
f(x) =
n
i=0
a
n
x
n
K[x]
(where a
n
= 1). If f is not separable, then by the rst lemma from the
previous section
f
(x) =
n
i=1
ia
i
x
i1
= 0
which implies that ia
i
= 0 for each i = 1, . . . , n. Since K has char-
acteristic zero, it follows that a
i
= 1 for all i = 1, . . . , n. So f is the
constant polynomial a
0
K[x]. There are two problems with this, (1)
constant polynomials are not irreducible by denition, and (2)
0 = f(u) = a
0
but no member of F can have irreducible polynomial equal to zero.
Therefore, it must be the case that f is separable.
(iv) Give an example of a non-separable nite-dimensional exten-
sion. A nite-dimensional extension is an algebraic extension, which
is separable when the ground eld is perfect. Use an innite eld of
positive characteristic.
Example. Consider the extension
Z
2
(x
2
) Z
2
(x).
As a vector space
Z
2
(x)
Z
2
(x
2
)
= f
1
/g
1
+ xf
2
/g
2
: f
i
/g
i
Z
2
(x
2
),
so this extension is of dimension 2. An extension is separable if and
only if every irreducible polynomial has no multiple roots (in some
splitting eld over the ground eld). But x
2
+1 is irreducible in Z
2
(x
2
)
and factors as (x + 1)
2
in Z
2
(x).
2006 C2.
Let K be a eld of 9 elements. Prove from scratch that K admits
an extension of degree 2 and any two such extensions are isomorphic.
Proof. Let F be a splitting eld of f(x) = x
3
22
x over K. Recall
that K is a splitting eld of g(x) = x
3
2
x over its prime subeld Z
3
.
Since all u K satisfy u
3
2
= u it follows that
u
3
22
= u
(3
2
)
2
= u
3
2
3
2
=
_
u
3
2
_
3
2
= u
3
2
= u.
PART C 79
In other words, each element of K is a root of f. Accordingly, F is a
splitting eld of f over Z
3
.
Since char F = 3, the derivative f
(x) = 1 is relatively prime to

f(x). The multiple roots theorem implies that f has no multiple roots
in F. If E F denotes the set of roots of f, then of course [E[ = 3
22
.
Verify Z
3
E. Now, since F = Z
3
(E) it follows that F = E whence
[F[ = 3
22
. Now
3
22
= [F[ = [K[
[F:K]
=
_
3
2
_
[F:K]
shows that [F : K] = 2.
If F
1
is another extension of K with degree 2, then
[F
1
: Z
3
] = [F
1
: K] [K : Z
3
]
implies [F
1
: Z
3
] = 4, so [F
1
[ = 3
22
. Show that F
1
is also a splitting
eld of f over Z
3
, and use the fact that any two splitting elds of fover
Z
3
are isomorphic.
2005 Qual
Part A
2005 A1 unnished.
2005 A2.
(a) Exhibit 2 distinct Sylow 2-subgroups of S
5
and an element of S
5
that conjugates one into another. Consider the permutations
a = (1 2 3 4), b = (1 2)(3 4)
from S
5
(the symbol 5 is xed by both). Observing that
(1 2 3 4)
=
4,
(1 2)(3 4)
= 2, (1 2 3 4)
1
= (1 4 3 2) and
ba = (1 2)(3 4)(1 2 3 4) = (2 4 3)
a
1
b = (1 4 3 2)(1 2)(3 4) = (2 4 3)
shows that
(1 2 3 4), (1 2)(3 4)
_
D
8
.
As [S
5
[ = 5! = 2
3
15, a subgroup H < S
5
is a Sylow 2-subgroup if
and only if [H[ = 8. Certainly then,
(1 2 3 4), (1 2)(3 4)
_
is a Sylow
2-subgroup.
For any S
5
(1 2 3 4)
1
=
_
(1) (2) (3) (4)
_
, (1 2)(3 4)
1
=
_
(1) (2))((3) (4)
_
.
In particular,
(4 5)
(1 2 3 4), (1 2)(3 4)
_
(4 5)
1
=
(1 2 3 5), (1 2)(3 5)
_
.
These Sylow 2-subgroups are indeed distinct since 5 is xed by one but
not the other.
(b) How many elements of order 7 are there in a simple group of
order 168?
Solution. Let G be a simple group of order 168 = 7 24. Sylows
theorems imply that n
7
1 (mod 7), but the hypothesis that G is
simple forces n
7
,= 1. Another well-known theorem states that n
7
24,
and it follows that the only possibility is n
7
= 8.
Three facts:
81
82 2005 QUAL
(1) Each Sylow 7-subgroup of G contains exactly 6 elements of
order 7.
Justication: A Sylow 7-subgroup P of G has exactly 7
elements since 7 is the highest power of 7 dividing 168. As
a group of prime order, P is cyclic. By Lagranges theorem,
each non-identity element of P generates P; there are exactly
6 such elements in P.
(2) Every element of order 7 belongs to some Sylow 7-subgroup of
G.
Justication: Any a G with order 7 generates a cyclic
subgroup a of G with order 7. A such, a is contained in a
Sylow 7-subgroup P of G; in particular a P. (More is true,
a actually is a Sylow 7-subgroup of G.)
(3) Distinct Sylow 7-subgroups of G have trivial intersections.
Justication: If P and Q are distinct Sylow 7-subgroups of
G and x is a non-identity element of P Q, then x generates
both P and Q whence P = Q. Contradiction.
Taken together, it is clear that G contains 8 6 = 48 elements of order
7.
2005 A3.
(1) Prove that D
8
is not isomorphic to D
4
Z
2
.
(2) If [G[ = p
n
then G contains normal subgroups of order p
i
for
each i = 0, . . . , n.
1. D
8
has generators x, y such that [x[ = 8, [y[ = 2 and xy = yx
1
.
Yet D
4
Z
2
has generators (a, 0), (b, 0), (e, 1) and (ab, 0) = (ba
1
, 0).
The latter contains no element of order 8.
2. Induct on n. When n = 0, G (and the problem) is trivial.
Assume a group of order p
n1
has normal subgroups of order p
i
for
i = 0, . . . , n 1.
Let [G[ = p
n
, as a p-group G has non-trivial center. Lagranges
theorem says that [C(G)[ divides p
n
, so in particular p divides [C(G)[.
Cauchys theorem gives an element a C(G) of order p. Thus a <
C(G) has order p and is normal in G. By Lagranges theorem the
group G/a has order p
n1
. The induction hypothesis gives normal
subgroups N
i
G/a of order p
i
for each i = 0, . . . , n 1. Corollary
I 5.12 in [2] says that
N
i
G/a N
i
= K
i
/a
PART A 83
where a < K
i
G. By Lagranges theorem [K
i
[ = p
i+1
. As i runs
through 0, . . . , n1, i+1 runs through 1, . . . , n. As the trivial subgroup
is (trivially) normal in G, the result follows.
2005 A4. Let F be a eld and let R be the subset of F[x] consisting
of polynomials whose coecient of x is 0. Prove that F is a subring
of R. Prove also that R is not a UFD by showing that x
6
R has two
dierent factorizations in R into irreducibles.
Proof. There is an injective homomorphism of rings : F R
whereby
F a a + 0x + 0x
2
+ R.
Since F is (in particular) a ring, im is a ring.
The monomial x
6
R has factorizations
x
3
x
3
= x
6
= x
2
x
2
x
2
.
The term x
3
is irreducible in R: If x
3
= x
i
x
j
then i + j = 3. If one
of i, j = 0 then that term in the factorization is a unit. Elseif i, j 1
then one of them must be 1 and this purported factorization does not
actually occur from within R. Similarly, x
2
is irreducible in R: Here
i + j = 2 forces both to be equal to 1 if neither is equal to 0.
2005 A5. Let R be a commutative ring. Assume R[x] is a prin-
cipal ideal domain.
(a) Prove that R is a domain.
Proof. There is an injective ring homomorphism : R R[x]
whereby r r + 0x + 0x
2
+ . . . (the constant polynomial f(x) = r).
By denition R im < R[x]. A subring of a domain is a domain, so
R must be a domain.
(b) Prove that the ideal (x) generated by x is prime.
Proof. It must be shown that fg (x) implies f (x) or g (x).
Note that
(x) = hx : h R[x] = p(x) R[x] : p(0) = 0 =
_
n
i1
a
i
x
i
: a
i
R
_
.
Suppose that f =
m
j=0
b
j
x
j
, g =
p
k=0
c
i
x
i
and fg =
q
l=0
d
l
x
l
;
observe d
0
= b
0
c
0
. Now, fg (x) if and only if 0 = fg(0) = b
0
c
0
.
Using the new fact that R is a domain, it follows that one of b
0
or c
0
is equal to zero and hence that one of f, g (respectively) is an element
of (x).
84 2005 QUAL
(c) Explain why (x) is then maximal.
Proof. Suppose I is an ideal of R[x] which properly contains (x).
Using the hypothesis that R[x] is a principal ideal domain, there exists
h R[x] such that I = (h). The statement (x) < (h) is equivalent to
h[x. But (x) is prime if and only if x is prime in R[x], and in a princi-
pal ideal domain primes and irreducibles coincide; so x is irreducible.
Accordingly, h is either an associate of x or a unit of R[x]. The former
cannot hold because (h) is supposed to properly contain (x), therefore
h R[x]
and it follows that (h) = R[x]. Hence (x) is maximal in

R[x].
(d) Conclude that R must be a eld.
Proof. A standard theorem on quotient rings indicates that R[x]/(x)
is a eld if and only if (x) is maximal. Of course
R[x]/(x) = a + (x) : a R
which provides a function
R[x]/(x) a + (x) a R
that is easily veried to be an isomorphism.
Part B
2005 B1. Let : R S be a homomorphism of commutative
unitary rings, I < R, J < S ideals. Prove or disprove each of the
following assertions.
(i) (I) is an ideal in S. No. In order for this to happen must
be surjective.
Counterexample. Inclusion : Z , where (n) = n/1. Z is
an ideal (of itself), but (Z) is not. For example, (1) = 1/1 (Z) but
1
2
(1) =
1
2
/ (Z).
(ii) If J is a prime ideal, so is
1
(J). Yes.
Proof. Recall from Math 521B that
1
(J) is an ideal of R. Mum-
ble some stu about being proper and nonzero or whatever. If a, b R
satisfy ab
1
(J), then by denition (a)(b) = (ab) J. As J is
prime, it follows that (a) J or (b) J; but this is exactly what it
means for a
1
(J) or b
1
(J). Hence
1
(J) is prime.
PART B 85
(iii) If J is a maximal ideal, so is
1
(J). No. In order for this to
happen must be surjective.
Counterexample. Let : [x] denote the canonical injec-
tion (r) = r + 0x + 0x
2
+. . . . From 2005 A5(c), the ideal (x) of [x]
is maximal. (It is straightforward to see that it is prime, and in a PID,
which [x] is because is a eld, prime and maximal are the same.)
But
1
_
(x)
_
= r : r + 0x + (x) = 0
which is clearly not maximal.
2005 B2 unnished.
2005 B3.
(1) Determine with proof Hom
Z
(1, Z).
(2) Determine with proof whether 1 is a projective Z-module.
1. The answer is 0. By way of contradiction suppose is a nonzero
element of the Homset. Let x 1 satisfy (x) = n ,= 0. Multiply
by 1 to get n = n(x/n), implying 1 = (x/n). What is (x/2n)?
Again using 1 we have 2(x/2n) = (x/n) = 1. This means that
(x/2n) = m is an integer satisfying 2m = 1. Impossible, whence
there is no such .
2. The answer is no. By way of contradiction: The Z-module 1 is
projective if and only if 1 is a direct summand of a free Z-module. In
particular, 1 would be a submodule of a free Z-module. The integers
form a principal ideal domain, so 1 would be free as a Z-module because
submodules of free PID-modules are free. This just means that 1 would
be a direct sum of copies of Z. But 1 is a divisible abelian group, and
since a direct summand of a divisible abelian group is divisible (easily
veried) the net result is that the integers form a divisible abelian
group. This is false.
2005 B4. Determine with proof whether the following Z-modules
are injective.
(1)
(2) Z
12
.
Again the argument relies heavily on the result that divisible abelian
groups are exactly injective Z-modules. So this solution should include
(at least) a sketch of this fact. See Hungerford IV.6.
1. The rationals form a divisible abelian group, so they are injec-
tive. The argument is simple: given a rational number and a nonzero
86 2005 QUAL
integer, nd another rational number that when multiplied by said
integer yields the original rational.
2. The ring Z
12
is not divisible. Use y = 2 and n = 12. There does
not exist x Z
12
satisfying nx = y.
Part C
2005 C1. Let p be a prime integer and let F
p
n denote the eld
with p
n
elements in a xed algebraic closure of F
p
.
(a) Show that F
p
n F
p
m if and only if n[m.
Proof. () Include the prime subeld in the tower
Z
p
F
p
n F
p
m.
But then
m = [F
p
m : Z
p
] = [F
p
m : F
p
n] [F
p
n : Z
p
] = [F
p
m : F
p
n] n
so n[m.
() The eld F
p
m is the splitting eld of
f = x
p
m
x Z
p
over Z
p
. Since the characteristic of the ground eld is p, f
= 1
and consequently f has p
m
distinct roots in F
p
m. This means that the
irreducible factors of f are separable, therefore Z
p
F
p
m is a (nite-
dimensional) Galois extension.
Let G = Aut
Zp
F
p
m. By the Fundamental Theorem of Galois The-
ory, [G[ = m and there is a one-to-one correspondence between sub-
groups H < G and intermediate elds Z
p
E F
p
m. Recall that G
is cyclic with generator : u u
p
. In particular, since n[m there is a
unique subgroup, namely
H =
n
< G
of order [H[ = m/n and index [G : H] = n. By the Fundamental
Theorem,
n = [G : H] = [H
: Z
p
]
so H
= F
p
n F
p
m.
(b) Let q be another prime integer and let
E =
_
i=1
F
p
(q
i
)
.
Show that E is an innite extension of F
p
which is not algebraically
closed.
PART C 87
Proof. It is clear that E is an innite-dimensional extension of
Z
p
. Consider the non-constant polynomial
f = x
p
q+1
x Z
p
[x].
If E is algebraically closed then E contains a splitting eld of f over Z
p
,
i.e. E F
p
q+1. But there exists i N (large) such that F
p
q+1 F
p
(q
i
)
,
so by part (a), q + 1[q
i
. This is false, so E cannot be algebraically
closed.
2005 C2.
(a) Factor the polynomial x
9
1 over .
Solution. Recall
x
9
1 =
d|9
g
d
(x)
where g
d
denotes the dth cyclotomic polynomial over . The denition
of the latter is
g
d
(x) =
(x )
where the product is extended over the set of (distinct) primitive dth
roots of unity (in a xed algebraic close of the rationals). Since the
divisors of 9 are 1, 3, and 9:
g
1
(x) = (x 1)
g
3
(x) =
x
3
1
g
1
(x)
= x
2
+x + 1
g
9
(x) =
x
9
1
g
1
(x)g
3
(x)
= x
6
+x
3
+ 1.
which gives
x
9
1 = (x 1)(x
2
+x + 1)(x
6
+x
3
+ 1).
(b) Is x
6
+x
3
+ 1 irreducible over ? Explain.
Solution. It suces to show that x
6
+ x
3
+ 1 = g
9
is irreducible
over Z. Let f Z[x] be a monic irreducible factor of g
9
. Because g
9
is
monic and has no multiple roots it will suce to show that every root
of g
9
is a root of f, whence g
9
[f and equality holds.
Let be a primitive 9th root of unity (i.e. a root of g
9
) and suppose
f() = 0. Given any prime p ,[9 it can be shown (via the division
algorithm, canonical projection, et cetera) that f(
p
) = 0 as well. Now,
if k is any integer in 0, . . . , 9 which is coprime to 9, then k can be
written as p
1
p
r
where no p
i
divides 9. Repeatedly using the fact
88 2005 QUAL
that
p
is a root of f whenever is a root of f, it can be shown that
,
p
1
, . . . ,
k
are all roots of f. This list comprises all primitive 9th
roots of unity, so g
9
[f.
(c) Give the Galois group of x
6
+x
3
+ 1 over . Explain.
Solution. Each primitive 9th root of unity (root of g
9
) generates
the set of all 9th roots of unity and hence generates the set of all
primitive 9th roots of unity (roots of g
9
). Thus a splitting eld of g
9
over is () where denotes a primitive 9th root of unity. The
Galois group of
9
is G = Aut
(). Note that G is also the Galois

group of x
9
1 over .
Any G is completely determined by its action on , but since
sends roots of x
9
1 to roots of x
9
1 it must be true that () =
i
for some i 0, . . . , 8. Similarly
1
() =
j
for some j 0, . . . , 8.
Since
1
= 1 =
1
it follows that i is a unit in Z
9
. The function
dened by
(G, ) i
_
(Z
9
)
,
_
is a monomorphism of groups. Accordingly, G is abelian and isomor-
phic to a subgroup of (Z
9
)
.
The characteristic of is zero, and so the (algebraic) extension
() is Galois. The Fundamental Theorem of Finite-Dimensional
Galois Theory gives
(9) = [() : ] = [G[
(Z
9
)
= (9)
and it follows that G (Z
9
)
.
2005 C3.
(a) Dene what it means for a polynomial f [x] to be solvable
by radicals over .
Definition. f is solvable by radicals over the rationals if and only
if there exists a radical extension F of and a splitting eld E of f
over such that F E .
Equivalently (since the rationals have characteristic zero), f is solv-
able by radicals if and only if the Galois group of f over the rationals
is a solvable group.
(b) Show that f(x) = x
5
4x + 2 is not solvable by radicals.
Theorem. If f [x] is an irreducible polynomial of prime degree
p and has exactly two complex roots, then the Galois group of f over
the rationals is isomorphic to S
p
.
PART C 89
Eisensteins criterion (with p = 2) indicates that f is irreducible
over the rationals. Calculus can be used to produce the graph of f,
which shows that f has exactly three real (and hence exactly two com-
plex) roots. By the theorem, the Galois group of f over the rationals
is S
5
, but this group is not solvable.
2005 C4.
Give an example of a nite extension E F of elds with innitely
many intermediate elds and explain why your example works.
Example. Let
E = Z
2
(x
2
, y
2
) Z
2
(x, y) = F.
Observe rst that the dimension of F over E is 4: Consider the
intermediate eld
E Z
2
(x
2
, y) F
which implies
[F : E] = [F : Z
2
(x
2
, y)] [Z
2
(x
2
, y) : E]
As vector spaces
Z
2
(x
2
, y)
E
= f
1
/g
1
+ yf
2
/g
2
: f
i
/g
i
E
F
Z
2
(x
2
,y)
= f
1
/g
1
+ xf
2
/g
2
: f
i
/g
i
Z
2
(x
2
, y),
so [F : E] = 2 2.
Next it will be shown that F is not a simple extension of E: If
F = E(w) then w
2
2
E. Since E(x, y) = F it follows that u
2
E for
all u F. In particular w
2
E, and so w is a root of the polynomial
f(z) = z
2
w
2
E[z].
Note that w / E implies f is irreducible over E. But then [F : E] =
2 < 4, impossible.
Finally, we are ready to exhibit an innite collection of proper in-
termediate elds. For each i N consider
E E(x +yx
2i
) F.
Since F is not a simple extension of E, these elds are all proper. If,
for some i ,= j two of these intermediate elds coincide, then E(x +
yx
2i
) contains x +yx
2i
and x +yx
2j
and hence must also contain their
dierence
E(x + yx
2i
) x +yx
2i
x yx
2j
= y(x
2i
x
2j
).
Since x
2i
x
2j
E(x + yx
2i
), it follows that y E(x + yx
2i
) as well.
Now it is clear that E(x + yx
2i
) contains Z
2
(x
2
, y). This containment
is certainly proper, because x +yx
2i
/ Z
2
(x
2
, y).
90 2005 QUAL
The tower
E Z
2
(x
2
, y) E(x +yx
2i
) F
leads to the equation
4 = [F : E]
= [F : E(x +yx
2i
)] [E(x +yx
2i
) : Z
2
(x
2
, y)] [Z
2
(x
2
, y) : E]
= [F : E(x +yx
2i
)] [E(x +yx
2i
) : Z
2
(x
2
, y)] 2.
One of these factors must be equal to 1, but this is impossible: E(x +
yx
2i
) is a proper subeld of F and it was shown earlier that E(x +
yx
2i
) ,= Z
2
(x
2
, y).
In conclusion, no two of these intermediate elds coincide. This
shows that the extension E F admits an innite collection of inter-
mediate elds.
2004 Qual
Part A
2004 A1 unnished.
2004 A2.
Proof. By the First Isomorphism Theorem
G

/
Inn(G)

/
Aut(G)
G/C(G)

9
s
s
s
s
s
s
s
s
s
s
where is an isomorphism G/C(G) Inn(G). All subgroups of a
cyclic group are cyclic, so Inn(G) is cyclic. Evidently G/C(G) is also
cyclic; from the G/Z theorem it follows that G is abelian.
2004 A3. Let G be a nite group and H G a normal subgroup
of prime order p. Prove that H is contain in each Sylow p-subgroup of
G.
Proof. Express [G[ = p
n
m with (p, m) = 1. Sylows rst theorem
says that there exist subgroups G
i
of order p
i
such that
H G
2
. . . G
n
.
Since G is nite, Syl
p
(G) = P < G : [P[ = p
n
. In particular, H < G
n
and the latter is a Sylow p-subgroup. Now we know that H is contained
in some Sylow p-subgroup of G. Sylows second theorem indicates that
Syl
p
(G) = xG
n
x
1
: x G.
Now H < G
n
implies that for each P Syl
p
(G) there exists x G such
that xHx
1
< P. As H is normal in G, we have H = xHx
1
< P.
91
92 2004 QUAL
Part B
2004 B1. Let I and J be ideals in a commutative unitary ring R
such that I +J = R and let A be a left R-module. Show that
A
(I J)A

A
IA

A
JA
.
Remark. There is a more elegant proof involving high-level prop-
erties of ideals.
Proof. Assume A is unitary so
IA =
_
<
r
i
a
i
: r
i
I, a
i
A
_
,
the denitions of (I J)A and JA are similar. The canonical projection
epimorphisms
1
: A
A
IA
,
2
: A
A
JA
can be combined to form a homomorphism of R-modules
: A
A
IA

A
JA
whereby (a) = (a + IA, a + JA). The rst isomorphism theorem for
R-modules indicates that
:
A
ker
im ,
so it suces to prove (1) ker = (I J)A and (2) is surjective.
For (1), observe a ker if and only if a IA JA. It must be
shown that the latter equals (I J)A. This is accomplished by rst
showing that IA JA = (IJ)A and then that IJ = I J.
It is clear from the denitions that (IJ)A IAJA. Since I +J =
R and R is unital, there exist x I, y J such that x +y = 1. Given
a IA JA,
a = 1a = (x +y)a (IJ)A
hence the reverse inclusion holds and IA JA = (IJ)A.
The ideal IJ is always contained in I J, and if r I J then
r = rx + ry similarly to before. But x I, r J, and rx = xr imply
xr IJ; similarly ry IJ. So r IJ and IJ = I J.
It follows that (IJ)A = (I J)A, thus
ker = IA JA = (IJ)A = (I J)A
as desired. It remains to prove (2), that is surjective.
PART B 93
Given (a
1
+IA, a
2
+ JA) A
_
IA A
_
JA, observe
(ya
1
+xa
2
) = (ya
1
+ IA, xa
2
+JA)
=
_
(1 x)a
1
, (1 y)a
2
_
= (a
1
+IA, a
2
+ JA).
So is an R-module epimorphism.
2004 B2. Let p be a prime integer and let Z(p
) be the subgroup
of the additive group Z/Z generated by the cosets of the form 1 + p
i
with i N. Let A be a subgroup of a nitely generated group B. Show
that any homomorphism f : A Z(p
) extends to a homomorphism
g : B Z(p
).
2004 B3. Give an example of a monomorphism f : A B of
abelian groups and an abelian group C such that the induced homomor-
phism Hom
Z
(B, C) Hom
Z
(A, C) is not onto.
Remark. By IV 4.6 this is equivalent to nding a non-injective
Z-module. Lemma IV 3.9 says that injective unitary Z-modules coin-
cide with divisible abelian groups, so it suces to nd a non-divisible
abelian group. Exercise IV 3.4 indicates that Z and Z
n
are not divisible,
so either will do.
Of course, this is just a heuristic composed of unjustied (but true)
assertions. All that matters for the exam is that the example is correct.
Example. Let A = Z, B = , C = Z and consider the inclusion
homomorphism : Z . It must be shown that there exists f
Hom
Z
(Z, Z) such that for all g Hom
Z
(, Z), f ,= g .
Recall from undergraduate group theory that Hom
Z
(, Z) = 0. It
is clear that the identity 1
Z
Hom
Z
(Z, Z) does the job of f, since
g = 0 = 0 ,= 1
Z
= f.
2004 B4.
(a) Sketch the construction of A
R
B. Here, R is a ring, A = A
R
is a right R-module, and B =
R
B is a left R-module. Let F denote the
free group on the set A B = (a, b) : a A, b B, and let K F
be the subgroup generated by the elements
(a +a
, b) (a, b) (a
, b)
(a, b + b
) (a, b) (a, b
)
(ar, b) (a, rb)
94 2004 QUAL
where a, a
A, b, b
B, and r R. The quotient group F/K is

referred to as the tensor product of A and B over R. Notation is as
follows:
F/K A
R
B
(a, b) + K a b
r
i=1
n
i
_
(a, b) + K
_

r
i=1
n
i
(a
i
b
i
), n
i
Z.
(b) State the universal property of A
R
B. Continue using the
notation from (a).
Definition. An map f : AB G is middle linear if and only
if for all a, a
A, all b, b
B and all r R the identities

f(a + a
, b) = (a, b) + (a
, b)
f(a, b + b
) = (a, b) + (a, b
)
f(ar, b) = f(a, rb)
are satised.
The universal property of A B is as follows: If G is an abelian
group and f : A B G is middle linear, then there exists a unique
abelian group homomorphism

f : A
R
B G such that f =

f ,
where : A B A
R
B is the inclusion map. The tensor product
is determined uniquely up to isomorphism by this property.
2004 B5. Let R be a commutative ring and let X be an indeter-
minate. Let M be a maximal ideal of R[X] such that M R = P is a
maximal ideal of R. Show that M = fR[X] + PR[X] for some monic
f R[X] such that the image

f of f in (R/P)[X] is irreducible. (

f is
obtained from f by reducing coecients mod P.)
Proof. The quotient R/P is a eld because P is a maximal ideal
of R. This means that (R/P)[X] is a principal ideal domain. Since
M PR[X], the quotient M/PR[X] is an ideal
1
in (R/P)[X]. Thus
M/PR[X] =

f(R/P)[X] for some

f (R/P)[X]. Verify that M/PR[X]
is maximal
2
hence

f is irreducible, and can be taken to be monic..
1
PR[X] M R[X] gives M/PR[X] R[X]/PR[X]. But R[X]/PR[X]

=
(R/P)[X]: check that R[X] f

f (R/P)[X] satises ker = PR[X] by
looking at coecients. First isomorphism theorem.
2
Suppose there exists an ideal M/PR[X] J _ R[X]/PR[X]. Hungerford
I.5.11/12 (or whatever variant thereof for ideals in rings) sez that there exists a
unique M I _ R[X] such that J = I/PR[X]. The maximality of M forces
I = M.
PART B 95
Our

f is also manifestly nonzero, and so
(1) (f, PR[X])R[X] _ PR[X].
In particular, this means 1 / (f, PR[X])R[X], which is to say that
(2) fR[X] + PR[X] ,= R[X].
Taking (1) and (2) together we get
PR[X] _ fR[X] + PR[X] _ R[X].
It follows
3
that M = fR[X] + PR[X].
2004 B6.
Give an example of an integral domain R which is not a UFD al-
though each element of R factors into irreducibles in R. Justify your
assertions.
Lemma. Let N : Z[
10] Z be dened as
N(a + b
10) = a
2
10b
2
.
Then for all u, v Z[
10], N(uv) = N(u)N(v) and u Z[
10]
if
and only if N(u) = 1.
Proof. The rst part is tedious and straightforward to verify. For
the second, (): If uv = 1 then
1 = N(uv) = N(u)N(v) =
_
N(u), N(v)
_
_
(1, 1), (1, 1)
_
.
(): If
N(a +b
10)
= 1 then
1 = a
2
10b
2
= (a +b
10)(a b
10)
implies that a + b
10 is a unit with inverse a b
10.
Now for the example
Example. Let R = Z[
10]. Note that R is Noetherian, so each

element has a factorization into irreducibles (recursive argument on
principal ideals of divisors). But
2 3 = 6 = (4
10)(4 +
10).
By the lemma 2, 3, 4
10 are irreducible. If 2 = uv then

assuming u, v are not units
4 = N(2) = N(uv) = N(u)N(v) = N(u) = 2 = N(v).
3
Something like this: I = fR[X] +PR[X] is an ideal lying strictly between PR[X]
and R[X] such that I/PR[X] = M/PR[X]. What else could I be then but M?
This is not rigorous.
96 2004 QUAL
But then u = a + b
10 implies that 2 = a
2
10b
2
, impossible for
a, b Z. Similarly for 3. If 4 +
10 = uv then
6 = N(u)N(v) =
_
N(u), N(v)
_
= 2, 3.
This implies that there exist a, b, c, d Z such that a
2
10b
2
= 2 and
c
2
10d
2
= 3, impossible.
Part C
Do four out of the six problems.
2004 C1.
Suppose that F is an extension eld of K. Let u, v F be algebraic
over K and assume that the degree of u is prime to the degree of v over
K. What is the degree of the extension K(u, v) over K? You must
provide a proof for your answer.
The answer is mn.
Proof. Suppose that [K(u) : K] = n and [K(v) : K] = m. The
hypothesis is (m, n) = 1, which implies that [m, n] = mn. Let | =
a
1
, . . . , a
n
be a basis for K(u)
K
and 1 = b
1
, . . . , b
m
be a basis for
K(v)
K
. Observe that 1 is a spanning set for K(u, v)
K(u)
, so [K(u, v) :
K(u)] m. Thus
[K(u, v) : K] = [K(u, v) : K(u)] [K(u) : K] mn = [m, n].
But m and n divide [K(u, v) : K], so [m, n] divides [K(u, v) : K].
In particular
[m, n] [K(u, v) : K]
which implies that
[K(u, v) : K] = [m, n] = mn.
2004 C2.
(a) Let F be an extension eld of K. Dene the Galois group of F
over K. Suppose that L is a subeld of F containing K. When do we
say that L is closed? Let G be the Galois group of F over K. Let H
be a subgroup of G. When do we say that H is closed?
Definition. The Galois group of F over K is the set
Aut
K
F = Aut(F) Hom
K
(F, F)
under the operation of function composition.
PART C 97
The Galois group will be denoted G from now on.
Recall that, for the intermediate eld K L F
L
= G : (u) = u u L = Aut
L
F
L
= v F : (v) = v L
.
In general L L
but not conversely.

Definition. The intermediate eld L is referred to as closed when
L = L
.
In light of the remark made before the denition, to show that L is
closed it suces to demonstrate that L
L.
The priming operation for subgroups H < G is similar
H
= u F : (u) = u H
H
= G : (u) = (u) u H
= Aut
H
F.
In general H < H
but not conversely.

Definition. The subgroup H is referred to as closed when H =
H
.
As before, to prove that H is closed it is enough to verify that
H
< H.
(b) Prove that if H is a subgroup of G then its xed eld is closed.
Proof. It must be shown that H
= (H
= H
. The denition
of H
is
H
= v F : (v) = v H
.
Certainly H
. Select v H
and H. Since H H
it
follows that H
. But then (v) = v by the denition of H
. This
means v H
; hence H
< H
which shows that H
is closed.
2004 C4.
Suppose that K is the eld of rational numbers and let f be an
irreducible polynomial of degree 3 and discriminant D. Suppose D ,= 0.
Prove that D > 0 if and only if f has three real roots. What conclusion
can you draw if D > 0?
Proof. It was not stated, but shall be assumed that f K[x].
Suppose that f factors in some splitting eld over F K as
f(x) = (x u
1
)(x u
2
)(x u
3
) F[x].
Recall that
D = (u
1
u
2
)
2
(u
1
u
3
)
2
(u
2
u
3
)
2
.
98 2004 QUAL
Since the coecients of f are rational, it follows that f has either
exactly three real roots or one real root and two complex roots (which
occur as a conjugate pair).
First it will be proven that if f has three real roots then D > 0. If
u
i
1 for all i, then each multiplicand in the expression dening D is
non-negative. In fact, since K is separable the roots of f are distinct,
so each term is positive. Accordingly D > 0. By contraposition,D < 0
implies f does not have three real roots. As noted earlier, the latter
means that f has one real and two complex roots (conjugate pair).
To complete the characterization it will be shown that if f has one
real and two complex roots then D < 0. Suppose that u
1
1 and
u
2
= a +bi, u
3
= a bi C. The formula for D is
D = (u
1
a bi)
2
(u
1
a +bi)
2
(a + bi a +bi)
2
,
which when expanded shows that D < 0. By contraposition, if D > 0
then f has three real roots.
2004 C6.
Dene a cyclotomic extension of order n of a eld K. Dene the
nth cyclotomic polynomial over K. Suppose that F
8
is a cyclotomic
extension of order 8 over the eld of rational numbers. Determine
the Galois group of F
8
over .
Definition. A cyclotomic extension of order n of a eld K is a
splitting eld F
n
K of x
n
1 K[x].
Definition. The nth cyclotomic polynomial over K is
n
(x) =
(n)
j=1
(x
j
)
where the product is extended over all distinct primitive nth roots of
unity
j
in some xed algebraic closure of K.
As described, F
8
is a splitting eld of x
8
1 over the rationals. It
is a theorem that
x
8
1 =
d|8
d
(x).
The set R F
8
of roots of x
8
1 is exactly the set of 8th roots of
unity, so F
8
= (R). If is a primitive 8th root of unity (i.e. a root
of
8
) then = R, and so
F
8
= (R) = ().
The Galois group Aut
F
8
will be denoted G.
PART C 99
The irreducible polynomial of over is
8
, which has degree
deg
8
= (8) = 4. Since the eld of rationals has characteristic zero,
the algebraic extension F
8
is Galois. According to the Fundamen-
tal Theorem of Finite Dimensional Galois Theory
[G[ = [F
8
: ] = [() : ] = deg
8
= 4.
Each G is completely and uniquely determined by its action () =
i
for some i 0, . . . , 7. Similarly, () =
j
and since
1
= 1 =
1
it follows that i is a unit modulo 8.
The map dened by
G i Z
8
is easily seen to be a monomorphism of groups, so G is isomorphic to
a subgroup of the units modulo 8. But [G[ = [Z
8
[, so it must be the
case that G is isomorphic to the group of units modulo 8 itself. Note
that
G Z
8
Z
2
Z
2
.
2004 Qual, version 2 (October)
Part A
2004v2 A1 unnished.
2004v2 A2.
(a) Solvable quotient implies solvable group. Let G be a nite group
and H a normal subgroup of G. Show that if H and G/H are solvable,
then G is solvable also.
Proof. Let

G denote G/N. By hypothesis there are solvable series
e = N
0
N
1
. . . N
n
= N
e =

G
0

G
1
. . .

G
m
=

G.
By the standard theorem characterizing the subgroups of a quotient,
for each i = 1, . . . m there exists a subgroup G
i
of G containing N such
that
G
i
= G
i
/N,
so actually
e = G
1
/N G
2
/N . . . G
m
/N = G/N.
By the Third Isomorphism Theorem,
G
i+1
/

G
i
=
_
G
i+1
/N
_
_
_
G
i
/N
_
G
i+1
/G
i
.
Accordingly, the factors in the penultimate display are abelian. Putting
the two series together gives
e N
1
. . . N
n
= N = G
0
G
1
. . . G
m
= G,
which is a solvable series in G.
(b) Solvable factors implies solvable product. If H, K are solvable
normal subgroups of G with HK = G and H K = e, then G is
solvable.
Lemma. If G H K, then G/H K.
101
102 2004 QUAL, VERSION 2 (OCTOBER)
Proof. Consider the canonical projection homomorphism : H
K K. Note im = K, ker H. There is also the canonical
injection homomorphism : H H K, with im H. Observe
im = ker . By the rst isomorphism theorem,
H K
/
K
HK
ker
/
im
i.e. H K/ ker im . Now, H K G and ker = im H
implies
H K/ ker G/H
and since im = K, it follows that G/H K.
Proof of (b). The standard theorem on internal direct products
indicates that G HK, so it suces to prove that HK is solvable.
Using part (a)
G here plays the role of G in (a)
H here plays the role of H in (a), solvable
K here plays the role of G/H in (a), solvable
hence G is solvable.
2004v2 A3. Let S be a multiplicative subset of the commutative
ring R and let I be an ideal of R. Show that S
1
Rad(I) = Rad(S
1
I),
where Rad(J) = x R : x
n
J for some positive integer n.
Proof. Verify that S
1
Rad(I) Rad(S
1
I).
Conversely, choose
y
t
Rad(S
1
I). Then
y
n
t
n
= P
x
s
for some x
I, s S, n 1. Therefore there exists a
S such that s
(y
n
s xt
n
) =
0, that is to say s
y
n
s = s
xt
n
I. Since I is an ideal we now have
(s
)
n
y
n
s
n
I so that s
ys Rad(I).
Let z s
ys and t
= s
st. Then tzt = tyt
and hence t(yt
zt) =?,
so that
y
t
=
z
t
, where z Rad(I).
2004v2 A4. Duplicates 1995 A3.
2004v2 A5. Let G be a nite simple group having a subgroup H
of index n. Show that G is isomorphic to a subgroup of S
n
.
Remark. This is similar to [2002 A3(a)].
Proof. Using [G : H] = n, write
G/H = a
i
H : i = 1, . . . , n.
PART B 103
The group G acts on the set G/H by left multiplication,
g a
i
H = (ga
i
)H,
note (ga
i
)H = a
j
H for some j 1, . . . , n. This action induces a
group homomoprhism : G S
n
whereby
(g)(i) = j g a
i
H = a
j
H.
Recall from undergraduate algebra that ker G. As G is simple,
either ker = G xor ker = e. Certainly is not the trivial homo-
morphism, unless G is the trivial group (in which case there is nothing
to prove). By the First Isomorphism Theorem
G
/
S
n
G/e

/
im
i.e. G G/e im < S
n
.
Part B
(1) Decompose the following as a direct sum of cyclic groups
(a) Z
12
Z
19
(b) Hom(Z
12
, Z
9
)
(2) Let A, B be nitely generated abelian groups. Prove that
Hom(A, B) is nitely generated.
Lemma. Z
n
Z
m

= Z
gcd(m,n)
Proof. See JMDs solution for 2009 B3, or do it NMs way: First
show that for any abelian group A and all m > 0, AZ
m

= A/mA. Use
basic properties of . Next use cyclic group theory to show Z
n
/mZ
n

=
Z
gcd(m,n)
.
Lemma. Hom
Z
(Z
m
, Z
n
)

= Z
gcd(m,n)
.
Proof. Hungerford IV.4.Ex1(a,b). Use NMs method.
1(a). The greatest common divisor of 9 and 12 is 3, so Z
12
Z
9

= Z
3
by the Lemma.
1(b). By the second Lemma, Hom(Z
12
, Z
9
)

= Z
3
.
2. Suppose A is generated by x
1
, . . . , x
n
. All elements of A have
the form x
m
1
1
. . . x
mn
n
where m
i
Z. Dene a homomorphism Z
n
A
of abelian groups by extending e
i
x
i
to be Z-linear, where e
i
is the
ith standard basis element of Z
n
. This msp is surjective. Let K = ker
this map, so there is an exact sequence
0 K Z
n
A 0.
Apply the left exact contravariant functor Hom
Z
(, B)
0 Hom(A, B) Hom(Z
n
, B) Hom(K, B).
This shows that Hom(A, B) is a submodule of Hom(Z
n
, B); if we can
show that the latter is nitely generated then since the underlying ring
Z is Noetherian it will follow that Hom(A, B) is nitely generated.
Dene a homomorphism of abelian groups Hom(Z
n
, B) B
n
by
extending ((e
1
), . . . , (e
n
)) to be Z-linear. Verify that this is
an isomorphism, using the structure theorem Hungerford II.2.1 to get
surjectivity. Since B is nitely generated, so too is B
n
and Hom(Z
n
, B).
Part C
PLEASE CIRCLE THE Four PROBLEMS YOU WANT TO BE
GRADED.
2004 C1.
Let F be the splitting eld over of the polynomial x
4
4x
2
1.
Let g(x) = x
3
+ 6x
2
12x 12. Does g(x) have a root in F? Prove
your answer.
Lemma (Eisensteins Criterion). For a unique factorization D and
its quotient eld F, if
f(x) =
deg f
i=0
a
i
x
i
D[x]
with deg f > 1 and p D is irreducible such that p ,[a
deg f
, but p[a
i
for
1 i < deg f and p
2
,[a
0
, then f is irreducible over F.
Theorem (Rational Root Theorem). If r/s is a root of the
polynomial
f(x) =
deg f
i=0
a
i
x
i
[x]
then s[a
deg f
and r[a
0
.
Solution. Eisensteins criterion (D = Z, p = 3) indicates that g
is irreducible over . In particular, g has no roots in . The rational
root theorem implies that if r/s is a root of f, then r/s = 1.
PART C 105
Neither of 1 are roots of f, so f has no rational roots and hence any
factorization of f in [x] into non-units
f(x) = h(x)k(x)
must satisfy deg h = 2 = deg k.
Exploring the purported factorization f = hk further, since f is
monic with constant term 1 it must be the case that
h(x)k(x) = (x
2
+ ax + 1)(x
2
+ bx 1)
for some a, b . Expanding the right-hand side produces the system
of equations
b +a = 0
ab = 4
b a = 0
which is inconsistent. Consequently, there exists no such factorization
of f. At this point all possible factorizations of f in [x] into non-units
have been shown to be impossible, and it follows that f is irreducible
over .
Substituting y = x
2
, solving for y using the quadratic formula, and
re-substituting shows that the roots of f are
_
2 +
5,
_
2 +
5,
_
2
5,
_
2
5.
Note that
F =
__
2 +
5,
_
2
5
_
has degree 4 over . The roots of f are distinct, which means that f
and also F are separable over . A standard theorem (Hungerford V
3.11) indicates that F is algebraic and Galois over . Another theorem
(Hungerford V 3.14) shows that F is normal over .
If g has a root u F, then g must split in F and accordingly the
splitting eld E of g over is contained in F. Now [E : ] divides
3! = 6. But
[F : Q] = [F : E] [E : ]
implies that 6[4, which is false. Therefore g has no root in F.
2004v2 C3.
Let F
12
be a cyclotomic extension of of order 12. Determine
Aut
F
12
and all intermediate elds. The eld F
12
is a splitting eld of
x
12
1 over the rationals. If denotes a primitive twelfth root of unity,
then generates set of all twelfth roots of unity. The latter is exactly
the set of roots of x
12
1, so
F
12
= ().
The extension F
12
is algebraic and consequently, since char = 0,
is Galois. Let G denote Aut
F
12
. The Fundamental Theorem of Finite-
Dimensional Galois Theory indicates that F
12
is Galois over every inter-
mediate eld E, and there is a one-to-one correspondence between in-
termediate elds E and subgroups H of G given by F
12
E E
G.
Furthermore, if L M F
12
then [M : L] = [L
: M
] and, in
particular,
[G[ = [() : ] = (12) = 4.
Any G is completely and uniquely determined by the image
() =
i
for some i 0, . . . , 11. Now
1
() =
j
similarly, and
since
1
= 1
F
12
=
1
it follows that i is a unit in Z
12
. The function
dened by
G i Z
12
is evidently a monomorhpism of groups, which places G in isomorphism
with a subgroup of Z
12
. But the group of units modulo 12 has order
(12) = 4, which is exactly the order of G, so it must be the case that
G Z
12
. Note that each element of Z
12
has order 2, so that
Z
12
Z
2
Z
2
.
The only proper subgroups of the latter are 0 Z
2
and Z
2
0,
hence the only intermediate elds are those given by the Galois corre-
spondence, which I do not care to write down in this document.
2004v2 C4.
Construct a eld with 49 elements and give the rules for its addition
and multiplication. If a is a generator, what is the multiplicative inverse
of 1 + a in terms of your set of minimal generators?
Proof. The ground will be K = Z
7
. Consider f(x) = x
2
3
K[x]. Since deg f = 2 it suces to observe that f has no roots in K
to conclude that f is irreducible over K. Thus in K[x] the ideal (f) is
maximal, and so
K[x]
(f)
= r
0
+r
1
x + (f) : r
i
K
PART C 107
is a eld with 49 elements. Note also that this eld is isomorphic to
F = r
0
+r
1
a : r
i
K, a
2
3 = 0.
In F, multiplication and addition is done according to the rule a
2
= 3.
The extension I will use is K F.
After experimenting with a few cases, one is led to
(1 + a)(3 + 4a) = 3 + 4a + 3a + 4a
2
= 3 + 12 + (4 + 3)a = 1.
2003 Qual
Part A
2003 A1. A group cannot be the union of two proper subgroups.
Proof. Let G be a group and suppose H, K < G satisfy H K =
G. Choose k G H, ( k K). For any h H, the element
hk / H. As H K = G, it must be that hk K. But k
1
K, so
(hk)k
1
= h K. This shows G K, so G = K.
2003 A2 unnished.
2003 A3. Let G
1
and G
2
be two non-trivial non-isomorphic simple
groups. Prove that any proper trivial normal subgroup of G
1
G
2
coincides with G
1
or G
2
.
Lemma. If : G H is a group homomorphism and N H,
then
1
(N) G.
Proof. The pullback of N is dened as
1
(N) = a G : (i) N.
Let a
1
(N) and g G. Then
(gag
1
) = (g)(a)(g
1
) = (g)(a)(g)
1
N
so gag
1

1
(N). As a, g were chosen arbitrarily it follows that
g
1
(N)g
1

1
(N)
for all g G, which is what it means for
1
(N) G.
Proof of 2003 A3. Let N G
1
G
2
. There are canonical injec-
tions (monomorphisms of groups)
i
: G
i
G
1
G
2
for i = 1, 2. By the
lemma,
1
i
(N) G
i
for each i. As G
i
are simple and non-isomorphic,
exactly one of the following cases holds:
(1)
1
1
(N) = G
1
and
1
2
(N) = e
G
2
(2)
1
1
(N) = e
G
1
and
1
2
(N) = G
2
109
110 2003 QUAL
Since G
1
G
2
G
2
G
1
, assume without loss of generality that (1)
is true. Then
N =
1
1
1
(N) =
1
(G
1
) = G
1
e
G
2
G
1
.
2003 A4. A free abelian group is either trivial or isomorphic to Z.
Definition. A group Gis free on a set X when there is an injective
map of sets : X A and for any group H and map of sets f : X H
there exists a unique group homomorphism

f : G H such that
f = f.
G
!

f
A
A
A
A
A
A
A
X
O
f
/
H
Proof. Let A be abelian and free on X. If X = then A is trivial
and there is nothing to prove.
If X = x is a singleton, then every reduced word in A has the
form x
n
for some n Z. Dene a map A x
n
n Z and verify
that it is an isomorphism of groups.
If [X[ 2, then there exist distinct elements x, y X. The group
A contains the reduced word
1 ,= xyx
1
y
1
A
and it follows that
xy ,= yx
whence A is not abelian. This is a contradiction to the hypothesis,
so if A is to be abelian then [X[ 0, 1, i.e. A is either trivial or
isomorphic to Z.
2003 A5. Describe up to isomorphism all groups of order 121.
Prove your answer.
Solution. Since 121 = 11
2
and 11 is prime, G is abelian. By the
Fundamental Theorem of Finite Abelian Groups,
G A Z
121
, Z
11
Z
11
For the proof that [G[ = p

2
implies G abelian, see [1995 A5].
PART B 111
2003 A6. Prove that if S is a Sylow subgroup of a nite group G,
then N
G
(N
G
(S)) = N
G
(S).
Proof. Note that S is a Sylow subgroup of any subgroup K < G
containing S. By Sylows second theorem, any conjugate of S is a
Sylow subgroup of G. If S
is a conjugate of S contained in a subgroup

K < G, then S
is a Sylow subgroup of K as well. In particular, we are

thinking of K = N
G
(S); here S N
G
(S) implies (by Sylows second
theorem again) that S is the only Sylow subgroup of N
G
(S).
Given x N
G
(N
G
(S)) it must be shown that x N
G
(S). For this
it suces to show that xSx
1
= S. By the denition of x
xN
G
(S)x
1
= N
G
(S).
Since S < N
G
(S) it follows that
xSx
1
< xN
G
(S)x
1
= N
G
(S).
So xSx
1
is another Sylow subgroup of N
G
(S), which necessitates
xSx
1
= S. Thus N
G
((N
G
(S)) < N
G
(S). The other containment
is obvious, hence equality holds.
Part B
2003 B1 unnished.
2003 B2 unnished.
2003 B3. Let R be a commutative ring with unique maximal ideal
M. Let A be the smallest subring of R containing the multiplicative
identity 1 of R. Show that A is ring isomorphic to either Z or Z/p
n
Z
for some p, n N with p prime. (Hint: consider the idempotents of R.)
Proof. Dene : Z A by (1) = 1
R
, and extend to be a
ring homomorphism. Because of As denition (Z) = A. The rst
isomorphism theorem for rings yields Z/ ker

= A. If is injective
then A

= Z, and if is not injecive then since the only ideals of Z
are mZ (m Z) we have A

= Z/mZ. It remains to show that m = p
n
for some prime p and some n 0. For simplicity let us identify A and
Z/mZ.
Idea: let m = p
n
1
1
p
nr
r
. If r > 1 then (p
1
), (p
2
) are maximal ideals
in A. End of rigor. Do something like this: there exist s
1
, . . . , s
r
Z
such that s
1
p
1
+ + s
r
p
r
= 1. Therefore (not being careful) (p
1
) +
+ (p
r
) = R. Whatever.
2003 B4 unnished.
112 2003 QUAL
2003 B5. Let
0 A

B

C 0
be an exact sequence of R-modules.
(a) What part of the sequence
0 Hom
R
(C, D)

Hom
R
(B, D)

Hom
R
(A, D) 0
is exact for every left R-module D?
Answer. The answer is the rst part
0 Hom
R
(C, D)

Hom
R
(B, D).
Given that im = C, it must be checked that if , : C D satisfy
() = () then = . The new assumption can be reworded as
= , i.e. for all b B we have (b) = (b). As b
runs through all elements of B, im = C guarantees that (b) runs
through all elements of C; it follows that = .
(b) Show that the remaining part of the new sequence is exact if and
only if the original sequence is split exact. The remaining part of the
new sequence is the end
Hom
R
(B, D)

Hom
R
(A, D) 0.
Given that is injective, it must be checked that for all f : A D
there exists g : B D such that f = g . A priori there is no reason
why this should be true.
Proof. Assume that the original sequence is split exact; in partic-
ular assume there exists : B A such that = 1
A
. Given f as
above, observe that g = f does the job.
Conversely, assume that
0 Hom
R
(C, D)

Hom
R
(B, D)
is exact for all R-modules D. Specialize to the case D = A and consider
f = 1
A
.
2003 B6. Let K be a eld and let M and N be nitely generated

modules over the polynomial ring K[x]. Suppose that N has invariant
factors (x 2)(x 3)
2
, (x 2)(x 3)
2
(x 5) and M has invariant
factors (x 1), (x 1)(x 2)
2
, (x 1)(x 2)
2
(x 3).
PART C 113
(a) Give the elementary divisors of M N. Recall that K[x] is a
principal ideal domain (if I is an ideal of K[x] choose f I of minimal
degree, then I = (f) using the division algorithm). The structure
theorem IV 6.12 indicates that
M
K[x]
(x 1)

K[x]
(x 1)(x 2)
2

K[x]
(x 1)(x 2)
2
(x 3)
N
K[x]
(x 2)(x 3)
2

K[x]
(x 2)(x 3)
2
(x 5)
.
Elementary divisors are the prime-power divisors of the invariant fac-
tors. Accordingly the elementary divisors of M are
x 1, x 1, (x 2)
2
, x 1, (x 2)
2
, x 3
and those of N are
x 2, (x 3)
2
, x 2, (x 3)
2
, x 5.
The elementary divisors of M N are given by concatenating these
lists.
(b) Give the invariant factors of M N. Using the power grid
on the elementary divisors gives
(x 2)
(x 2)
2
(x 1)(x 3)
(x 2)
2
(x 1)(x 3)
2
(x 2)
2
(x 1)(x 3)
2
(x 5)
as the invariant factors of M N.
Part C
2003 C2. Let F be a eld extension of K.
Dene the Galois group G of F over K.
Definition. The Galois group is
G = Aut
K
F := Aut(F) Hom
K
(F, F).
Now dene what one means by a stable subeld of this extension.
Definition. An intermediate eld F E K is stable whenever
im [
E
= E for all G.
114 2003 QUAL
Prove that if E is a stable intermediate eld, then the Galois group
of F over E is a normal subgroup of G.
Proof. Certainly Aut
E
F G. Let G and Aut
E
F. Given
u E, since E is stable it is true that im [
E
= E = im
1
[
E
. By
choice [
E
= 1
E
. So
1
(u) =
1
(e
) =
1
(e
) = e
E.
Conversely, prove that if H is a normal subgroup of G, then the
xed eld of H is a stable subeld of the extension F K.
Proof. For all H and any G there exists
H such
that =
. Given
u H
= u F : (u) = u H
observe that
(u) =
(u) = (u)
so (u) H
.
2003 C3. Let F K be elds.
(i) Dene the maximal algebraic extension of K in F.
Definition. The maximal algebraic extension of K in F is the in-
termediate eld K A F that is algebraic over K and not property
contained in any other algebraic extension of K which is contained in
F. In other words, if
K A E F
and E is algebraic over K, then E = A.
Such an extension always exists by Zorns lemma.
(ii) Suppose that for every extension F K, the maximal algebraic
extension of K in F is K. Prove that K is algebraically closed.
Proof. Let K be an algebraic closure of K. In particular K is
algebraic over K. In the extension K K, then, A = K. But A = K
by hypothesis, so K = K is algebraically closed.
(iii) Now suppose K is algebraically closed and let F K. Prove
that the maximal algebraic extension of K in F is K.
Lemma. K is an algebraic closure of K if and only if for all alge-
braic extensions E K there exists a K-monomorphism E K.
This is proved after the proof of (iii).
PART C 115
Proof. Here the setup is
K = K A F.
But, by the lemma, K is an algebraic closure of itself if and only if
for all algebraic extensions E K there exists a K-automorphism
E K. In particular there exists a K-monomorphism A K, and it
follows that A = K.
Proof of the lemma. () K is algebraic over K by denition,
suppose that E is also algebraic over K. If E is algebraically closed,
then K
K
E and the proof halts. If not, then let E be an algebraic
closure of E. Since E E K and E K is algebraic it follows that
E is an algebraic closure of K. Thus there exists a K-isomorphism
: E
K
K, which gives a K-monomorphism [
E
: E K.
() Assume K K is algebraic and let E be an algebraic closure
of K. Note that E is an algebraic closure of K. By hypothesis there
is a K-monomorphism : E K, but since K E we have the
inclusion map K E. It follows that K = E.
2002 Qual
Part A
2002 A1. Show that if a positive integer d divides the order n of
a nite cyclic group G, then G has a unique subgroup of order d.
Proof. Say G = a. Ill show that a
n/d
is the one and only
subgroup of order d. From undergraduate group theory
[a
n/d
[ =
n
(n, n/d)
=
n
n/d
= d,
so [a
n/d
[ = d. Suppose H < G has order d. All subgroups of a cyclic
group are cyclic, so H = a
m
where m is the least positive integer
such that a
m
H. Also note that m[n, so (m, n) = m and
d = [a
m
[ = [a
(m,n)
[ =
n
(m, n)
=
n
m
which implies m = n/d. Thus H = a
n/d
.
2002 A2. Can there be a simple group of order 200?
Proof. Let G be a group of order 200 = 2
3
5
2
. Sylows third
theorem gives
n
2
1 (mod 2), n
5
1 (mod 5)
and a well-known theorem indicates that
n
2
[25, n
2
[8.
Accordingly, n
5
= 1. Say P is the unique Sylow 5-subgroup of G.
Sylows second theorem states that
Syl
5
(G) = xPx
1
: x G
but it was just shown that xPx
1
= P for all x G. Hence P G,
and the latter cannot be simple.
2002 A3.
117
118 2002 QUAL
(a) If H is a subgroup of G and [G : H] = n, then H contains a
normal subgroup K G and [G : K] divides n!.
Proof. The group G acts on the set
G/H = a
i
H : i = 1, . . . , n
by left multiplication
g a
i
H = (ga
i
)H = a
j
H.
Dene a group homomorphism : G S
n
by
(i) = j (ga
i
)H = a
j
H.
The kernel K = ker is a normal subgroup of G, and
g K i : (ga
i
)H = a
i
H.
Taking a
i
= e gives g K implies gH = H, which is to say g H.
Thus K < H, as desired.
By the rst isomorphism theorem
G

/
proj
S
n
G/K

/
im
the group G/K is isomorphic to im < S
n
. Therefore, by Lagranges
theorem,
[G : K] = [G/K[ = [im [
n!
(b) If G is nite and n = p is the least prime dividing [G[, then H
is normal.
Proof. Let : G S
p
and K = ker be as before, and denote
[H : K] = k so that
[G : K] = [G : H][H : K] = pk.
Since G/K im , pk[p! and hence k[(p 1)!. All prime divisors of
(p 1)! are less than p, while all prime divisors of k are greater than
or equal to p (since p is the smallest prime dividing [G[). This forces
k = 1, and consequently H = K G.
PART B 119
2002 A4. Show that each nite p-group is solvable.
Definition. A sequence of subgroups
G = G
0
> G
1
> > G
t
= e
is a solvable series provided that
(1) G
i
G
i+1
(2) G
i
/G
i+1
is abelian.
Theorem (II 8.5). G is solvable if and only if G has a solvable
series.
Proof. If n = 0 then G is trivial and there is nothing to prove.
For n 1, Sylows rst theorem gives subgroups G
i
of G with order
p
i
for i = 1, . . . , n, each of which is normal in some subgroup of order
p
i+1
. Putting these together, there is a sequence of subgroups
e G
1
G
2
. . . G
n1
G
n
.
Certainly G
n
= G, since [G
n
[ = p
n
= [G[. It is also clear that
G
i+1
/G
i
= p
(i+1)i
= p
whence the factors of this series are cyclic, hence abelian. Summarily,
there exists a solvable series in G.
2002 A6. Let R be a commutative unitary ring. An ideal I or R
is said to be primary if a, b R and ab I imply a I or b
n
I for
some n N. Assume I is primary and S is a multiplicative subset of
R such that S I = . Show that S
1
I is a primary ideal of S
1
R.
Proof. Let
a
s
b
t
S
1
I. In other words
ab
st
=
i
u
for some i I, u
S. This means there exists v S such that v(abu ist) = 0. That is
to say (ab)(vu) I, and so either ab I or (vu)
n
I for some n (I
is primary). Well if (vu)
n
I then (vu)
n
I S which contradicts
something. So it must be the case that ab I. Again either a I or
b
m
I for some m (I is primary). If a I then
a
s
S
1
I. If b
m
I
then (
b
t
)
m
=
b
m
t
m
S
1
I, which completes the proof.
Part B
2002 B4.
(i) Duplicates 2007 B1(a).
120 2002 QUAL
(ii) True or false: If 0 A B C 0 is exact, then A
R
B
A
R
C implies B C. The answer to this is an emphatic no. For
example,
Z

Z
Z but , Z as abelian groups.
Proof that
Z

Z
Z. The sequence
0 Z

/Z 0
is exact. Apply the left-exact covariant functor
Z
, and write
for
Z
to get a new exact sequence
0 Z
1
Q
1
Q
_
/Z
_
.
Facts:
(1) Z
This is true by virtue of the more general fact that R
R
M M. Dene a middle-linear map by (n, p/q) np/q. The
universal property gives a unique homomorphism : Z
agreeing with the middle-linear map on the simple tensors.
has inverse : p/q 1 p/q, whence is injective. It is
clearly surjective.
(2)
_
/Z
_
0
All simple tensors do this
_
p/q +Z
_
r/s =
_
p/q +Z
_
qr/qs
=
_
pq/q +Z
_
r/qs
= 0 r/qs = 0.
Composing with the isomorphisms gives
0
(1
Q
)

(1
Q
)
0,
still exact. Now
im ( 1
) = ker( 1
) = .
So ( 1
) : .
Summarily
Z

Z
Z.
2002 B5. True or false?
(i) A torsion-free module is projective. FALSE is torsion-free
(since it is a domain) as a Z-module, but it is not a projective Z-module.
(ii) A projective module is torsion-free. FALSE Use an R that is
not a domain. As a module for itself, R = Z
4
is free on 1. Obviously
R is not torsion-free since ann
R
(2) = (2) _ 0.
PART C 121
(iii) A torsion-free module is free. FALSE As a Z-module, is
torsion free but has no basis. This assertion is easy to show using least
common multiples and clever cancellation: no two elements of are
linearly independent over Z, and , Z so any purported basis of
must have size at least 2.
2002 B6. True or false?
(i) A free module is torsion-free. FALSE As a module for itself,
R = Z
4
is free on 1, but the annihilator of 2 R is nonzero.
(ii) A projective module is free. FALSE Recall from undergrad-
uate group theory that Z
6
Z
2
Z
3
. Since Z
6
is free on 1 as a
module for itself it follows that Z
2
is a projective Z
6
-module. But Z
2
has no basis over Z
6
(easy to verify).
(iii) A free module is projective. TRUE Let P be free on X. Given
a diagram
X

/
P
A
f
/
B
/
0
dene a map of sets g : X A by f(x) = a f
1
((x)), well-dened
because f is surjective. The universal property of free gives a unique
homomorphism g : P A agreeing with g on X.
X

/
g
! g
.~~
~
~
~
A
f
/
B
/
0
f g(x) = (x) x X
That f g = is true by virtue of the setup.
Part C
2002 C1.
(i) Prove that if f K[x] is a polynomial of degree n, then there
exists an extension of K in which f has a root.
Proof. Let p K[x] be an irreducible factor of f, say of degree
r; it suces to construct an extension of K in which p has a root. The
ideal (p) is maximal in K[x], so the quotient
F =
K[x]
(p)
=
_
a
0
+ a
1
x +. . . a
r1
x
r1
+ (p) : a
i
K
_
is a eld. Identifying K with its image under the monomorphism a
a+(p), it can be assumed that F is an extension of K. Let u = x+(p).
122 2002 QUAL
Then, writing p(x) =
r
i=1
b
i
x
i
, the ith term in the expression of p(u)
has the form
b
i
_
i
j=1
_
i
j
_
x
i
(p)
ij
_
= b
i
x
i
+ (p).
Thus, p(u) = p(x) + (p) = 0 + (p).
(ii) Consider the example f(x) = x
3
5x2 [x], let u be a real
root of this polynomial. What is the natural basis of (u) and write
the element x
4
3x + 1 as a linear combination of the basis elements.
Solution. As a vector space over ,
(u) = a
0
+a
1
u +a
2
u
2
: a
i
, f(u) = 0.
Polynomial division gives
x
4
3x + 1 = xf(x) + 5x
2
x + 1,
so that in (u),
u
4
3u + 1 = 5u
2
u + 1.
2002 C3.
F is an algebraic closure of K if and only if F is algebraic over
K and for any algebraic eld extension E
1
of another eld K
1
and
isomorphism of elds : K
1
K, extends to a monomorphism
E
1
F.
Theorem. Let : A L be an isomorphism of elds, S = f
i
:
i I A[x] a set of polynomials of positive degree, and S
= f
i
:
i I L[x]. If B is a splitting eld of S over A and M is a splitting
eld of S
over L, then extends to an isomorphism M L.

Proof of 2002 C3. () It needs only to be shown that F is
algebraically closed over K. Let E be an algebraic closure of F, K
F E. Since E is algebraic and algebraically closed over F it is also
algebraic and algebraically closed over K. With = 1
K
, by hypothesis
there exists a monomorphism : E F, so actually E = F because
inclusion F E is a monomorphism. So F is algebraically closed over
K.
() Use the Theorem with A = K, L = K
1
, S = the irreducible
elements of K[x], S
= the irreducible elements of K

1
[x], B = F, and
M = some algebraic closure of E
1
. Note that M is an algebraic closure
of K
1
as well. The theorem provides an isomorphism : M F which
yields a monomorphism
[
E
1
: E
1
F
PART C 123
which extends by denition.
2002 C5. Let F be an algebraic closure of K = Z
p
. Prove that
(a) F is algebraic Galois over K.
Theorem. F K is Galois if and only if F is a splitting eld of
a set of separable polynomials T K[x] of positive degree.
Proof. Since F is (by denition) a splitting eld of the irreducible
polynomials in K[x] it suces to show that each irreducible f K[x]
is separable. If u is a multiple root of f, then f = (x u)
2
g for some
g K[x]. But then f
= (xu)[2g +(xu)g
] so u is also a root of f
.
Since f is irreducible and deg f
< deg f these two polynomials must

be coprime. Say fh + f
k = 1 for some h, k K[x]. But then 0 = 1,

which is no good.
It must be, then, that either f is separable or f
= 0. Assuming
the latter, write f = a
0
+ +x
n
so that
0 = f
= a
1
+ 2a
2
x + + nx
n1
.
Now ia
i
= 0 for i = 1, . . . , n. This means that those i for which a
i
,= 0
are divisible by p, so actually f(x) = r(x
p
) for some r K[x]. But
every element in K is a pth power and K has characteristic p, so
f(x) = r(x
p
) = r(x)
p
shows that f is not irreducible. This is impossible and all other alter-
natives have been exhausted: f must be separable.
(b) The map : F F given by u u
p
is a non-identity K-
automorphism of F.
Proof. The K-monomorphism part is easy. Surjectivity comes
from the fact that F is algebraic over K. Given u F it is algebraic
over K, and so [K(u) : K] = d is nite. In other words K(u) is a
nite eld of cardinality p
d
. Now it follows from the K-monomorphism
part that [
K(u)
is surjective, which means there is v K(u) such that
(v) = u. But v F.
(c) What is the xed eld of the subgroup ?
By denition (write it out)
is the set of u F which are the

roots of x
p
i
x for all i N. Now F
p
i is the splitting eld of said
polynomial over K, so it follows that
iN
F
p
i
which is nothing but K itself.
2001 Qual
Part A
2001 A1.
2001 A2. A group of order 77 has exactly one subgroup of order
11
Proof. Let G be a group with order 77 = 7 11. Sylows theorems
imply n
1
1 1 (mod 7) and n
11
[7. Clearly then, n
7
= 1; i.e. there
exists exactly one Sylow 11-subgroup of G, call it E. A remark on
[2], p.94 says that every 11-subgroup H of G is contained in a Sylow
11-subgroup of G. In other words, each such H is contained in E.
Obviously then, H = E, so E is the unique subgroup of G with order
11. Note that E G by either Sylows second theorem or by the
result which says E G when [G : E] (= 7 in this case) equals the
least prime dividing [G[.
If you arent satised quoting Hungerford, suppose that E, E
are
two subgroups of G of order 11. As groups of prime order, both are
cyclic. Cyclic subgroups of the same order which are unequal have
trivial intersection see 2008 A2(iii) so EE
< G is a subgroup of
order 11 11 = 121. Contradiction!
2001 A3. Prove that if G is a group and G/C(G) is cyclic, then
G is abelian.
Definition. The center of G is
C(G) = c G : a G, ca = ac.
The details of why C(G) G are omitted. The group operation on
G/C(G) is
_
gC(G)
__
hC(G)
_
= (gh)C(G).
Proof. Combine the hypothesis with the group operation described
above to see
G/C(G) =
__
aC(G)
_
i
: i Z
_
=
_
a
i
C(G) : i Z
_
for some generator a G.
125
126 2001 QUAL
Given g, h G, by hypothesis their cosets in G/C(G) have the
forms
gC(G) = a
i
C(G), hC(G) = a
j
C(G).
For the coset of g, this equality implies that for all c C(G) there
exists c
C(G) such that gc = a

i
c
. A similar statement applies to h.

In particular for e C(G) there are c, c
C(G) such that

g = ge = a
i
c, h = he = a
j
c
.
Punchline:
gh = a
i
ca
j
c
= a
i
a
j
cc
c C(G)
= a
i
a
j
c
c c, c
C(G)
= a
j
a
i
c
c associativity
= a
j
c
a
i
c c
C(G)
= hg.
2001 A4. Duplicates 2003 A4
2001 A5. If H is a cyclic normal subgroup of G, then every
subgroup of H is normal in G.
Proof. The setup is K < H G, H = h. Every subgroup of
a cyclic group is cyclic, so there exists k Z such that K = h
k
.
The hypotheses H G and H = h combine to give: for all g G,
ghg
1
H. The goal is to show that for all g G, gh
k
g
1
K.
Given g G arbitrary, say ghg
1
= h
n
(again using H = h).
Then
gh
k
g
1
= g h. . . h
. .
k-fold
g
1
= ghg
1
. . . ghg
1
. .
k-fold
=
_
h
n
_
k
and
_
h
n
_
k
=
_
h
k
_
n
K.
Part B
2001 B1. Typo: should be Noetherian
Let R be a Noetherian integral domain. Show that R is a principal
ideal domain if each nitely generated torsion-free R-module is free.
Proof from pg. Let I be an ideal. Since R is Noetherian I is a
nitely generated ideal. I is torsion free, so by hypothesis I is free.
It is clear that every set of generators for I as an ideal are a spanning
set as a free R-module. That is, I = (a
1
, . . . , a
n
)R = a
1
R + +
PART C 127
a
n
R. Let x
1
, . . . , x
n
be a linearly independent set of generators, so
I = (x
1
, . . . , x
n
)R.
Claim m 1. If not then let x
1
,= x
2
be basis elements. Then
for all r
1
, r
2
R we have r
1
x
1
+ r
2
x
2
= 0 implies r
1
= 0 = r
2
. But
x
1
, x
2
R implies x
2
x
1
x
1
x
2
= 0 and it follows that x
1
= 0 = x
2
,
impossible.
So every nitely generated ideal is principal, I = xR.
2001 B2.
Show that if Hom
R
(D, ) preserves exactness of 0 A B
C 0 for each D, then the sequence splits.
Proof. Call the rst map of the given sequence f and the second
g. The functor Hom
R
(D, ) is covariant and left exact, so the induced
sequence
0 Hom
R
(D, A)
f
Hom
R
(D, B)
g
Hom
R
(D, C)
is exact. Assume that it is also right exact, so that g is surjective.
This means that, given : D C, there exists : D B such that
= g() = g .
Take D = C and = 1
C
to conclude.
2001 B5. Let R be a commutative unitary ring. Show that each
injective R-module is divisible.
Proof. Select y M and a nonzero non-zero divisor r R. We
must nd x M such that rx = y. Begin with the ideal (r) M.
Dene an R-module morphism f : (r) M by extending r y.
Check that this is well-dened: sr = tr implies sy = ty. Because M is
injective there exists g : R M closing the following diagram.
0
/
(r)ubset
s
/
f
R
g
.{v
v
v
v
v
M
Use x = g(1) and verify that it works.
Part C
2001 C2.
(a) Determine the splitting elds and its degree over the rationals
for x
6
4.
128 2001 QUAL
Write f = x
6
4. The roots of f are
_
6
4 exp(in/3) : 0 n 5
_
and the splitting eld of f over the rationals is
F = (
6
4 exp(in/3) : 0 n 5).
This characterization is redundant. Let be a primitive 6th root of
unity, say =
1+i
3
2
. Then F = (
6
4, ) = (
6
4,
3, i) and since
6
4 =
3
2 it follows that
F = (
3
2,
3, i)
and there is a tower
(
3
2) (
3
2,
3) F.
.
Since
3
2 / ) the polynomial x
3
2 is irreducible over , so
[(
3
2) : ] = 3. Since
2 / (
3
2) the polynomial x
2
3 is irreducible
over (
3
2), so [(
3
2,
3) : (
3
2)] = 2. Since i / (
3
2,
3) the
polynomial x
2
+1 is irreducible over (
3
2,
3), so [F : (
3
2,
3)] =
2. Using the tower,
[F : ] = [F : (
3
2,
3)] [(
3
2,
3) : (
3
2)] [(
3
2 : ] = 12.
.
(b) Prove that the polynomial x
p
n
x is separable over Z
p
.
Proof. In Hungerford, separable applies only to irreducible poly-
nomials, which f = x
p
n
x = x(x
p
n
1
1) is not. Re-interpret separable
in this context to mean no multiple roots.
In general, if u is a multiple root of f = (x u)
n
g (n 2) then
f
= n(x u)
n1
g + (x u)
n
g
shows that u is also a root of f
. Here,
f
= 1 because the characteristic of Z

p
is p. Since f
has no roots at
all, it cannot be the case that f has a multiple root.
2001 C3.
For any integer r 1, let F
p
r be a nite eld of cardinality p
r
.
Prove that
F
p
n F
p
m n[m.
.
Proof. () Including the prime subeld in the tower
Z
p
F
p
n F
p
m
observe that
m = [F
p
m : Z
p
] = [F
p
m : F
p
n] [F
p
n : Z
p
] = [F
p
m : F
p
n] n.
PART C 129
() The eld F
p
m is the splitting eld of
f = x
p
m
x Z
p
over Z
p
. Since the characteristic of the ground eld is p, f
= 1
and consequently f has p
m
distinct roots in F
p
m. This means that the
irreducible factors of f are separable, therefore Z
p
F
p
m is a (nite-
dimensional) Galois extension.
Let G = Aut
Zp
F
p
m. By the Fundamental Theorem of Galois The-
ory, [G[ = m and there is a one-to-one correspondence between sub-
groups H < G and intermediate elds Z
p
E F
p
m. Recall that G
is cyclic with generator : u u
p
. In particular, since n[m there is a
unique subgroup, namely
H =
n
< G
of order [H[ = m/n and index [G : H] = n. By the Fundamental
Theorem,
n = [G : H] = [H
: Z
p
]
so H
= F
p
n F
p
m.
2001 C5.
Determine the splitting eld of x
p
x1 over Z
p
and prove explicitly
that the Galois group is cyclic.
Since a
p
= a for all a Z
p
it is clear that f = x
p
x 1 has no
roots in Z
p
. The fact that f
= 1 indicates that f is irreducible and

separable over Z
p
.
If u
0
, . . . , u
p1
denote the roots of f, then a splitting eld is given
by F = Z
p
(u
0
, . . . , u
p1
). It is easy to verify that f(u +i) = 0 for each
i Z
p
, accordingly, u
0
, . . . , u
p1
= u +i : 0 i p 1. It follows
that F = Z
p
(u). The Galois group is G = Aut
Zp
Z
p
(u).
Proof. Any G is completely determined by its action on u:
(u) = u + i, for some i Z
p
.
Certainly the map
G i Z
p
is a monomorphism of groups, from which it follows that G is isomor-
phic to a subgroup of Z
p
. In particular [G[ divides p, consequently G
is either trivial or isomorphic to Z
p
. Certainly G is not trivial! It must
be the case that G Z
p
.
2001 C6.
130 2001 QUAL
Determine the Galois group of (x
2
2)(x
2
3)(x
2
5) over . Call
this polynomial f [x]. The following result will be useful
Theorem (V 3.8). Let : K L be an isomorphism of elds, S =
f
i
a set of polynomials (of positive degree) in K[x], and S
= f
i
the corresponding set of polynomials in L[x]. If F is a splitting eld of

S over K and M is a splitting eld of S
over L, then is extendible

to an isomorphism : F M.
Eisensteins criterion (p = 2) indicates that x
2
2 is irreducible over
the rationals. The latter has characteristic zero, so this polynomial is
also separable and its splitting eld is Galois over . But the splitting
eld of x
2
2 over is just (
2). Each element of Aut
2) must
send a root of x
2
2 to another root of x
2
2. Accordingly,
Aut
2) = 1
(
2)
,
1
:
2.
It is straightforward to verify that
3 / (
2), so x
2
3 is irre-
ducible over (
2). Certainly then, the splitting eld of x

2
3 over
(
2) (or equivalently the splitting eld of (x

2
3)(x
2
2) over )
is (
2,
3). Before invoking the theorem, note that the setup is

as follows: K = L = (
2), = 1
(
2)
,
1
, S = S
= x
2
3,
and F = M = (
2,
3). Now by the theorem, each element of

Aut
2) extends to a -automorphism of (
2,
3). Since each

such map must send roots of x
2
3 to roots of x
2
3, it is clear that
the only options are
1
(
2)
= 1
(
2,
3)
1
(
2)
=
2
:
1
=
3
:
1
=
4
:
3
This process yields
Aut
(
2)
(
2,
3) = 1
(
2,
3),
2
,
3
,
4
.
which is comprised of three elements of order 2 and the identity.
To nish the problem the process in the previous paragraph must
be iterated. It is straightforward to verify that
5 / (
2,
3), so
x
2
5 is irreducible over (
2,
3). Certainly then, the splitting eld

of x
2
5 over (
2,
3) (or equivalently the splitting eld of (x

2
2)(x
2
3)(x
2
5) over ) is (
2,
3,
5). The setup for the theorem

is as follows: K = L = (
2,
3), = 1
(
2,
3)
,
2
,
3
,
4
, S = S
=
x
2
5, F = M = (
2,
3,
5). Now by the theorem, each element

PART C 131
of Aut
(
2)
(
2,
3) extends to an element of Aut
2,
3,
5).
Each such map must send roots of x
2
5 to roots of x
2
5, it is clear
that the only options are
1
(
2,
3)
= 1
(
2,
3,
5)
1
(
2,
3)
=
5
:
2
=
6
:
2
=
7
:
3
=
8
:
3
=
9
:
4
=
10
:
4
=
11
:
5.
So, nally, the Galois group of f over is
Aut
2,
3,
5) = 1
(
2,
3,
5)
,
5
,
6
,
7
,
8
,
9
,
10
,
11
.
This group is comprised of seven elements of order two and the identity,
so it is isomorphic to Z
2
Z
2
Z
2
.
Let F be the splitting eld of this polynomial. Determine all
subelds L F which are Galois over . The Galois group (Z
2
)
3
is
abelian, so all subgroups are normal. In the Galois correspondence this
translates to all intermediate elds are stable and Galois over . There
are 11 subgroups of the Galois group, seven each isomorphic to Z
2
Z
2
and Z
2
. Their xed elds are the intermediate elds asked for in the
prompt.
2000 Qual
Part A
2000 A1 unnished.
2000 A2 unnished.
2000 A3 unnished.
2000 A4. True or False: There exists G non-abelian with [G[ =
125, [C(G)[ = 25.
False!
Proof. If [G[ = 125 = 5
3
and [C(G)[ = 125 = 5
2
, then [G/C(G)[ =
5. As a group of prime order, G/C(G) is cyclic. By the G/Z theorem
G is abelian.
Part B
2000 B1. Give example of abelian groups A, B satisfying the fol-
lowing.
(i) A B = 0. Use A = Z
2
and B = Z
3
. This is easy to see.
(ii) AA A. Z
2
Z
2
Z
2
because any simple tensor does this
a b = (ab) 1 = (ab)(1 1)
so that the tensor product is just 1 1. The generator obviously has
order 2, and the claim follows
2000 B2.
Let R be a ring with identity. Assume that M is a simple mod-
ule for R. Prove that M is cyclic and that any non-zero R-module
endomorphism of M is an isomorphism.
Proof. Recall that a simple R-module is a non-zero R-module
with no proper submodules. As such, there exists a M 0. By
denition the cyclic submodule Ra is a non-zero submodule of M,
since the latter is simple it must be that Ra = M.
Let be a non-zero R-module endomorphism of M. Such a map is
uniquely and completely determined by (a), since (ra) = r(a) for
133
134 2000 QUAL
all m = ra M = Ra. It follows that (a) ,= 0, and so im = R(a)
is a non-zero submodule of M. Another appeal to simplicity gives
im = M. It is straightforward to see that is injective.
2000 B3. Let P be an R-module. Then P is projective if and only
if P is a direct summand of a free R-module.
Proof. Prove Hungerford IV.3.4(i) (ii) (iii), then (iii)
(i).
2000 B4.
If A is any abelian group, compute Hom
Z
(Z
m
, A).
Solution. Recall
A[m] = a A : ma = 0.
Dene f : Hom A[m] by f() = (1). Since is a homomorphism
of abelian groups it is assured that (1) A[m]. Observe that
f( +) = [ + ](1) = (1) + (1) = f() + f().
Given a A[m], the homomorphism : Z
m
A[m] whereby : 1 a
is such that f() = a, hence f is surjective. If f() = 0 then by
denition (1) = 0, and since is completely determined by the image
of 1 it follows that = 0. Thus f is injective, and consequently
Hom
Z
(Z
m
, A) A[m].
What can you say about Z
m
? The denition of said object is
Z
m
= Hom
Z
(Z
m
, Z).
Using the previous part of this question and the fact that Z is an
integral domain
Z
m
Z[m] = a Z : ma = 0 = 0.
Alternatively, one observes that the only nite subgroup of Z is the
trivial one. Since the image of Z
m
under any linear functional is a
nite subgroup of Z, said map must just be the zero homomorphism.
2000 B5.
PART B 135
Prove that a free module over a PID is torsion-free.
Proof. Let F be the free R-module on some non-empty set X,
with R a PID. Choose a F 0 and recall that the annihilator of a is
the ideal of R given by
ann
R
(a) = r R : ra = 0.
By hypothesis there exists r R such that ann
R
(a) = (r). Since F is
free, a has the form
a =
n
i=1
r
i
x
i
(r
i
R 0, x
i
X).
It suces to show r = 0. By denition
0 = ra =
i
(rr
i
)x
i
= rr
i
= 0 (i = 1, . . . , n).
Since R is a domain it follows that r = 0 or r
i
= 0. If r = 0 the
proof is complete. Elseif r
i
= 0 for all i then a = 0, contradiction. In
conclusion, ann
R
(a) = 0 so R is torsion-free.
Give an example to show that the converse is false. As an abelian
group (Z-module), is torsion free and not free.
Why is not a free Z-module. Suppose that =
Z
is free on
some nonempty X. Choose r/s, r
/s
X such that (r, s) = 1 = (r
, s
).
Let
n = s
[r, r
]
r
n
= (s
)
[r, r
]
r
both are integers since r, r
[[r, r
] = lcm(r, r
). Then
n(r/s) + n
(r
/s
) = s
[r, r
]
r
r
s
+ (s
)
[r, r
]
r
= [r, r
] [r, r
] = 0
so that X is linearly dependent when [X[ 2. If [X[ = 1 then X
cannot span , since , Z as abelian groups.
2000 B6.
136 2000 QUAL
What are the invariant factors of Z
n
Z
m
, regarded as a module
over Z?
Example. 360 = 2
3
3
2
5 7
0
and 525 = 2
0
3 5
2
7. The
elementary divisors of Z
360
Z
525
are 2
3
, 3, 3
3
, 5, 5
3
, 7 so the invariant
factors are given by
1 3 5 1 = 15
2
3
3
2
5
2
7 = something.
Solution of 2000 B6. Factor
n =
k
i=1
p
a
i
i
and m =
k
i=1
p
b
i
i
where p
i
= p Z : p[n or p[m and a
i
, b
i
0 are integers. Arrange
the primes in increasing order: i < j implies p
i
< p
j
. The elementary
divisors of Z
n
Z
m
are p
a
i
i
, p
b
i
i
for all a
i
, b
i
> 0. The invariant factors
are given by
k
i=1
p
min{a
i
,b
i
}
i
and
k
i=1
p
max{a
i
,b
i
}
i
.
2000 B8.
Prove that if q is the minimal polynomial of a linear transformation
of a vector space E, then deg q dimE.
Proof. Assume dim
K
E = d (K is a eld) and : E E is a
linear transformation with minimal polynomial q. Recall that q is the
unique monic polynomial of minimal degree which annihilates , (i.e.
q() = 0) and divides all other polynomials annihilating .
The characteristic polynomial p of
p = det(xI
d
A) K[x]
(where A is the matrix of with respect to some basis of E) is monic
of degree d. The content of the Cayley-Hamilton theorem is q[p, which
implies that deg q deg p = d.
1999 Qual
Parts A and B
1999 A1 unnished.
1999 A2 unnished.
1999 AB5. Show that the direct sum of an arbitrary family of
injective abelian groups is injective. (Hint: divisible.)
Proof. Injective abelian groups are divisible abelian groups per
Hungerford IV.3.9. Let A =
iI
A
i
be a direct sum of injective
abelian groups. Then each A
i
is a divisible abelian group. Select y A
and n a nonzero integer. Write y = (a
i
: i I) with almost all a
i
= 0.
Dene x = (x
i
: i I) A as follows: for each i I if a
i
= 0 then set
x
i
= 0; if a
i
,= 0 then choose x
i
A
i
satisfying nx
i
= a
i
, possible by
the divisibility of A
i
. By design nx = y, so A is divisible. Therefore A
is an injective Z-module.
The result that divisible abelian groups are exactly the injective
Z-modules is critical here, so this solution should be accompanied by
(at least) a sketch of Hungerford IV.3.9.
1999 AB7.
(a) Give an example of an integral domain R and ideals I and J of
R such that IJ ,= I J.
Remark. 2003 B1 indicates that under the assumption that R is a
PID, IJ = IJ implies I+J = R. The contrapositive of this statement
is that (still assuming R is a PID) I + J ,= R implies IJ ,= I J. So
it suces to get a PID and nd ideals whose sum is not R.
137
138 1999 QUAL
Example. Let R = Z, I = (2) and J = (6). [Remember that
(2, 6) = 2 implies that (2) + (6) = (2) ,= Z.] By denition
IJ = 2x
1
6y
1
+ + 2x
n
6y
n
: x
i
, y
i
Z
= 12z
i
: z
i
Z
= (12)
while
I J = x : 2[x and 6[x
= x : 6[x = (6).
Here, IJ _ I J.
(b) Show that if I +J = R then IJ = I J.
Proof. The inclusion IJ I J is always true. The the reverse,
use the x +y = 1 trick.
1999 AB8.
Give an example of a monomorphism f : A B of abelian groups
and an abelian group C such that the induced homomorphism

f :
Hom
Z
(B, C) Hom
Z
(A, C) is not onto. The induced homomorphism
is dened by

f() = f, where : B C. Diagramatically,

f gives
an arrow between arrows as follows:
0
/
A
f
/
f()
~
~
~
~
~
~
~
C
Example. Let A = Z, B = , f = : Z , and C = Z. For
= 1
Z
: A B there is no : Z such that = 1
Z
because
Hom
Z
(, Z) = 0.
0
/
Z

/
1
Z

Z
PART C 139
1999 AB10. Let
A =
_
_
2
1
1
1
1 1
1
2
1
_
_
Find the minimal polynomial and the rational canonical form of A.
Solution. A is in Jordan form. The roots of the minimal (and
characteristic) polynomial are = 1, 1, 2. The multiplicity of as a
root of q
A
is exactly the size of the largest Jordan block corresponding
to . Accordingly,
q
A
= (x + 1)(x 1)
2
(x 2).
The characteristic polynomial of A is
p
A
= det(xI
8
A) = (x2)(x+1)(x1)(x1)
2
(x+1)(x2)(x+1),
which also gives the list of elementary divisors. Using the power grid,
the invariant factors are
(x + 1) = q
1
(x + 1)(x 2)(x 1) = q
2
(x + 1)(x 2)(x 1)
2
= q
3
= q
A
.
The rational canonical form of A is the direct sum of the companion
matrices of the invariant factors
A C =
_
_
1
0 1 0
0 0 1
2 1 2
0 1 0 0
0 0 1 0
0 0 0 1
2 3 1 3
_
Part C
Do 4 problems.
1999 C3.
140 1999 QUAL
Prove or disprove: The polynomial f(x) = x
625
x1 is irreducible
over Z
5
. By way of contradiction, suppose f is irreducible.
Let F = Z
5
(u) where u is a root of f. Then [F : Z
5
] = 5
4
implies
[F[ = [Z
5
[
[F:Z
5
]
= 5
5
4
.
This means that F is a splitting eld of x
5
5
4
x over Z
5
. In particular,
the equations
u
5
5
4
= u and u
5
4
= u + 1
hold. Raise the second equation to the power of 5
4
:
_
u
5
4
_
5
4
= (u + 1)
5
4
= u
5
4
+ 1
5
4
= (u + 1) + 1 = u + 2.
And again:
_
_
u
5
4
_
5
4
_
5
4
= (u + 2)
5
4
= u
5
4
+ 2
5
4
= (u + 1) + 2 = u + 3.
The suggests the equation (which I am not going to prove)
_
_
_
u
5
4
_
.
.
.
_
_
5
4
= u + n
where the dots denote n-fold exponentiation.
Now, since 625 = 5
4
it follows that
u
5
625
= u
5
5
4
= u.
Consider
u = u
5
625
= u
5
u
5
624
= u
5
u
5
4156
= u
5
(u +156) = u
5
(u +1) = u
6
+u
5
.
Accordingly, x
6
+ x
5
x divides f and so the latter is reducible. This
contradicts the initial assumption that f is irreducible, so it must in
fact be reducible over Z
5
.
1999 C4.
Prove or disprove: Let F be a degree 4 extension of . If any proper
subeld of F is Galois over , then F is Galois over .
In the following example, F E with F E and E
normal but F not normal. Being a normal extension is equiva-
lent to being a splitting eld of a set of polynomials. Being algebraic
(comes for free here) and Galois (what I want) is equivalent to being
a splitting eld of a set of separable polynomials. So, showing that
the polynomials used here are separable it sucient to conclude that
F E and E are Galois but F is not Galois.
PART C 141
Counterexample. Consider the real fourth root of 2,
4
2 1
and its square
2 1. The tower
(
2) (
4
2)
is such that
(1) (
2)
and
(2) (
2) (
4
2)
are Galois, but
(3) (
4
2)
is not Galois. It suces to prove that in (1) and (2), the larger elds
are splitting elds of a separable polynomial over the smaller elds, but
not so in (3).
For (1), let f(x) = x
2
2 [x]. The roots of f in (
2) are
2, all of which are distinct, so (
2) is indeed the splitting eld

of the separable polynomial f over . It follows that extension (1) is
Galois.
For (2), let g(x) = x
2
2 (
2)[x]. The roots of g in (

4
2)
are
4
2, all of which are distinct, so (

4
2) is indeed the splitting

eld of the separable polynomial g over (
2). Hence g is separable

over (
2). It follows that extension (2) is Galois.

For (3), consider h(x) = x
4
2 [x]. This is a separable polyno-
mial (over ) having
4
2 as a root, but (
4
2) is not the splitting eld

of h over . Indeed, if F denotes the splitting eld of h over , then
i
4
2 F (
4
2). Consequently, extension (3) is not Galois.

1999 C6.
Let F > k be elds.
(1) If [F : k] = , then tr.d.
k
F > 0.
False. The extension
(
2,
3, . . . ,
p, . . . ) >
(p is prime) is an innite-dimensional algebraic extension. But
algebraic extensions have zero transcendence degree: For any
element u F has an irreducible polynomial over k which
annihilates it, which means is a maximal algebraically inde-
pendent subset of F.
(2) If [F : k] < then tr.d.
k
F = 0.
True. If F is a nite-dimensional extension of k, then F
is algebraic over k: If u F then the set 1, u, . . . , u
[F:k]
is
142 1999 QUAL
linearly dependent over k and there exist a
0
, . . . , a
[F:k]
k (not
all zero) such that
[F:k]
i=1
a
i
u
i
= 0. But algebraic extensions
have zero transcendence degree.
(3) If tr.d.
k
F < , then tr.d.
k
F = [F : k].
False. The trivial extension k > k has dimension one and
transcendence degree zero.
1998 Qual
Part A
1998 A1 unnished.
1998 A4. Let R be an integral domain. Then R is Euclidean if
and only if R is a principal ideal domain if and only if R is a unique
factorization domain.
Proof. The implications ED PID UFD are standard, see
Hungerford III.3.7/9. An explanation of the main ideas of each proof
should be sucient.
A counterexample to UFD PID is Z[x]. See Hungerford III.6.Ex1(a,b).
Since the integers are a unique factorization domain, so is Z[x]; see Gal-
lian 17.6.
A counterexample to PID ED is Z[(1 +
19i)/2]. Dummit and

Foote 8.5 and the following example show that its not a ED, and the
example following 8.9 on p.282 shows that it is a PID.
143
1997 Qual
Part B
1997 B2.
Let A be a left module over a ring R. Construct a canonical R-
module homomorphism : A A
, where for any left or right R-

module B, B
:= Hom
R
(B, R). Give a sucient condition on R for
to be an isomorphism.
Solution. is dened by (a)(f) = f(a). Check that is a ho-
momorphism of R-modules. If R 1 and A is free of nite rank, then
is an isomorphism.
Let B = b
1
, . . . , b
n
be a basis for A.
is injective Let a ker . For all f A
,
0 = (a)(f) = f(a)
= f(r
1
b
1
+ +r
n
b
n
)
= r
1
f(b
1
) + + r
n
f(b
n
).
This is true in particular for the elements f
j
A
dened by
f
j
(b
k
) =
jk
. So
0 = r
1
f
j
(b
1
) + +r
n
f
j
(b
n
)
for each j = 1, . . . , n implies a = 0.
is surjective Take for granted that A
is free on f
j
and A
is free on g
j
: j = 1, . . . , n where g
j
(f
k
) =
jk
. (Too easy
to write down) It suces to show that, for all g
j
, there exists
a A such that (a) = g
j
. Well, let a = b
j
and observe that
for all f
k
(b
j
)(f
k
) = f
k
(b
j
)
=
kj
=
jk
= g
j
(f
k
).
So is surjective.
Therefore is an isomorphism of R-modules.
145
146 1997 QUAL
1997 B3.
Let R be a nonzero commutative ring with unity such that every
submodule of any free R-module of nite rank is free. Prove that R is
a PID..
Proof. As a module for itself, R is free on 1. Let I be an ideal
of R. By denition I is an R-submodule of R. By hypothesis I is free,
and so rank I rank R = 1. Now, the rank of I cannot be zero for
in this case I = is not an ideal. So I is free on a for some a I.
Since R is commutative,
I = Ra = aR = (a)
is a principal ideal. Accordingly, R is a PID.
1997 B4.
Prove that any maximal linearly independent subset X of a (left)
vector space V over a division ring D is a basis of V .
Proof. By denition X is a basis of X < V . If there is a vector
v V W then consider the set X v. Suppose r, r
1
, . . . , r
n
D
and x
1
, . . . , x
n
X satisfy
rv +r
1
x
1
+ +r
n
x
n
= 0.
Re-arrange to see that
v = (r
1
)(r
1
x
1
+ + r
n
x
n
) X,
contradictory to the choice of v unless r = 0. But then the linear
independence of X implies r
i
= 0 for all i. This means that X
v is linearly independent, contradictory to the hypothesis that X is
maximal linearly independent. Consequently, there is no such v. In
other words X = V , i.e. X spans V .
Part C
1997 C1.
(i) Dene separable extension and perfect eld.
Definition. A polynomial is separable if and only if all of its roots
in some splitting eld are distinct. An algebraic element of a eld is
separable if and only if its irreducible polynomial is separable. An
algebraic extension eld is separable if and only if all of its elements
are separable.
Definition. A eld K is perfect if any of the following (equivalent)
conditions are true:
(1) every irreducible polynomial in K[x] is separable,
PART C 147
(2) every algebraic closure K of K is Galois over K,
(3) every algebraic extension of K is separable,
(4) char K = 0 or char K = p and K = K
p
.
(ii) Prove that every nite eld is perfect.
Proof. It suces to show that (4) is true. Suppose that K is a
nite eld; there exists a prime p and a non-negative integer n such
that [K[ = p
n
and char K = p.
To see that every a K has the form a = b
p
for some b K,
consider the map : K K whereby (b) = b
p
. Observe that is
operation-preserving
(bc) = (bc)
p
= b
p
c
p
= (b)(c)
by commutativity of multiplication in K and
(b +c) = (b +c)
p
= b
p
+ c
p
= (b) + (c)
by the binomial theorem and the fact that char K = p.
Recall that the multiplicative group of the units of K is isomorphic
to the additive group of integers modulo p
n
1,
_
K
,
_
_
Z
p
n
1
, +
_
.
Evidently, each b K
satises the equation b

p
n
1
= 1
K
, thus b
p
n
= p.
But this shows that is surjective, since (0) = 0. In other words,
K = K
p
and it follows that K is perfect.
1997 C2.
Prove from scratch that every nite eld admits algebraic extensions
of arbitrarily large degree.
Proof. Consider F
q
, where q is a prime power. Let n 1 be
arbitrary and let E be a splitting eld of
f = x
q
n
x
over F
q
. Note that [E : F
q
] divides q
n
! and is hence nite, so E is an
algebraic extension of F
q
. Let U = roots of f, so that E = F
q
(U).
Once E is shown to be a splitting eld of f over the prime subeld
Z
p
F
q
it will follow that [E[ = q
n
, and so
[E[ = q
n
= [F
q
[
[E:Fq]
= q
[E:Fq]
whence [E : F
q
] = n.
Note that F
q
is the splitting eld of g = x
q
x over Z
p
; in other
words F
q
= Z
p
(V ), where V is the set of all roots of g. To prove that E
is a splitting eld of f over Z
p
it suces to prove that V U, whence
E = F
q
(U) = Z
p
(U, V ) = Z
p
(U).
148 1997 QUAL
If v is a root of g, then v
q
= v. But then
v
q
n
= (v
q
)
q
n1
= (v)
q
n1
= = v
so V U.
1996 Qual
Part B
1996 B1. Describe all maximal ideals in the polynomial ring C[x
1
, . . . , x
n
].
Prove your answer.
Solution. All maximal ideals in C[x
1
, . . . , x
n
] look like
(x
1
a
1
, . . . , x
n
a
n
)C[x
1
, . . . , x
n
].
Induct on n to prove it. For n = 1 we are looking at C[x], which is
a principal ideal domain. Since the complex numbers are algebraically
closed every polynomial splits. The maximal ideals, which correspond
to the irreducible polynomials, are those ideals of the form (xa)C, a
C.
Assume that the maximal ideals of C[x
1
, . . . , x
n1
] are exactly those
ideals of the form (x
1
a
1
, . . . , x
n
a
n1
)C[x
1
, . . . , x
n1
]. Suppose M is
a maximal ideal in C[x
1
, . . . , x
n1
]. Then there exist a
1
, . . . , a
n1
C
such that
M = (x
1
a
1
, . . . , x
n
a
n1
)C[x
1
, . . . , x
n1
].
The quotient C[x
1
, . . . , x
n1
]/M is a eld and
C[x
1
, . . . , x
n
]/M[x
n
]

= C[x
1
, . . . , x
n
]/MC[x
1
, . . . , x
n
]
is a principal ideal domain. The maximal ideals of this domain are (x
n
a
n
)(that) by the base case. Therefore its preimage (x
1
a
1
, . . . , x
n

a
n
)C[x
1
, . . . , x
n
] under canonical projection is maximal in C[x
1
, . . . , x
n
].
1996 B2.
Let R be a ring. A left R-module M is called simple i M ,= 0 and
M has no proper submodule. Prove that for any simple left R-module
the endomorphism ring Hom
R
(M, M) is a division ring.
Definition. A division ring is a ring with identity 1 ,= 0 in which
every non-zero element is a unit.
149
150 1996 QUAL
Proof. The ring structure of Hom
R
(M, M) is as follows
[f + g](m) = f(m) + g(m)
[fg](m) = f g(m)
[rf](m) = rf(m).
Note that the identity 1
M
Hom
R
(M, M), and 1
M
,= 0 since M ,= 0. check the action
The image of a given f Hom
R
(M, M) is a submodule of M.
Recalling that ker f, im f are submodules of M, there are four cases to
consider (using simplicity).
ker f im f conclusion
M M M = 0, forbidden
0 M f is an automorphism
M 0 f = 0
0 0 M = 0, forbidden
If f ,= 0 then it must be an automorphism and it therefore has an in-
verse f
1
which is also an automorphism. In conclusion, Hom
R
(M, M)
is a division ring.
1996 B3.
Consider the C[x]-module structure on C
n
dened by p(x) v =
_
p(A)
_
(v), where v C
n
, p(x) C[x], and A is the n n matrix
below. Decompose C
n
according to the structure theorem for modules
over a principal ideal domain.
A =
_
_
_
_
_
_
_
_
_
0 0 0 0 0
i 0 0 0 0
0 i 0 0 0
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0 0 0 0 0
0 0 0 i 0
_
_
_
_
_
_
_
_
_
Since C is a eld C[x] is a principal ideal domain. The structure
theorem indicates that
C
n
j=1
C[x]
(q
j
)
where q
j
are the invariant factors of A. Recall that q
t
= q
A
is the
minimal polynomial of A. The characteristic polynomial of A is
p
A
(x) = det(xI
n
A) = x
n
because A is lower-triangular. Obviously A ,= 0, matrix multiplication
shows that A
k
,= 0 for k = 2, . . . , n 1. So q
A
= x
n
= p
A
. But
PART C 151
the characteristic polynomial is the product of the invariant factors, so
t = 1 and the structure theorem gives
C
n
C[x]
(x
n
)
.
Part C
1996 C1. Suppose G is a Galois extension eld of degree 7 over
F. What can you say about a group of automorphisms of E whose
elements x each element of F? Prove your answer.
Solution. Any such group is a subgroup of the Galois group of E
over F
G = Aut
F
E = Aut(E) Hom
F
(E, E).
The extension E F is nite-dimensional and Galois, so the Funda-
mental Theorem of Galois Theory indicates that [G[ = [E : F] = 7.
Given G 1
E
, Lagranges theorem implies that the order of
divides 7. Since is not the identity, [[ = 7, which means
G =
is a cyclic group. Since the order of any subgroup of G divides the
order of G (Lagrange), any such group as described in the problem
statement is either trivial or equal to G itself.
1996 C3.
Let f be a polynomial of degree n over F and let E be a splitting
eld of f over F. Prove that the degree of E over F is at most n!.
A stronger result is true: [E : F] divides n!.
Proof. Strong induction on n!.
When n = 1, f(x) = ax + b with a, b F and a ,= 0. The root of
f is b/a F, so E = F and [E : F] = 1.
Assume the statement is true for k = 1, . . . , n 1. If all roots
of f belong to F then the proof is complete since E = F as before.
Otherwise, let u E F be a root of f and let q F[x] denote us
irreducible polynomial. Suppose deg q = r. Since q[f, the latter can
be factored over F as
f(x) = q(x)h(x), h(x) F[x]
where deg h = n r.
In case f is irreducible over F, it follows that deg h = 0 and conse-
quently r = n. Over F(u), f can be factored as
f(x) = (x u)
m
f
1
(x), f
1
F(u)[x]
152 1996 QUAL
where m 1 and deg f
1
= n m. Now E is a splitting eld of f
1
over
F(u), so by the induction hypothesis [E : F(u)] divides (n m)!, say
with quotient s
1
. Then
[E : F] = [E : F(u)][F(u) : F]
and [F(u) : F] = r = n implies
[E : F]s = (n m)! n
which divides n!. Certainly then [E : F] n!.
In case f is reducible over F, let u = u
1
, . . . , u
t
be the distinct
roots of q in E. By design, F(u
1
, . . . u
t
) is a splitting eld of q over
F. The induction hypothesis indicates that [F(u
1
, . . . , u
t
) : F] divides
r!, say with quotient s
2
. Now E must be a splitting eld of h over
F(u
1
, . . . , u
t
), and since deg h < n the induction hypothesis gives that
[E : F(u
1
, . . . , u
t
)] divides (n r)!, say with quotient s
3
. Then
[E : F] = [E : F(u
1
, . . . , u
t
)][F(u
1
, . . . , u
t
) : F]
implies
[E : F]s
2
s
3
= r! (n r)!
which divides n! (with quotient
_
n
r
_
). Certainly then [E : F] n!.
1996 C4.
What can you say about the cardinality of an algebraic closure of
the eld of 3 elements? Sketch a proof of your answer.
Theorem. If E is an algebraic extension of F, then
[E[
0
[F[.
Let E denote an algebraic closure of F
3
. The answer to this question
is [E[ =
0
.
Proof. Since 3
0
=
0
, by the theorem [E[
0
. If [E[ <
0
,
then E is nite. But in this case E cannot be algebraically closed, for
the non-constant polynomial
f(x) = 1 +
uE
(x u)
has no root in E.
1996 C5.
PART C 153
Prove that there exists an extension eld of F
4
of degree 2 over F
4
.
Proof. Write F
4
= 0, 1, u
1
, u
2
and recall that as an additive
group, F
4
Z
2
Z
2
. For the purposes of addition we can make the
identications
0 (0, 0), 1 (1, 1), u
1
(1, 0), u
2
(0, 1).
As a multiplicative group, F
4
Z
3
(the latter is written additively) so
for the purposes of multiplication we can make the identications
1 0, u
1
1, u
2
2.
Consider p(x) = x
2
+ u
1
x + 1 F
4
[x]. If p is reducible then it must
split, so to show that p is irreducible it suces to check that p has no
roots in F
4
. Using the identications above, compute
p(0) = 1, p(1) = u
1
p(u
1
) = 1, p(u
2
) = u
2
.
Accordingly, p is irreducible over F
4
. Adjoining a root of p, call it v,
to F
4
yields an extension of degree 2.
1995 Qual
Part A
1995 A1. Dene normal subgroup.
Definition. A subgroup N of a group G is normal if and only if
N is invariant under each inner automorphism of G. There are several
other equivalent characterizations of this concept.
Give an example of a subgroup that is normal and one that is not.
Example (A normal subgroup). Consider 2Z < Z. Each inner
automorphism of Z is given by
n x +n x
for some x Z. Now, n 2Z if and only if n = 2k for some k Z.
But then
x + n x = x + 2k x = x x + 2k = 2k 2Z
shows that 2Z is invariant under any inner automorphism of Z.
That example really is as trivial as it seems since every subgroup
of an abelian group is normal.
Example (A non-normal subgroup). [1994 A3(b)]
1995 A2.
1995 A3. If [G[ = p
n
, then C(G) _ e. In other words, p-groups
have nontrivial centers.
Proof. By Lagranges theorem, [C(G)[ divides p
n
. From the class
equation
[G[ =
n
i=1
[G : C
G
(x
i
)] = [C(G)[ +
m
j=1
[G : C
G
(x
j
)]
[G : C
G
(x
j
)] divides p
n
for each x
j
under consideration. As the sum
in the rightmost term of the above equality is extended over the dis-
tinct conjugacy classes of the elements x
j
G C(G), certainly
155
156 1995 QUAL
[G : C
G
(x
j
)] 2 for each j. In order for the divisibility conditions
observed earlier to be met, it is clear that p
[G : C
G
(x
j
)] for all j.
But now, the fact that p divides both [G[ and
m
j=1
[G : C
G
(x
j
)]
forces p to also divide [C(G)[. In other words
[C(G)[ = pk, for some positive integer k.
Now it is clear that [C(G)[ contains at least p elements.
See the remarks made in (A.3.1) if further clarication is necessary.
1995 A5. Show that every group of order p
2
, p a prime, is
abelian.
Proof. Since G is a p-group, C(G) _ e by problem 3 from
the present exam. By Lagranges theorem, [C(G)[ divides p
2
; since
p is prime it is clear that C(G) consists of either p or p
2
elements.
Certainly G is abelian in the latter case, for then G = C(G). Otherwise,
if [C(G)[ = p then another application of Lagranges theorem gives
[G/C(G)[ = p. As a group of prime order, G/C(G) is cyclic; by the
G/Z theorem G is abelian.
Part B
1995 B1.
(a) Dene: an R-module is projective. An R-module P is projective
when, given any diagram of R-modules and R-module homomorphisms
with bottom row exact
P
g
A
f
/
B
/
0
there exists G Hom
R
(P, A) such that fG = g.
(b) Prove that if R is a eld then any R-module is projective.
Proof. The argument proceeds as follows: Every module over a
eld is free and all free modules (over any ring) are projective.
Let M =
R
M be an R-module. To prove that R is free it suces
to show that any linearly independent subset of M is contained in a
basis of M. Then this statement can be applied to (trivially linearly
independent) to conclude that M has a basis.
PART B 157
Let X be a linearly independent subset of M and let S denote the
set
4
of all linearly independent subsets of M containing X. Note S ,=
because X X. Use to make S into a poset. Choose an arbitrary
chain C
i
: i I from S. The set C =
i
C
i
is a candidate for an
upper bound of the chain in S. Evidently X, C
i
C for all i. It must
be shown that S is linearly independent. Suppose c
1
, . . . , c
n
C and
r
1
, . . . , r
n
R satisfy
r
1
c
1
+ +r
n
c
n
= 0.
There are nitely many c
j
, so there exists m(j) N such that c
j
C
m
for all j. The linear independence of C
m
implies r
j
= 0 for all j, hence
C is linearly independent. Thus every chain in S has an upper bound in
S. It follows from Zorns lemma that S contains a -maximal element
B. Pedantically, X B and B is linearly independent.
By denition B is a basis of B. If W ,= M then choose m MW
and consider B m. If r, r
1
, . . . , r
n
R and b
1
, . . . , b
n
B satisfy
rm +r
1
b
1
+ +r
n
b
n
= 0
then re-arrange to see that
m = (r
1
)(r
1
b
1
+ +r
n
b
n
) W
which is contradictory unless r = 0. But then the linear independence
of B implies that r
j
= 0 for all j. This means that B m is a linear
independent set which is -larger than B, a contradiction. Therefore
W = M after all, so M is free on B.
Given a diagram as described in part (a), augment it to include
B M
B

/
M
g
A
f
/
D
/
0
For all b B there exists a
b
A such that f(a) = g(b), because f
is surjective. Dene a map of sets h : B A by h(b) = a
b
. The
universal property of free provides a unique R-module homomorphism
H : M A such that H = h
B
h
/
M
!H
.~}
}
}
}
g
A
f
/
D
/
0
4
S T(M) so it is a set
158 1995 QUAL
By virtue of the setup, fH = g, therefore M is a projective R-module.
1995 B4. R is a commutative unital ring, and all modules are

unitary.
Let M be an R-module and let H be the set of all m in M such
that rm = 0 implies r = 0 or m = 0.
(1) Prove that if R is a principal ideal domain and M is cyclic
then H is a submodule of M.
(2) Is the result in part (1) true of the hypothesis that R is a
principal ideal domain is deleted? Give a proof or a coun-
terexample.
1. There exists a M such that M = Ra. The set H comprises
zero and those elements of M that are not torsion. By Hungerford
IV.6.6, M = F M
t
where F is free of nite rank. In fact, rank
R
F 1
since M has one generator. We have H M
t
= 0 by denition, since if
m H0 and rm = 0 it must be the case that r = 0. The directness
of the sum gives H F. If a M
t
then M = M
t
and F = 0, so H = 0
is a submodule. If a / M
t
then M
t
= 0 and M = F. Claim F H.
This is clear, because F

= M/M
t
means that all nonzero elements of
F are torsion-free: that is the dening condition of H.
That proof was awkward.
2. Let R = Z
4
= M. This ring is not a domain, and is cyclic
on 1 as a module for itself. Going through the elements one-by-one
shows H = 0, 1, 3 which is not a submodule since 2 1 = 2 / H, for
example.
Part C
1995 C3 (incomplete).
(a) Let p be a prime and F = Z
p
. Determine the number of irre-
ducible polynomials of type x
2
+cx+d F[x]. There are p
2
polynomials
of the indicated type. But which of these are irreducible over F?
If d = 0 and c ,= 0 then
x
2
+cx = x(x +c)
is irreducible for all p possible values of c.
If c = 0 and d ,= 0 then
x
2
+ d
is reducible if and only if d is a square in Z
p
. There are
p1
2
+1 squares
in Z
p
, so there are that many reducible polynomials of the form x
2
+d.
PART C 159
So far there are at most
p
2
p
p 1
2
1
irreducible polynomials of the indicated form. If c, d ,= 0 then
x
2
+cx +d = (x a)(x b)
if and only if c = (a +b) and d = ab. Yep.
(b) Let f be one of the polynomials described in (a). Prove that
K = F[x]/(f) is a eld containing p
2
elements, and that the elements
of K have the form a + b where a, b F and is a root of f in K.
Prove also that every such element a + b with b ,= 0 is a root of an
irreducible quadratic polynomial in F[x].
Proof. The rst two statements are easy; = x + (f).
If u = a + b then doing some algebra gives
u
2
(2a +bc)u + (a
2
+abc +b
2
d) = 0.
Yep.
1994 Qual
Part A
1994 A1.
1994 A2 improvement needed. Let G be a cyclic group of
order n and H be a cyclic group of order m. Describe all group homo-
morphisms G H.
Solution. Write G = a, [a[ = n and H = b, [b[ = m. Let
: G H be a group homomorphism. We have subgroups ker < G
and im < H; these are both cyclic since G, H are cyclic. Recall that
for all x G and all k Z,
(x
k
) = (x)
k
.
Observe that all y H satisfy y
m
= e
H
, so
(a
m
) = (a)
m
= e
H
= a
m
ker = a
m
< ker .
Now, a
n
= e
G
and (e
G
) = e
H
imply
e
H
= (a
n
) = (a)
n
.
Combining these results shows that
(a)
divides (m, n). Certainly

im =
(a)
_
, and so [im [ divides (m, n).
In particular, must be the trivial homomorphism when (m, n) = 1
. Otherwise, (a) = b
k
for k[ gcd(m, n).
1994 A3.
(a) Every subgroup of index 2 is normal. Let H < G with [G :
H] = 2. The set of left cosets of H in G is
G/H = H, aH
for some a G. If Ha ,= aH then Ha = h, meaning for all h H
there exists h
H such that
ha = h
.
But then
a = h
1
h
= a H = aH = H
161
162 1994 QUAL
a contradiction. Accordingly, aH = Ha which means H G.
(b) A subgroup of index 3 that is not normal. It is well-known ([2],
I.6 Exercise 4(a)) that
S
3
=
(1 2), (1 2 3)
_
.
Here, the order of (1 2) is 2; by Lagranges theorem
6 = [S
3
[ =
_
S
3
:
(1 2)
_
_
(1 2)
_
=
_
S
3
:
(1 2)
_
_
2
which means that the index of
(1 2)
_
is 3. Now
(1 2 3)(1 2)(1 2 3)
1
= (1 2 3)(1 2)(1 3 2)
= (2 3) /
(1 2)
_
shows that
(1 2)
_
is not normal in S
3
.
1994 A4 unnished.
Part B
1994 B2.
Consider a commutative diagram of R-module and homomorphisms
as below such that each row is exact. If two of the vertical maps are
injective, what about the third? Consider all three cases.
0
/
A
f
/
B
g
/
C
/
0
0
/
A
/
B
/
C
/
0
(1) If , are injective, then is injective. This is the Short Five
Lemma. Let b ker . Since g
is a homomorphism, g
(b) =
g
(0) = 0. By commutativity 0 = g
(b) = g(b), hence g(b)

ker . But is injective, so g(b) = 0. In other words b ker g.
By exactness b im f, say b = f(a) for some a A. (Note
f(a) ker .) By commutativity 0 = f(a) = f
(a), so
(a) ker f
. But f
is injective, so (a) = 0. Since is also

injective, a = 0 whence b = f(a) = 0. In conclusion ker = 0.
(2) If , are injective, then nothing. In the following situation
0
/
4Z

1
/
Z

1
/
Z
4
/
0
0
/
2Z

2
/
Z

2
/
Z
2
/
0
PART B 163
with (4n) = 2(2n) 2Z and (n) = n injective, the pigeon-
hole principle indicates that , no matter how it is dened,
cannot be injective.
(3) If , are injective, then nothing. Let : [x] denote
the canonical injection whereby r r + 0x + . . . . In the
following situation (-modules)
0
/
(x)

/
[x]
e
0
/
0
0
/
im
/
[x]
/
(x)
/
0
with e
0
(f) = f(0), (f) = f +im (identifying with its image
under the obvious isomorphism), = 1
[x]
and (r) = rx, the
only possibility for is the zero map.
1994 B4.
Which of the implications free projective torsion-free hold?
Give proofs or counters.
Proof that free projective. This is standard and has al-
ready been proven several times in this document.
Example (projective , free). Regard Z
6
as a module for itself, in
this way it is free on 1. The Fundamental Theorem of Finite Abelian
Groups gives Z
6
Z
2
Z
3
. Thus Z
2
is a projective Z
6
-module, but it
has no basis.
Example (free , torsion-free). Z
6
=
Z
6
Z
6
is free on 1 but is
certainly not torsion-free.
Example (torsion-free , free). =
Z
is torsion-free but has no
basis.
Example (projective ,torsion-free). Z
2
=
Z
6
Z
2
is projective since
Z
6
= Z
2
Z
3
but has no basis.
Z
4
=
Z
4
Z
4
is free in 1 and hence projective, but is certainly not
torsion-free.
Example (torsion-free , projective). Q =
Z
is torsion-free but
not projective.
Compute Hom
Z
(Z
2
, 1).
Proof. Briey: since 1 is torsion-free the homset must be zero.
1994 Version 2 Qual

Part A
1994v2 A1 unnished.
1994v2 A2 unnished.
(1) Find all the subgroups of Z
18
.
(2) Find all the ideals in the ring Z
18
.
(3) Which of these ideals are maximal and which are prime?
1.
1994v2 A3 unnished.
1994v2 A4. Let G be a group of order p
n
. Assume that the center
Z(G) has order p
n1
. Prove that G is Abelian.
Proof. Lagranges theorem implies that [Z(G)[ divides p
n
. As p
is prime, it follows that [Z(G)[ = p
k
for some positive integer k; by
hypothesis k p 1, p. If k = p 1, then [G/Z(G)[ = p, so the
quotient group G/Z(G) is cyclic. By the G/Z theorem, G is abelian.
Elseif k = p, then G = Z(G) is clearly abelian.
All p-groups have non-trivial centers. When its center is suciently
large, a p-group is abelian.
Part B
1994v2 B3.
A module M for a ring R is called simple if M ,= 0 and M has
no proper submodule, i.e. if N < M is a submodule then N = 0 or
N = M. Let M and M
be two simple modules and let : M M
be a
module homomorphism. Prove that is either zero or an isomorphism.
Proof. An easy verication shows that ker and im are sub-
modules of M and M
, respectively. By hypothesis there are four cases

165
166 1994 VERSION 2 QUAL
ker im conclusion
0 0 M = 0 forbidden
M 0 = 0
0 M is an isomorphism
M M
= 0 forbidden
The result follows.
1994v2 B4.
Let I be an ideal in a ring R. Prove that if R/I is a free R-module,
then I = 0.
Proof. Assume R is unital. Then, as a module for itself, R is free
on 1. Canonical projection : R R/I is an R-module epimor-
phism. Lemma IV 2.10 indicates that R/I is a free R/I-module on
(1). Regard R/I as an R-module on which I acts trivially. Thus
R/I is a free R-module on (1) and by Exercise IV 2.13 R/I
R,
where the direct sum is
(1)
-fold. But this just means R/I R,

from which it follows that I = 0.
1994v2 B5.
(a) Prove that a module P over a ring R is projective if and only
if, for any surjective homomorphism of R-modules : N N
the
corresponding homomorphism of abelian groups

: Hom
R
(P, N)
Hom
R
(P, N
) is surjective.
Proof. P is projective means that for any diagram of R-modules
and R-module homomorphisms (with bottom row exact)
P
A
/
B
/
0
there exists an R-module homomorphism P A making the diagram
commute.
() Assuming P is projective, suppose : P N
is given. Dia-
gramatically
P
.~|
|
|
|
N
/
N
/
0
By hypothesis there exists : P N satisfying =

(), hence

is surjective.
PART B 167
() Assuming

is surjective and : N N
is surjective, let
: P N
be given. The diagram is

P
/
N
/
0
By hypothesis there exists : P N such that

() = . But
(b) = shows that indeed closes the diagram.

(b) Z
2
Z
Z
5
= 0.
Proof. For any simple tensor a b Z
2
Z
Z
5
,
a b = (ab) 1 = ab(1 1)
so that
Z
2
Z
Z
5
= 1 1
is a cyclic group. Since
2(1 1) = 2 1 = 0 1 = 0
5(1 1) = 5 1 = 1 5 = 1 0 = 0
it follows that
1 1
divides 2 and 5, thus also (2, 5) = 1. This means

that the cyclic group Z
2
Z
Z
5
has order one, so Z
2
Z
Z
5
0.
1993 Qual
Part A
1993 A6. Let R be a local ring and f : R R
a surjective ring
homomorphism. Prove that R
is local. In other words, a quotient of

a local ring is local.
Proof. The ring R
has at least one maximal ideal. By way of

contradiction suppose that I ,= J are distinct maximal ideals in R
.
Show that f
1
(I), f
1
(J) are maximal ideals in R. If M is the unique
maximal ideal of R then it must be the case that f
1
(I) = M = f
1
(J).
Then, since f is surjective,
I = f f
1
(I) = f f
1
(J) = J,
which shows that R
is a local ring.
Why f
1
(I), f
1
(J) are maximal ideals in R: It suces to show
that f
1
(J) is maximal because the proof for f
1
(I) is symmetric and
the pullback of an ideal under a homomorphism is an ideal. Suppose
that K is an ideal of R such that f
1
(J) K R. Homomorphisms
preserve ideals, so f(K) is an ideal of R satisfying J f(K) R
.
Since J is maximal either f(K) = J or f(K) = R
.
If f(K) = J then K f
1
(J) then equality holds.
If f(K) = R
then choose any x R. Then f(x) R
= f(K)
so there exists x
K with f(x) = f(x
). Accordingly
5
xx

ker f f
1
(J) K. Accordingly x = x
+ (x x
) K, and
so K = R.
Therefore f
1
(J) is maximal.
Part B
1993 B2.
5
Useful fact: 0 every ideal of R
, so ker f f
1
(L) for every ideal L R
.
169
170 1993 QUAL
Give an example of a module D and an exact sequence 0 A
B C 0 which shows that neither Hom
R
(, D) nor Hom
R
(D, )
is an exact functor. Show that this example has the desired property.
Example (Hom
R
(, D) is not exact). Let D = Z
2
=
Z
Z
2
. For
Hom
Z
(, Z
2
) use
0 Z
2
Z
4
Z
2
0.
Apply the functor
0 Hom
Z
(Z
2
, Z
2
) Hom
Z
(Z
2
, Z
4
) Hom
Z
(Z
2
, Z
2
) 0
and recall that Hom
Z
(Z
n
, Z
m
) Z
(n,m)
. So the above sequence is
isomorphic to
0 Z
2
f
Z
2
g
Z
2
0.
Assume this last sequence is exact. The map f is an injection between
sets of the same size, so it is a bijection. By exactness, Z
2
= im f =
ker g, so g = 0 and the sequence is actually
0 Z
2
Z
2
0 0
which is nonsense.
What I was thinking: This asks for a non-injective abelian group. Re-
member that injective abelian groups are divisible abelian groups, and vice-
versa. Examples of these are Z and Z
n
for any n. Trying the smallest n
worked out, the trick is to get a 0 in an inconvenient place.
Example (Hom
R
(D, ) is not exact). For Hom
Z
(Z
2
, ) use
0 2Z

Z

Z
2
0.
Apply the functor
0 Hom
Z
(Z
2
, 2Z)

Hom
Z
(Z
2
, Z)

Hom
Z
(Z
2
, Z
2
) 0.
The right side of this new sequence cannot be exact: Every subgroup of
Z is either trivial or isomorphic to Z; in particular Z has no nite non-
trivial subgroups (Z-submodules). But if : Z
2
Z then im < Z
is a nite subgroup. Hence = 0, so Hom
Z
(Z
2
, Z) = 0. In a similar
way, Hom
Z
(Z
2
, 2Z) = 0. But Hom
Z
(Z
2
, Z
2
) Z
2
, so the new sequence
is actually
0 0

0

Hom
Z
(Z
2
, Z
2
) 0.
which is ridiculous.
1993 B3.
PART B 171
Give an example of a projective module which is not free and show
that this example has the desired property.
Example. Regard Z
6
as a module for itself and notice that it is free
on 1. The fundamental theorem of nite abelian groups indicates
Z
6
Z
2
Z
3
.
One of the equivalent conditions characterizing projective modules P
is that P is a direct summand of a free module. As Z
2
is exactly that,
we see that Z
2
is a projective Z
6
-module. There is no basis of Z
2
over
Z
6
, so the former is not free.
1993 B4.
Give an example of a submodule of a nitely-generated module that
is not nitely generated.
Example. Let R = Z[x
1
, x
2
, . . . ] and M = R. M is a nitely
generated R-module (by 1). Consider the R-submodule (ideal)
N = (x
1
, x
2
, . . . ).
By its denition N cannot be nitely generated over R: every polyno-
mial contains only nitely many indeterminants.
1993 B7.
A is a 9 by 9 matrix given in Jordan form.
A = J
1
(1)J
1
(1)J
1
(1)J
1
(1)J
1
(2)J
1
(2)J
2
(1)J
1
(1.)
Find the minimal polynomial of A. Inspection gives
q
A
(x) = (x + 1)(x 2)(x 1)
2
.
Find the eigenvalues of A. These are the entries along the diagonal
of A
= 1, 2, 1
(without multiplicity).
Find the rational canonical form of A. To do this the invariant
factors of A must be found. The minimal polynomial q
A
is the largest
invariant factor and the characteristic polynomial
p
A
(x) = (x + 1)
4
(x 2)
2
(x 1)
3
is the product of all the invariant factors. Thus
t1
j=1
q
j
= (x + 1)
3
(x 2)(x 1).
172 1993 QUAL
Given the form of q
A
and the divisibility relationships q
1
[ . . . [q
A
, the
invariant factors must be
(x + 1), (x + 1), (x + 1)(x 2)(x + 1), q
A
.
The rational canonical form is the matrix direct sum of the companion
matrices of these polynomials. Note that
(x+1)(x2)(x+1) = x
3
2x
2
x+2, q
A
= x
4
4x
3
+x
2
+4x4.
So
A C = (1) (1)
_
_
0 1 0
0 0 1
2 1 2
_
_
_
_
_
_
0 1 0 0
0 0 1 0
0 0 0 1
4 4 1 4
_
_
_
_
.
Part C
1993 C1.
(a) Dene simple algebraic extension F K of elds.
Definition. An algebraic extension F K is simple if F has the
form F = K(u) for some u F which is algebraic over K.
The degree of a simple algebraic extension is the degree of us min-
imum polynomial. This may be innite or a positive integer and is
denoted [F : K].
(b) Prove that if F K is a simple algebraic extension of degree
n, then [Aut
K
F[ n.
Proof. Let
q(x) = a
0
+a
1
x + +a
n
x
n
K[x]
be us minimum polynomial, choose Aut
K
F. Recall that 1, u, . . . , u
n1
is a basis of F = K(u) as a vector space over K. For all v F there

exist c
i
K (for 1 i n 1) such that
v = c
0
+ c
1
u + +c
n1
u
n1
(v) = c
1
(u) + + c
n1
(u)
n1
.
Thus the action of is completely determined by its action on u. In
particular
0 = (0) =
_
q(u)
_
= q
_
(u)
_
shows that (u) is also a root of q. There are at most n roots of q
contained in F, hence there are at most n distinct .
PART C 173
(c) An example where [Aut
K
F < n. Use a polynomial whose roots
are not all contained in a simple extension. The polynomial x
3
2 has
roots
u =
3
2 1,
3
2e
2i
3
,
3
2e
4i
3
C.
The extension F = (u) has degree 3 by virtue of the setup, but any
must send u to (u) = u, since the complex roots are not contained
in F. Consequently, the Galois group consists of only the identity.
1993 C3.
Let F be a nite eld and F
1
, F
2
be subelds of F. State and prove
a formula for [F
1
F
2
[ in terms of [F
1
[ and [F
2
[.
Lemma. F
p
a F
p
b if and only if a[b.
The formula that I will prove here is
[F
1
F
2
[ =
_
[F
1
[, [F
2
[
_
where (, ) denotes the greatest common divisor.
Proof. Recall that the GCD of two integers a, b is the unique
integer d such that
(1) d[a and d[b
(2) If c[a and c[b, then c[d.
Suppose that F = F
p
n and, using the lemma, F
1
= F
p
r
1 and F
2
= F
p
r
2 ,
where r
i
[n. Let m = [F
1
F
2
[.
It will suce to show that m satises the GCD property; an appeal
to uniqueness will yield m = (r
1
, r
2
) as desired. Certainly
F
1
F
2
F
1
, F
2
whence by the lemma m[r
1
, r
2
, which is property (1). Suppose that
K F
1
, F
2
. By the lemma [K[ = p
c
with c[r
1
, r
2
. But it is true that
K F
1
F
2
, and so (by the lemma again) c[m. This is property
(2).
1993 C5.
Prove or disprove that the degree of the splitting eld of a polynomial
of degree n divides n!.
Proof. Strong induction on n. Let K be a eld, f K[x] be a
polynomial of degree n, and F K the splitting eld of f over K.
Recall that F = K(roots off).
Base Case When n = 1, the situation is f(x) = ax + b, where
a, b K and a ,= 0. Clearly the root of f is b/a K, so F = K which
implies that [F : K] = 1.
174 1993 QUAL
Inductive Case Let n > 1 and assume that for k = 1, . . . , n1 the
degree of the splitting eld of a polynomial of degree k divides k!.
If all roots of f are elements of K, then [F : K] = 1 and there is
nothing to prove. Suppose u F K is a root of f. This element
is algebraic over K and has an irreducible polynomial, say g K[x].
Since f(u) = 0, g[f; say f = gh for some h K[x]. Denote deg g = r,
so deg h = n r.
If f is irreducible in K[x]: The polynomial h must be a constant,
so r = n. Factor f in K(u)[x] as f(x) = (x u)f
1
(x); certainly
deg f
1
= n 1. Now F is the splitting eld of f
1
over K(u), so the
induction hypothesis gives that [F : K(u)] divides (n 1)!. Say
[F : K(u)]q = (n 1)!.
But, since r = n,
[F : K] = [F : K(u)][K(u) : K] = [F : K(u)]n
shows that n! = [F : K(u)]nq = [F : K]q. In other words, [F : K]
divides n!.
If f is reducible in K[x]: Here deg h 1. Let u
1
, . . . , u
s
F be
the roots of g in F.
6
By denition, K(u
1
, . . . , u
s
) is the splitting eld of
g over K. Since [K(u
1
, . . . , u
s
) : K] = r < n, the induction hypothesis
indicates that [K(u
1
, . . . , u
s
) : K] divides r!. Now F is the splitting
eld of h over K(u
1
, . . . , u
s
). Because [F : K(u
1
, . . . , u
s
)] = n r < n,
it follows from the induction hypothesis that [F : K(u
1
, . . . , u
s
)] divides
(n r)!. Combining these two conclusions shows that
[F : K] = [F : K(u
1
, . . . , u
s
)] [K(u
1
, . . . , u
s
) : K]
divides r!(n r)!. But
_
n
r
_
=
n!
r!(n r!)
N
shows that [F : K] divides n!.
6
Work with the irreducible polynomial rst, then use what is left over to handle h.
Qual Seminar Homework
06/12/10
Classify all groups of order 45. This will use the fact that
[G[ = p
2
(p prime) implies G is an object in the category of abelian
groups.
Classification of groups of order 45. Sylows theorems im-
ply n
3
= 1 = n
5
; say P and Q are Sylow 3- and 5-subgroups of G, re-
spectively; note P, Q G. Also observe that [P[ = 9, [Q[ = 5 whence
Q Z
5
and, by the Fundamental Theorem of Finite Abelian groups,
P A Z
9
, Z
3
Z
3
.
If g P Q, then [g[ divides 5 and 9. Necessarily, [g[ = 1 which is to
say P Q = e. Certainly PQ = G; so [2], I 8.7 gives G = P Q,
i.e.
G Z
9
Z
5
or G Z
3
Z
3
Z
5
.
175
Exam Questions from Academic Year 20092010
Part B
201B Midterm
R will always denote a ring that has identity and all modules over
R will be assumed to be unitary. Questions 1 and 3 are worth 15 points
each. All others are worth 10 points and a perfect score is 45. All work
should be shown and reasoning explained. If you use a theorem, you
should try to quote it: at least the hypothesis and the conclusion that
you need.
201B Midterm Question 1. Prove that any simple R-module is
cyclic.
Proof. Let M be a simple R-module. If M = 0 then there is
nothing to prove, so assume M ,= 0. Choose a M and observe that
Ra < M is a non-zero submodule. By simplicity, Ra = M, showing
that M must be cyclic.
Give examples for the following:
(1) an example of R and a simple R-module.
R = Z and M = Z
2
(2) an example to show that a cyclic R-module need not be simple.
R = Z and M = 2Z > 4Z ,= 0
201B Midterm Question 2. Suppose that M and P are R-
modules and that P is a quotient of M. Prove that if P is projective,
then P is a summand of M, i.e. there exists a R-module (sic) N with
M N P.
Proof. P is a quotient of M means that there exists a surjective
R-module homomorphism M

P 0. P is projective means that
given any diagram of R-modules and R-module homomorphisms with
177
178 EXAM QUESTIONS FROM ACADEMIC YEAR 20092010
exact bottom row
P
g
A
f
/
B
/
0
there exists an R-module homomorphismh : P Asuch that fh = g.
In other words, said diagram can be closed.
It is a theorem ([2] IV 3.4) that P is projective if and only if for
every short exact sequence of the form
0 A
f
B
g
P 0
there exist R-modules J, K and R-module isomorphisms , , such
that the diagram
0
/
A
f
/
B
g
/
P
/
0
0
/
J

/
J K

/
K
/
0
is commutative.
A portion of the top row is given by the hypothesis that P is a
quotient of M, let ker < M play the role of A, with the inclusion
map standing in for f. The theorem provides
0
/
ker
/
M

/
P
/
0
0
/
J

/
J K

/
K
/
0
which shows that P is isomorphic to a direct summand of an R-module
which is isomorphic to M
P K J K M.
The result follows.
201B Midterm Question 3. Let R = C[x] be the ring of polno-
mials in one variable and let I be the ideal of R generated by x
2
and J
the ideal generated by x
3
. What are the elements of R/I?
Answer. The complex numbers form a eld, so R is a euclidean
domain. Given r =
d
i=0
a
i
x
i
R, its image in the quotient ring is, by
the euclidean algorithm show details on
the actual exam
r +I = a
1
x +a
0
+ (x
2
).
So R/I = a
1
x +a
0
+ (x
2
) : a
i
C.
201B MIDTERM 179
Write down explicitly the action of R on R/I.
Answer. R acts on R/I by left multiplication: given
f =
d
i=0
b
i
x
i
R, r +I = a
1
x +a
0
+ (x
2
) R/I
the action is dened by
f (r +I) =
_
f (a
1
x +a
0
)
_
+ (x
2
)
= . . .
= (b
0
a
1
+ b
1
a
0
)x + b
0
a
0
+ (x
2
).
Now, use this to prove that Hom
R
(R/I, R/J) = 0.
Proof. It suces to show that if belongs to said homset, then
it is necessary that = 0. In particular, such a satises
_
f (r +I)
_
= f (r +I).
The right-hand side of the above equation is
_
(b
0
a
1
+b
1
a
0
)x +b
0
a
0
+ (x
2
)
_
= (b
0
a
1
+ b
1
a
0
)x + b
0
a
0
+ (x
3
)
while the left-hand side is
_
d
i=0
b
i
x
i
_
_
a
1
x +a
0
x + (x
3
)
_
=
d
i=0
b
i
a
1
x
i+1
+
d
i=1
b
i
a
0
x
i
+ (x
3
)
= (b
1
a
1
+ b
2
a
0
)x
2
+ (b
0
a
1
+a
1
b
0
)x +b
0
a
0
+ (x
3
).
The two sides do not agree unless = 0.
201B Midterm Question 4. Let R and I be as in exercise 3.
Prove that the sequence
0 I

R

R/I
is exact but not split.
Proof. Exact means im = ker . This is clear. Split means
that one of the following equivalent statements is true
(1) is invertible: there exists an R-module homomorphism f :
R R/I such that f = 1
I
.
(2) is invertible: there exists an R-module homomorphism g :
R/I R such that g = 1
R
.
(3) Some other garbage about isomorphic short exact sequences
involving a direct sum of R-modules.
Condition (2) is equivalent to being injective. This is not true be-
cause by denition ker = I ,= 0, and an R-module homomorphism is
injective if and only if its kernel is 0.
201B Midterm Question 5. Prove that R is a free module over
itself.
Proof. R is unitary and unital, so 1 forms a basis of R over
itself.
Let M be another R-module which is a quotient of R and assume
that M is not isomorphic to R. Prove that there exists a non-zero ideal
I in R such that M R/I as R-modules.
Proof. Let : R M be a surjective R-module homomorphism.
Verify that ker is an ideal of R and use the rst isomorphism theorem.
Deduce that M is not free.

Proof. M , R implies ker is non-trivial. One would hope that
(1) would be a basis for M over R, but examining the image of a
non-trivial element of the kernel shows that the former is not linearly
independent.
201B Final Exam
R will always denote a unital ring and all modules over R will
be assumed to be unitary. All work should be shown and reasoning
explained. If you use a theorem, you should try to quote it: at least
the hypothesis and the conclusion which you need. All questions are
worth 10 points.
Final Exam Question 1. Let R = C[x] be the ring of polynomials
in one variable and let I be the ideal generated by (x 1)
3
and J the
ideal generated by (x2)
2
. Consider the corresponding left R-modules
M = R/I and N = R/J.
(1) Prove that N is not a simple module
(2) Prove that there is a surjective map of R-modules : M N
(3) Let K be the kernel of . Prove that K is not a direct sum-
mand of M, i.e. that there does not exists (sic) a submodule
K
1
of M such that M is the internal direct sum of K and K
1
.
Proof of (1). Submodules of N are in one-to-one correspondence
with ideals of R containing J ([2] IV 1.10). If J is not maximal then
there is a proper ideal J
of R such that J _ J
_ R and a proper
non-zero submodule N
of N such that N
= J
/J. To prove that J is

201B FINAL EXAM 181
not maximal it suces to show that (x 2)
2
is not irreducible to
exhibit factorization in R of (x 2)
2
whose terms are not units in R.
Well,
(x 2)
2
= (x 2)(x 2),
and x 2 / R
(it isnt a constant polynomial), so J must not be

maximal.
The above factorization corresponds to the ideal J
= (x 2) of R.
The corresponding submodule of N is
N
= r(x)(x 2) : r(x) R, (x 2) ,[r(x).

Proof of (2). The division algorithm allows M, N to be expressed
as follows
M =
_
a
0
+a
1
x +a
2
x
2
+
_
(x 1)
3
_
: a
i
C
_
N =
_
b
0
+b
1
x +
_
(x 2)
2
_
: b
i
C
_
.
The denition of also uses the division algorithm
: a
0
+a
1
x + a
2
x
2
+
_
(x 1)
3
_
b
0
+b
1
x +
_
(x 2)
2
_
where
a
0
+ a
1
x +a
2
x
2
= q(x)(x 2)
2
+ b
0
+b
1
x.
Straightforward is the verication that is an R-module homomor-
phism. Given b
0
+ b
1
x +
_
(x 2)
2
_
N arbitrary, the element b
0
+
b
1
x +0x
2
_
(x 1)
3
_
clearly is mapped by onto the former, hence is
surjective.
Proof. Inherent in the setup is a short exact sequence
0 K

M

N 0.
Recall the three equivalent conditions dening split exact
(1) is invertible
(2) is invertible
(3) There exist R-modules A, B and R-module isomorphisms , ,
such that
0
/
K
/
M
/
N
/
0
0
/
A

/
A B
/
B
/
0
is commutative.
Certainly condition (3) is equivalent to the statement K is a direct
summand of M. Hence to answer this question is suces to demon-
strate that one of the above properties fails.
It is simple to see that (2) fails, as an R-module homomorphism is
invertible if and only if it is injective if and only if its kernel is 0. But
4 4x +x
2
K 0
shows that the latter is false!
Final Exam Question 2. Let R be a commutative ring with iden-
tity. Prove that M is a cyclic module for R then M is either isomorphic
to R or must be a torsion module for R.
Proof. As a unitary R-module, R is free on 1. To say that
M is cyclic means M = Ra for some a M. Dene a map of sets
f : 1 M by f(1) = a; use the universal property of R (free) to
obtain an R-module homomorphism f : R M satisfying g(1) = a.
Observe that for all r R,
ra = rg(1) = g(r1) = g(r)
so that g is an epimorphism. If ker g = 0, then g : R M. Elseif
ker g ,= 0 then re-examining the equation displayed above with r
ker g 0 shows that a is not linearly independent, i.e. M is not free.
In this case, then, R , M.
Recall a theorem on nitely generated modules over a principal ideal
domain:
Theorem. Any nitely generated module over a principal ideal do-
main is isomorphic to a direct sum of a free module and a torsion
module.
Suppose now that R is a principal ideal domain and that M is gen-
erated by two elements. What are all the things you can say about M
by applying just this theorem?
Answer. This answer is weak. If M is torsion-free, then M is
free of rank at most two, so M R R or M R. If M is not
torsion-free, then M R R/(r) if exactly one of Ms generators is
linearly independent. In this case, if a denotes the other generator then
(r) = O
a
. If neither of Ms generators are linearly independent, then
M R/(r) R/(s) where (s) = O
b
.
Using the fact that principal ideal domains are unique factorization
domains, write r =
n
i=1
p
e
i
i
and s =
m
j=1
q
d
j
j
where p
i
, q
j
R are
201B FINAL EXAM 183
irreducibles (equiv. primes) and e
i
, d
j
are positive integers. Then the
last two cases of the previous paragraph become
R R
n
i=1
R
_
(p
e
i
i
)
R
n
i=1
R
_
(p
e
i
i
)
m
j=1
R
_
(q
d
j
j
)
respectively.
Final Exam Question 3. Suppose that R = C[x] and that M is
a cyclic module for R. Prove that M is either free of rank 1 or that M
is torsion.
Proof. a is a spanning set for M over R. If it is linearly inde-
pendent, then M has a basis of size 1 and is therefore free of rank 1.
Elseif a is not linearly independent, then there exists r R0 such
that ra = 0. Since M is cyclic, every m M has the form m = sa for
some s R. Apply r and use the commutativity of R
rm = r(sa) = s(ar) = s0 = 0.
Hence M is torsion.
Prove that any submodule of M is also cyclic.
Theorem. If G is a submodule of a free module F over a principal
ideal domain, then G is free and rank G rank F.
Proof. Choose a submodule N of M. Since M is cyclic, it has the
form M = Ra for some a M.
If M is free of rank 1, then by the theorem N is free of rank at most
1. If rank N = 1 then N = M and the result follows. If rank N = 0
then N = 0 and the result is trivial. Elseif M is torsion then there is
an isomorphism ([2], IV 6.4(iii))
M = Ra R/O
a
where O
a
< R is the order ideal of a. The R-submodules of R/O
a
are
in one-to-one correspondence ([2], IV 1.10) with the R-submodules of
R containing O
a
S/O
a
O
a
S R.
Certainly N is isomorphic to an R-submodule of R/O
a
, which means
that there exists an R-submodule S of R (as above) satisfying N
S/O
a
. Another application of ([2], VI 6.4(iii)) gives S/O
a
Sa, thus
N Sa,
proving that N is a cyclic R-module.
Suppose now that R = C[x, y]. Give an example of a cyclic module
for R and a submodule which is not cyclic.
Example. R is cyclic as an R-module over itself, since R 1 and
R = R1. The ideal (x, y) is an R-submodule of R which is not cyclic:
it must be shown that for all a (x, y) there exists b (x, y) such that
for all r R, b ,= ra. (Write down the denition of cyclic and negate
it.)
Suppose a = rx +sy (r, s R) is given. Note that both r, s cannot
be zero. If r = 0 and s ,= 0 then a = sy; use b = x. If r ,= 0 and s = 0
then a = rx; use b = y. If r, s ,= 0 then a = rx + sy; use b = x (or
b = y).
Final Exam Question 4. Let T : U W be a linear trans-
formation of nite-dimensional vector spaces with dim V = n, dim
W = m. Prove that there exist bases B of V and B
of W such that
the matrix of T with respect to these bases is of the form E
n,m
r
where
E
n,m
r
is the n m matrix with the identity r r matrix in the top
left-hand corner and zeros elsewhere.
Note that this is equivalent to proving that you can choose bases
(v
1
, . . . , v
n
) of V and (w
1
, . . . , w
m
) of W such that T(v
i
) = w
i
for
i = 1, . . . , r and T(v
i
) = 0 for i > r.
Proof. Let r = rank T. Recall the result stating
dimV = dimker T + dim
_
V/ ker T
_
.
By Sylvesters law of nullity, dim
_
V/ ker T
_
= r. Let X be a basis of
ker T, and Y be a basis of V/ ker T. Say
Y =
_
v
i
+ ker T : i = 1, . . . , r
_
.
To verify that
B = v
1
, . . . v
r
X = v
1
, . . . v
r
, v
r+1
, . . . , v
n
is a basis of V it suces to show that B is linearly independent. If
n
i=1
a
i
v
i
= 0, then
r
i=1
a
i
v
i
=
m
j=r+1
(a
j
)v
j
and since the term on the right belongs to ker T, so too does the term
on the left. But this means that, in V/ ker T,
r
i=1
(a
i
v
i
+ ker T) = 0 + ker T
201B FINAL EXAM 185
so since Y is linearly independent it follows that a
i
= 0 for all i =
1, . . . , r. Now
m
j=r+1
(a
j
)v
j
= 0
and the linear independence of X implies that a
j
= 0 for all j =
r + 1, . . . , m. Hence B is linearly independent.
Dene, for i = 1, . . . , r vectors
w
i
= T(v
i
) W.
It is straightforward to verify that the set w
1
, . . . , w
r
is linearly in-
dependent. Extend this set to a basis
B
= w
1
, . . . , w
r
, w
r+1
, . . . , w
m
of W and observe that T(v

i
) = w
i
for i = 1, . . . , r and T(v
i
) = 0 for
i > r. Accordingly
B
[T]
B
=
__
T(v
1
)
_
B
. . .
_
T(v
n
)
_
B
=
_
I
r
0
0 0
_
= E
n,m
r
as desired.
Final Exam Question 5. Suppose that
A =
_
0 1
1 0
_
over a eld K. What is its minimal polynomial if K is the eld of real
numbers?
Answer. The characteristic polynomial of A is p
A
(x) = x
2
+ 1
1[x]. As p
A
is irreducible over 1, the minimal polynomial of A must
be m
A
(x) = p
A
(x).
What is the Jordan canonical form of A when K is the eld of
complex numbers?
Answer. The eigenvalues of A are = i C. As A is 2 2 and
has 2 distinct eigenvalues, it is diagonalizable over C
A D =
_
i 0
0 i
_
.
Certainly J = D in this case.
Does it have a Jordan canonical form over the reals? If not, what
can one say about A? In other words, what kind of matrix is A similar
to?
Answer. There are no eigenvalues of A in 1, so J does not exist
here. However, A has a rational canonical form over R (it will be seen
shortly that A is already in this form).
Since m
A
(x) = x
2
+ 1 (over 1), A has a single invariant factor:
q = x
2
+ 1. By [2], VII 4.7 A is similar to the companion matrix of q
A C =
_
0 1
1 0
_
.
But this shows that A is already in rational canonical form, so there is
nothing else (that I can think of) to say about A.
Final Exam Question 6. Suppose that A is an nn matrix over
K and that K is an eigenvalue of A. Use the division algorithm
for K[x] and the denitions of the minimal polynomial and eigenvalues
to prove that (x ) divides the minimal polynomial of A.
Proof. Let q
1
, . . . , q
t
be the invariant factors of A; recall that
q
1
[q
2
[ . . . [q
t
and q
t
= q
A
is the minimal polynomial of A. Associated
with each q
i
is a unique companion matrix D
i
whose characteristic
polynomial is p
D
i
= q
i
. The matrix A has rational canonical form
A D =
t
i=1
D
i
where denotes matrix direct sum, i.e. block diagonal form. As A
and D are similar, their characteristic polynomials are equal. But
p
D
= p
D
1
. . . p
Dt
= q
1
. . . q
t
whence p
A
= q
1
. . . q
t1
q
A
.
The characteristic polynomial of A is dened as
p
A
(x) = det(A xI
n
).
An element K is an eigenvalue of A if and only if Av = v for
some non-zero v K
n
. Equivalently, is an eigenvalue of A if and
only if det(A I
n
) = 0, and since the evaluation homomorphism
K[x] f(x) f(0) K commutes with the determinant (cf. [2007
B5]), it follows that is an eigenvalue of A if and only if p
A
() = 0.
In other words the eigenvalues of A are exactly the roots of p
A
.
Given an eigenvalue of A, divide the characteristic polynomial by
(x )
p
A
(x) = q(x)(x ) + r, r C
and observe that p
A
() = 0 implies r = 0, meaning (x )[p
A
. Since
K[x] is a principal ideal domain ( K is a eld) and (x ) is irre-
ducible over K, it is prime in K[x]. Thus (x ) divides q
1
. . . q
t1
q
A
201B FINAL EXAM 187
implies that (x )[q
i
for some i. Each q
i
divides q
A
, whence (x
)[q
A
.
Suppose that A is a 66 matrix with entries in the complex numbers
and with minimal polynomial (x2)
2
(x+3). What are the eigenvalues
of A? Justify your answer.
Answer. The proof of the previous statement shows that p
A
and
q
A
have exactly the same roots (not considering multiplicity). As the
roots of p
A
are the eigenvalues of A, it is clear that in this case the
eigenvalues of A must be the roots of q
A
, i.e. = 3, 2.
What are all possible Jordan forms of A?
Answer. One theorem about Jordan canonical form states that
the size of the largest Jordan block of As Jordan form which corre-
sponds to an eigenvalue of A is exactly the multiplicity of as a root
of q
A
. So the largest block for 3 is 1 1 and the largest block for 2
is 2 2.
Given that q
A
= (x2)
2
(x+3), the possible lists of invariant factors
and their corresponding versions of p
A
are
(x 2), (x 2)
2
, q
A
= p
A
= (x 2)
5
(x + 3)
(x + 3), (x + 3)(x 2), q
A
= p
A
= (x 2)
3
(x + 3)
2
(x 2), (x 2)(x + 3), q
A
= p
A
= (x 2)
4
(x + 3)
2
.
Use the fact that the algebraic multiplicity of equals the sum of
the sizes of the Jordan blocks corresponding to to get the three
possibilities
J J
2
(2) J
2
(2) J
1
(3) J
1
(3)
J J
2
(2) J
2
(2) J
1
(2) J
1
(3)
J J
2
(2) J
1
(2) J
1
(3) J
1
(3) J
1
(3)
Final Exam Question 7. Let
A =
_
(1 + t) (1 t)
t (t + 2)
_
be a 2 2 matrix with entries in the polynomial ring in one variable.
Is A an invertible matrix?
Answer. The contraposition to [2] IV 3.5(iv) is if det A is not a
unit in the underlying ring, then A is not invertible. Well,
det A = 2(t
2
+t + 1)
and since this is a polynomial of positive degree there is no chance that
it is a unit. Hence A is not invertible.
Let B and C be 22 matrices with entries in the complex numbers.
Prove that B +C is invertible for all but nitely many values of .
Proof. Say
B =
_
a b
c d
_
. C =
_
a
_
.
Direct computation shows that det B +C =
2
(. . . ) +(. . . ) +(. . . ).
Since C is algebraically closed, there are , C such that
det A = ( )( ).
Recall that a matrix over C is invertible if and only if its determinant is
nonzero. Accordingly, B +C is invertible if and only if ,= , .
Part C
There were no midterms in Part C. Included here are the corrected
homework problems.
Assignment 1
Graded III 3 #3, III 6 #1, 10
III.3 #3. Let R be the subring a + b
10 : a, b Z of the eld
of real numbers.
(a) The map N : R Z given by
N : a + b
10 a
2
10b
2
satises N(uv) = N(u)N(v) for all u, v R and N(u) = 0 if and only
if u = 0.
Proof. Say u = a +b
10 and v = c +d
10.
N(uv) = N(u)N(v) for all u, v R
We have
N(uv) = N
_
(a +b
10)(c + d
10)
_
= N
_
(ac + 10bd) + (ad +bc)
10
_
= (ac + 10bd)
2
10(ad +bc)
= a
2
c
c
10a
2
d
2
10b
2
c
2
+ 100b
2
d
2
= (a
2
10b
2
)(c
2
10d
2
)
= N(u)N(v).
ASSIGNMENT 1 189
N(u) = 0 if and only if u = 0
Observe that N(u) = a
2
10b
2
= 0 if and only if a
2
= 10b
2
.
The only integer solution of this equation is (a, b) = (0, 0).
(b) u is a unit in R if and only if N(u) = 1.
Proof. If uv = 1 then 1 = N(1) = N(uv) = N(u)N(v), which
implies N(u) = 1 = N(v). Conversely, if [N(a + b
10)[ = 1 then
1 = a
2
10b
2
= (a +b
10)(a b
10)
so a + b
10 is a unit.
(c) 2, 3, 4, 4 +
10, 4
10 are irreducible elements of R.

Proof. If 2 = uv for some non-units u, v R then
4 = N(2) = N(u)N(v)
implies N(u) = 2 = N(v). But this means there are integers a, b
satisfying a
2
10b
2
= 2, which is not true. A similar argument works
for the case 3 = uv where u, v are non-units. In this case we are led
to the equation a
2
10b
2
= 3, which also has no integer solutions. For
4 +
10 = uv (resp. 4
10 = uv) with u, v non-units we nd that

N(u), N(v) = 2, 3 (resp. = 2, 13) which again leads to equations
with no integer solutions.
Therefore, it must be true that 2, 3, 4, 4 +
10, 4
10 are irre-
ducible elements of R.
(d) 2, 3, 4, 4 +
10, 4
10 are not prime elements of R.

Proof. Denote J = 2, 3, 4, 4 +
10, 4
10 and observe that

no element of J divides any other element of J. For example, 2 does
not divide 4 +
10, for in this case there would be u R such that

2u = 4 +
10. But then

4N(u) = N(2)N(u) = N(4 +
10) = 6
implies 4[6, false. Note also that
2 3 = 6 = (4 +
10)(4
10).
For any x J the above equation gives examples of y, z R such that
x
yz but x ,
y and x ,
z, which is the logical negation of the denition

of x is prime.
x y z
2 4 +
10 4
10
3 4 +
10 4
10
4 +
10 2 3
4
10 2 3
Therefore, no element of J is prime.
III.3 #4.
Show that in the integral domain of Exercise 3 every element can be
factored into a product of irreducibles, but this factorization need not
be unique (in the sense of Denition 3.5(ii)).
Proof. Begin with a sketch of why each element can be factored
into a product of irreducibles. Let a + b
10 R be given. If it is
irreducible then halt. Elseif it is not irreducible, then factor
a +b
10 = u
1
v
1
(u
1
, v
1
/ R
).
If one of u
2
, v
2
are irreducible, then halt. If not, taking norms gives
a
2
10b
2
= N(u
1
)N(v
2
) Z. Since Z is a UFD, N(n
1
) has a prime-
power factorization
N(u
1
) = p
e
1
1
p
en
n
where p
i
are prime and e
i
are positive integers.
Factor u
1
= u
2
v
2
, where u
2
, v
2
/ R
. Take norms again to see that

N(u
2
)[N(u
1
). Since v
2
is not a unit, N(v
2
) ,= 1 so N(u
2
) ,= N(u
1
).
By design, the divisors of N(u
2
) are a proper subset of the divisors of
N(u
1
).
Continuing this process gives a sequence of elements u
j
, the divisors
of whose norms form a strictly decreasing sequence of subsets of the
set of divisors of N(u
1
). Certainly there will eventually be an m such
that u
m
= u
m+1
v
m+1
and N(u
m+1
) = 1, (at this point the divisors of
N(u
1
) are exhausted). This shows that u
m
is irreducible, and since
a +b
10 = u
1
v
1
= = u
m
v
m
v
m1
. . . v
1
it is clear that a +b
10 has an irreducible factor. Iterating this argu-

ment will produce an irreducible factorization of a +b
10.
Every x R can be factored into a product of irreducibles, but
2 3 = 6 = (4 +
10)(4
10)
shows that said factorization need not be unique.
III.3 #6.
ASSIGNMENT 1 191
(a) If a and n are integers, n > 0, then there exist integers q and r
such that a = qn +r, where [r[ n/2.
Proof. The ordinary division algorithm gives integers p and s such
that a = pn + s, where 0 s < n. If s n/2 the proof is complete,
use (q, r) = (p, s). Elseif n/2 < s < n then use (q, r) = (p + 1, s n):
(p + 1)n + (s n) = pn +n + s n
= pn +s
= a
and n/2 < s < n implies n/2 < s n < 0 (so [s n[ n/2).
(b) The Gaussian integers Z[i] form a Euclidean domain with (a+
bi) = a
2
+ b
2
.
Lemma. The function satises the following properties
(1) (a +bi) = 0 if and only if a + bi = 0;
(2)
_
(a + bi)(c +di)
_
= (a +bi)(c + di);
(3) a +bi is a unit if and only if (a +bi) = 1;
(4) if (a + bi) is prime, then a +bi is irreducible in Z[i].
Proof sketch. If x = a +bi and y = c +di then
(1) a
2
+b
2
= 0 implies a
2
= b
2
, which necessitates a = 0 = b.
(2)
(xy) =
_
(ac bd) + (ad + bc)i
_
= (ac bd)
2
+ (ad +bc)
2
= a
2
c
2
2abcd +b
2
d
2
+a
2
d
2
+ 2abcd +b
2
c
2
= a
2
c
2
+a
2
d
2
+ b
2
c
2
+b
2
d
2
= (a
2
+b
2
)(c
2
+ d
2
)
= (x)(y).
(3) If xy = 1 then 1 = (1) = (xy) = (x)(y) implies (x) =
1 = (y).
If (a + bi) = 1 then 1 = a
2
+ b
2
= (a + bi)(a bi) shows
that a +bi is a unit.
(4) Follows directly from (2) and (3).
Proof of (b). That

_
(a+bi)(c+di)
_
(a+bi)(c+di) follows
directly from property (2) above. If x, y Z[i] and y ,= 0 then xy
1
[i], the eld of quotients of Z[i]. Say xy

1
= s + ti, where s, t .
Now let m be the integer nearest s, and let n be the integer nearest
t. (These integers may not be uniquely determined, but that does not
matter). Thus, [ms[ 1/2 and [n t[ 1/2. Then
xy
1
= s +ti = (mm+s) + (n n + t)i
= (m+ni) +
_
(s m) + (t n)i
_
.
So,
x = (m+ni)y +
_
(s m) + (t n)i
_
y.
The division condition of the denition of a Euclidean domain is satis-
ed with q = m+ni and
r =
_
(s m) + (t n)i
_
y.
Clearly q Z[i], and since r = x qy, we also know r Z[i]. Finally,
(r) =
_
(s m) + (t n)i
_
(y)
=
_
(s m)
2
+ (t n)
2
_
(y)
_
1
4
+
1
4
_
(y) < (y)
III.3 #7. What are the units in the ring of Gaussian integers Z[i]?
Solution. The previous result tells us that x = a + bi is a unit
in Z[i] if and only if (x) = a
2
+ b
2
= 1. As the square of any integer
is non-negative, if (x) is to equal 1 we must have a = 0 xor b = 0.
These cases correspond to x = i, 1, so
Z[i]
= i, 1.
III.6 #1.
(a) If D is an integral domain and c is an irreducible element in
D, then D[x] is not a principal ideal domain.
Proof. Consider the ideal I = (x, c) of D[x]. If there exists h
D[x] with I =
_
h(x)
_
then, since c, x I there are f(x), g(x) D[x]
satisfying
c = h(x)f(x) x = h(x)g(x).
As D is an integral domain, the rst indicates
0 = deg c = deg h + deg f,
which means h(x) is a constant polynomial. Say h(x) = a, then
c = af(x) x = ag(x)
Using the irreducibility of c, the equation c = af(1) implies that a or
f(1) is a unit. Now a cannot be a unit because in that case I = D[x],
which is false (witness x
2
D[x] I). But then there is no way for the
ASSIGNMENT 1 193
polynomial ag(x) to be monic, so the equation x = ag(x) is not valid.
In conclusion, there can exist no such h.
(b) Z[x] is not a principal ideal domain.
Proof. This is a corollary of part (a).
(c) If F is a eld and n 2, then F[x
1
, . . . , x
n
] is not a principal
ideal domain.
Proof. Observe that
_
F[x
1
, . . . , x
n1
]
_
[x
n
] = F[x
1
, . . . , x
n
].
Noting that F[x
1
, . . . , x
n1
] is an integral domain, it suces to show
that x
n
is irreducible over F[x
1
, . . . , x
n1
]. Appealing to (a) will com-
plete the proof.
If there exist f, g F[x
1
, . . . , x
n1
] such that x
n
= fg then
1 = deg x
n
= deg f + deg f.
Accordingly, either f or g is a constant polynomial. Since F[x
1
, . . . , x
n1
]
is an integral domain and x
n
,= 0, it follows that neither f nor g is the
zero polynomial. Since F is a eld, all non-zero constant polynomi-
als are units in F[x
1
, . . . , x
n1
]. Consequently, x
n
is irreducible, as
desired.
III.6 #2.
If F is a eld and f, g F[x] with deg g 1, then there exist
unique polynomials f
0
, f
1
, . . . , f
r
F[x] such that deg f
i
< deg f
i+1
for
all i and
f = f
0
+f
1
g + f
2
g
2
+ +f
r
g
r
.
.
Proof. Note that F[x] is a euclidean domain. Divide f by g to
obtain unique polynomials f
0
, q
0
F[x] satisfying
f = f
0
+ q
0
g, deg f
0
< deg g.
Iterate, dividing q
0
by g to obtain unique polynomials f
1
, q
1
F[x]
satisfying
f = f
0
+ (f
1
+ q
1
g)g, deg f
1
< deg g.
Continuing, this process will eventually terminate, giving polynomials
f
r1
, f
r
D[x] satisfying
f = f
0
+ [f
1
+ + (f
r1
+f
r
g)g . . . ]g, deg f
r
< deg g.
III.6 #3. Let f be a polynomial of positive degree over an integral
domain D.
(a) If char D = 0 then f
,= 0.
Proof. If f =
n
j=1
a
j
x
j
then its formal derivative is
f
(x) = a
1
+ 2a
2
x + + (n 1)a
n1
x
n2
+na
n
x
n1
.
If deg f = 1 then f
(x) = a
1
, so if f
(x) = 0 then deg f ,= 1, which is a

contradiction. If deg f 2 and f
(x) = 0 then in particular 2a

2
x = 0
for all x D, contradictory to the assumption that char D = 0.
(b) If char D = p ,= 0, then f
= 0 if and only if f is a polynomial

in x
p
(that is, f = a
0
+a
p
x
p
+a
2p
x
2p
+ +a
np
x
np
).
Proof. () If
f = a
0
+a
p
x
p
+a
2p
x
2p
+ +a
np
x
np
then
f = (p)a
p
x
p1
+ (2p)a
2p
x
2p1
+ + (np)a
np
x
np1
and for each d D we have
f(d) = p(a
p
d
p1
) +p(2a
2p
d
2p1
) + +p(na
np
d
np1
) = 0 + 0 + + 0
since char D = p.
() Let f =
m
i=1
a
i
x
i
. The only non-zero terms of f are those
with p ,[a
i
and a
i
,= 0. Call these coecients b
j
, for j = 1, . . . , k (thus
the problem is reduced to considering a polynomial whose coecients
are non-zero and not divisible by p). The assumption is that f
= 0,
which means that jb
j
= 0 for all j. Since D is a domain, it follows that
j = 0 for all j, meaning that p[j for all j. Say j = pq
j
; now it is clear
that the only powers of x appearing in f are x
j
= x
pq
j
. In other words
f is a polynomial in x
p
.
III.6 #10.
(a) Let D be an integral domain and c D. Let
f(x) =
n
i=1
a
i
x
i
D[x] and f(x c) =
n
i=1
a
i
(x c)
i
D[x].
Then f(x) is irreducible in D[x] if and only if f(xc) is irreducible in
D[x].
Proof. This is immediate upon replacing x by x+c in each factor:
f(x) = g(x)h(x) f(x + c) = g(x +c)h(x +c)
If f = gh were a non-trivial factorization of f in D[x], then f(x +c) =
g(x + c)h(x + c) would be a non-trivial factorization of f(x + c) in
D[x].
ASSIGNMENT 1 195
(b) For each prime p the cyclotomic polynomial
f = x
p1
+x
p2
+ +x + 1
is irreducible in Z[x].
Proof. Observe that, by the binomial theorem,
F(x) = f(x+1) =
(x + 1)
p
1
x 1
= x
p1
+
_
p
1
_
x
p2
+
_
p
2
_
x
p3
+ +
_
p
1
_
.
Since every coecient except that of x
p1
is divisible by p, by Eisen-
steins criterion F is irreducible in Z[x]. It follows from part (a) with
D = Z and c = 1 that f is irreducible in Z[x].
Additional Exercise 1.
Let R be a unique factorization domain. Show that every irreducible
element c R is prime.
Proof. Suppose a, b R satisfy c
ab, i.e. suppose there exists

x R with cx = ab. Assume c ,[a. Factor a and b into irreducibles:
a = k
1
k
n
and b = l
1
l
m
(noting that these are unique up to
associates) to get
cx = k
1
k
n
l
1
l
m
.
By uniqueness, c must be associate to some k
i
or some l
j
; as c does
not divide a we must have uc = l
j
for some u R
and some j
1, , m. Thus
b = l
1
l
j1
(uc)l
j+1
l
m
which means c
b. Accordingly, c is prime in R.
Exercise 2.
R is an integral domain and p is a non-zero element of R. Show
that p is prime if and only if (p) is a prime ideal in R.
Proof. () Assume that p is prime. As p a non-zero non-unit in
R, the principal ideal (p) generated by p is a proper ideal of R. Suppose
a, b R satisfy ab (p); this gives (ab) (p). Theorem III 3.2(i) gives
p
ab, and by hypothesis p
a or p
b. Another application of III 3.2(i)

gives (a) (p) or (b) (p). Therefore (p) is a prime ideal in R.
() Assume that (p) is a prime ideal. First note that R as
an integral domain is commutative, also recall that by denition
(p) ,= R and for all a, b R: ab (p) implies a (p) or b (p). The
statements ab (p), a (p), and b (p) yield the statements
(ab) (p), (a) (p), and (b) (p) respectively. Accordingly,
the statement ab (p) implies a (p) or b (p) yields the statement
(ab) (p) implies (a) (p) or (b) (p). By III 3.2(i), the nal
statement is equivalent to the denition of p is prime.
Exercise 3.
Let a
1
, , a
n
be elements of a commutative ring R. Show that if R
is a unique factorization domain then there exists a greatest common
divisor of a
1
, , a
n
.
Proof. Each a
i
(i = 1, , n) has a factorization
a
i
= c
m
i,1
1
c
m
i,t
t
for distinct irreducibles c
j
R and non-negative integers m
i,j
. Dene,
for each j = 1, , t, a non-negative integer k
j
= minm
i,j
: i =
1, , n. The candidate for a greatest common divisor of a
1
, , a
n
is
d = c
k
1
1
c
kt
t
.
There are two conditions to verify
d
a
i
for all i = 1, , n.
For j = 1, t and i = 1, , n let n
i,j
= m
i,j
k
j
; note
n
i,j
0 for all i, j. For i = 1, , n let q
i
= c
n
i,1
1
c
n
i,t
t
. Then
dq
i
= c
k
1
1
c
kt
t
c
n
i,1
1
c
n
i,t
t
= a
i
shows that d
a
i
for all i.
If b
a
i
for all i then b
d.
The hypothesis b
a
i
for all i means (by unique factoriza-
tion) that b = c
l
i,1
1
c
l
i,t
t
for l
i,j
m
i,j
. The uniqueness of
this factorization implies that l
i,j
is independent of i, so we
can write l
i,j
= l
j
. The for all i portion of the hypothesis
forces l
j
k
j
for all j; thus h
j
:= k
j
l
j
0. But now
d = c
l
1
1
c
lt
t
c
h
1
1
c
ht
t
= bq
shows that b
d.
Assignment 2
Graded VI 1 #7, 17, Additional Exercise 5.
Unless specied otherwise F is always an extension eld of the eld
K.
VI.1 #1.
ASSIGNMENT 2 197
(a) [F : K] = 1 if and only if F = K.
Proof. (): Assume that F has dimension 1 as a vector space
over K, in other words there exists a basis of one element for F over
K. Theorem IV.2.1 and its proof show that F is isomorphic to a direct
sum of copies of K indexed by the basis, but this just means that F is
isomorphic to K. As K F we must have F = K.
(): If F = K then the set 1
K
forms a basis of F over K. But
this just means that [F : K] = 1.
(b) If [F : K] is prime, then there are no intermediate elds between
F and K.
Proof. Let E be an intermediate eld between F and K, i.e.
suppose F E K. By Theorem 1.2, [F : K] = [F : E][E : K]; as
divisors of a prime it must be that either [F : E] = 1 or [E : K] = 1.
According to part (a), this means that F = E or E = K.
(c) If u F has degree n over K, then n divides [F : K].
Proof. We have a tower of elds
K K(u) F
which leads to [F : K] = [F : K(u)][K(u) : K] = [F : K(u)]n.
V.1 #3.
If u
1
, . . . , u
n
F, then the eld K(u
1
, . . . , u
n
) is (isomorphic to)
the quotient eld of the ring K[u
1
, . . . , u
n
].
Proof. Denote the eld of quotients of K[u
1
, . . . , u
n
] by
E =
_
f(u
1
, . . . , u
n
)
g(u
1
, . . . , u
n
)
: f, g K[u
1
, . . . , x
n
]; g(u
1
, . . . , u
n
) ,= 0
_
,
(we use representatives to stand in for the equivalence classes obtained
in Es construction freely) and observe that any eld containing both
K and u
1
, . . . , u
n
must also contain E. Thus E K(u
1
, . . . , u
n
). As
for the reverse inclusion, we note that E is a eld containing K and
u
1
, . . . , u
n
and obtain immediately that K(u
1
, . . . , u
n
) E.
V.1 #7.
If v is algebraic over K(u) for some u F and v is transcendental
over K, then u is algebraic over K(v).
Remark (Idea of the proof). Take the minimal polynomial f over
K(u) of which v is a root. use this to build a polynomial g over K(v)
of which u is a root. that this can be done must have something to do
with the fact that v is not the root of ANY polynomial over K.
Proof. Let f K(u)[x] be the monic irreducible polynomial of
degree n uniquely determined by the conditions f(v) = 0 and for all
g K(u)[x], g(v) = 0 if and only if f divides g. Say
f(x) = x
n
+
n1
x
n1
+ +
1
x +
0
,
i
K(u)
so that 0 = v
n
+
n1
v
n1
+ +
1
v +
0
. Each
i
(0 i n 1)
has the form
i
=
h
i
(u)
k
i
(u)
, h
i
, k
i
K[x]; k
i
(u) ,= 0.
Accordingly,
0 = v
n
+
h
n1
(u)
k
n1
(u)
v
n1
+ +
h
1
(u)
k
1
(u)
v +
h
0
(u)
k
0
(u)
.
Let
s
n
=
n1
j=0
k
j
(u) s
i
=
n1
j=0,j=i
k
j
(u)
(for 0 i n 1) so that
s
n
f(x) =
n1
i=0
s
i
x
i
0 =
n1
i=0
s
i
v
i
.
Expand s
i
=
m(i)
j=0
c
ij
u
j
with c
ij
K so that
0 =
n1
i=0
s
i
v
i
=
i
_
_
m(i)
j=0
c
ij
u
j
_
_
v
i
=
j
c
ij
u
j
v
i
=
j
c
ij
v
i
u
j
=
j
_
i
c
ij
v
i
_
u
j
=
j
r
j
(v)u
j
.
Since v is transcendental over K, r
j
,= 0 for all j. Thus
g(x) =
j
r
j
(v)x
j
K(v)[x]
is a polynomial over K(v) of which u is a zero.
ASSIGNMENT 2 199
Example (Example illustrating the idea of the proof). Say f(x) =
x
2
+
1
x +
0
, where
i
=
h
i
(u)
k
i
(u)
. Then
f(x) = x
2
+
h
1
(u)
k
1
(u)
x +
h
0
(u)
k
0
(u)
0 = v
2
+
h
1
(u)
k
1
(u)
v +
h
0
(u)
k
0
(u)
.
Clear fractions to get
0 = k
1
(u)k
0
(u)
. .
s
2
(u)
v
2
+k
0
(u)h
1
(u)
. .
s
1
(u)
v +k
1
(u)h
0
(u)
. .
s
0
(u)
and write s
i
(u) = c
i2
u
2
+c
i1
u +c
i0
for c
ij
K (assuming m(i) = 2 for
each i, for simplicity). Then
0 = (c
22
u
2
+c
21
u +c
20
)v
2
+ (c
12
u
2
+c
11
u +c
10
)v + (c
02
u
2
+c
01
u +c
00
)
= (c
22
v
2
+c
12
v + c
02
)u
2
+ (c
21
v
2
+c
11
v +c
01
)u + (c
20
v
2
+ c
10
v +c
00
)
so that
g(x) = (c
22
v
2
+c
12
v +c
02
)x
2
+(c
21
v
2
+c
11
v +c
01
)x+(c
20
v
2
+c
10
v +c
00
).
V.1 #14a.
If F = (
2,
3), nd [F : ] and a basis of F over .

Proof. Since
3 is of degree 2 over the degree of the extension

(
2,
3) over (
2) is at most 2. It is equal to 2 if and only if x

2
3 is
irreducible over (
2). As this polynomial has degree 2, it is reducible

if and only if it has a root in (
2), i.e. if
3 (
2). Suppose there

are a, b such that a+b
2 =
3. Then 3 = (a
2
+2b
2
)+2ab
2, and
if ab ,= 0 then this equation can be solved for
2 in terms of a, b. This
implies that
2 , false. If a = 0 then
3 = b
2, and multiplying
by
2 gives
6 , false. Thus
3 / (
2), which gives [F : ] = 4.

Elements in this eld include 1,
2,
3,
6; by the computations
above this set forms a basis for F over .
V.1 #17.
Find an irreducible polynomial f of degree 2 over the eld Z
2
. Ad-
join a root u of f to Z
2
to obtain a eld Z
2
(u) of order 4. Use the same
method to construct a eld of order 8.
Proof. The polynomial g(x) = x
2
+x+1 has no roots in Z
2
, so it
is irreducible over Z
2
. Any root u of g(x) is algebraic of degree 2 over
Z
2
, so 1, u is a basis for Z
2
(u) over Z
2
:
Z
2
(u) = a +bu : a, b Z
2
.
Arithmetic in this eld is carried out according to the rule u
2
+u+1 = 0.
By design #Z
2
(u) = 4. Alternatively,
Z
2
(u)
Z
2
[x]
(x
2
+ x + 1)

_
a + bx + (x
2
+x + 1) : a, b Z
2
_
.
The polynomial f(x) = x
3
+ x + 1 also has no roots in Z
2
and is
thus irreducible over Z
2
. Any root v of f is algebraic of degree 3 over
Z
2
, so 1, v, v
2
is a basis of Z
2
(v) over Z
2
:
Z
2
(v) = a +bv + cv
2
: a, b, c Z
2
.
Arithmetic in this eld is carried out according to the rule v
3
+v+1 = 0.
By design #Z
2
(v) = 8. Alternatively,
Z
2
(v)
Z
2
[x]
(x
3
+x + 1)

_
a +bx +cx
2
+ (x
3
+x + 1) : a, b, c Z
2
_
.
V.1 #19.
If u, v F are algebraic over K of degrees m and n respectively,
then
[K(u, v) : K] mn.
If m and n are relatively prime, then equality holds.
Proof. The sets U = 1, u, . . . , u
m1
and V = 1, v, . . . , v
n1
are bases for K(u) and K(v) over K, respectively. Regarding K(u, v) =
K(u)(v), V is a spanning set for K(u)(v) over K(u). Accordingly,
[K(u, v) : K(u)] m = [K(v) : K] with equality if and only if V
is linearly independent over K(u). Since [K(u, v) : K] = [K(u, v) :
K(u)][K(u) : K], the desired estimate holds. The extension degree
[K(u, v) : K] is divisible by both m and n since K(u) and K(v) are
subelds of K(u, v); hence it is divisible by [m, n] = lcm(m, n). Thus
[m, n] [K(u, v) : K]. If (m, n) = gcd(m, n) = 1 then [m, n] = mn
and the desired equality is true.
Let K E F be elds. Show that if E is a nite dimensional
extension of K and F is a nite dimensional extension of E then F is
a nite dimensional extension of K and
[F : K] = [F : E][E : K].
.
ASSIGNMENT 2 201
Proof. Let
| = u
i
: i = 1, . . . , n 1 = v
j
: j = 1, . . . , m
be bases for F over E and E over K, respectively. Dene J = uv :
u |, v 1 and observe that all Js elements are distinct: if
uv = u
then uv u
= 0 and since | is linearly independent

over E it follows that u = u
= 0. It will be enough to prove that W is

a basis of F over K, for once this is established we will have
[F : K] = #J = #| #1 = [F : E][E : K].
If f F, then f =
n
i=1
e
i
u
i
where e
i
E, u
i
| since | spans
F as a vector space over E. Because E is a vector space over K,
each e
i
can be expressed as e
i
=
m
j=1
k
ij
v
j
where k
ij
K, v
j
1.
Substituting for e
i
in the penultimate equation gives
f =
n
i=1
_
m
j=1
k
ij
v
j
_
u
i
=
j
k
ij
v
j
u
i
,
whence J spans F as a vector space over K.
As for linear independence, suppose k
ij
K satisfy
j
k
ij
v
j
u
i
=
0. For each i, let e
i
=
m
j=1
k
ij
v
j
E. Then
0 =
i
_
j
k
ij
v
j
u
i
_
=
i
e
i
u
i
and the linear independence of | over E implies e
i
= 0 for all i. In
turn, the linear independence of 1 over K implies that k
ij
= 0 for all
i, j. Consequently, J is linearly independent over K.
Exercise 2.
Let u be a real root of the irreducible polynomial x
3
3x1 [x].
We know that [u] is a eld. Compute the inverse of u
3
3u 1 in
[u].
Proof. For general interest, observe that u =
3
13
2
. By Theorem
1.6, the degree of u over is [(u) : ] = 3, which means 1, u, u
2
is a basis for (u) as a vector space over . Consider u

4
+ 2u
2
+ 3
(u)
_
= [u]
_
. We can express this polynomial in terms of the above
basis by rst using polynomial division in [x] and then passing to
(u) as follows
x
4
+ 2x
2
+ 3 = (x 2)(x
2
3x 1) + 3x
2
+ 7x + 5
which implies
u
4
+ 2u
2
+ 3 = (u 2)(u
2
3u 1) + 3u
2
+ 7u + 5
= 3u
2
+ 7u + 5.
Now we can compute the multiplicative inverse of 3u
2
+ 7u + 5 in
(u). Because x
2
3x 1 is irreducible in [x], we know that it is
relatively prime to 3x
2
+7x+5. As their greatest common divisor equals
1, there exists a [x]-linear combination of these two polynomials which
equals 1; that is to say there exist g(x), h(x) [x] such that
(x
2
3x 1)g(x) + (3x
2
+ 7x + 5)h(x) = 1.
From this we have, in (u),
1 = (u
2
3u 1)g(u) + (3u
2
+ 7u + 5)h(u)
= (3u
2
+ 7u + 5)h(u).
So h(u) is the multiplicative inverse of 3u
2
+7u+5 in (u). The poly-
nomial h(x) can be determined explicitly from the euclidean algorithm:
h(x) =
7
111
x
2
26
111
x +
28
111
so that
h(u) =
7
111
u
2
26
111
u +
28
111
.
Exercise 3.
Lindemanns theorem states that if
1
, . . . ,
n
are distinct algebraic
numbers and A
1
, . . . , A
n
are non-zero algebraic numbers, then the equa-
tion
A
1
e
1
+ +A
n
e
n
= 0
cannot hold. Using Lindemanns theorem show that for every non-zero
algebraic number the number e
is transcendental.
Proof. If e
is algebraic, then there exists f(x) = a

n
x
n
+ +
a
1
x + a
0
such that
(1) a
n
e
n
+a
n1
e
(n1)
+ + a
1
e
+a
0
= 0.
Observe that a
i
and i are algebraic for all i = 0, . . . , n since g
i
= xa
i
has a
i
as a root, and if h(x) = b
m
x
m
+ +b
0
has as a root then
h
i
(x) =
b
m
i
m
x
m
+ + b
1
x +b
0
has i as a root. Consequently, from Lindemanns theorem (along with
the observation that a
0
= a
0
e
0
) it follows that (1) cannot hold, which
is a contradiction. Therefore e
is transcendental.
ASSIGNMENT 3 203
Exercise 4. If F is a nite dimensional extension eld of K, then
F is nitely generated and algebraic over K.
Proof. Suppose F is a n-dimensional extension eld of K, i.e.
[F : K] = n. Specically, F has dimension n as a vector space over
K. Given u F, the n + 1-subset 1
K
, . . . , u
n
of F must be linearly
dependent over K. Thus, there exists (a
1
, . . . , a
n
) K
n
(0, . . . , 0)
such that
n
i=1
a
i
u
i
= 0.
Evidently, f(x) =
i
a
i
x
i
K[x] is a polynomial with u as a root;
in other words u is algebraic over K. As u was chosen arbitrarily, it
follows that F is algebraic over K.
As F is a n-dimensional K-vector space, there exists a basis v
i
:
1 i n of F over K. Accordingly, given u F we have a
i
K
such that u =
i
a
i
v
i
. But this means F = K(v
1
, . . . , v
n
).
Exercise 5.
Let K E F be elds. Show that if E is an algebraic extension
of K and F is an algebraic extension of E then F is an algebraic
extension of K.
Proof. Let u F be given. By hypothesis, u is algebraic over
E, say of degree n. In particular, u is algebraic over the extension
eld K(e
0
, . . . , e
n1
) of K, where e
i
are the coecients of us minimal
polynomial in E[x]. There is a tower of elds
K K(e
0
, . . . , e
n1
) K(e
0
, . . . , e
n1
)(u)
where
[K(e
0
, . . . , e
n1
)(u) : K(e
0
, . . . , e
n1
)] = n
and, since each e
i
is algebraic over K (by hypothesis)
[K(e
0
, . . . , e
n1
) : K] = m < .
Consequently, [K(e
0
, . . . , e
n1
)(u) : K] = nm < and it follows that
K(e
0
, . . . , e
n1
)(u) K is algebraic. In particular u is algebraic over
K, and since u was chosen arbitrarily it is clear that F K is algebraic.
Assignment 3
Dont know which problems were graded.
V.2 #3.
If 0 d , then Aut
d) is the identity or is isomorphic to

Z
2
.
Proof. Case d is a square in . Here (
d) = , and certainly
Aut
consists only of the identity automorphism.

Case d is not a square in . Here
d is algebraic of degree 2 over

(with minimal polynomial p [x]), so
(
d) = a +b
d : a, b .
Any Aut
d) acts as follows
(a +b
d) = (a) + (b)(
d) = a + b(
d).
Accordingly, any such is completely and uniquely determined by its
image of
d. But (
d) must also be a root of p, so the only options are

(
d)
d,
d. Clearly then, in this case Aut
d) = 1
,
where (
d) =
d.
V.2 #4.
What is the Galois group of (
2,
3,
5) over ?
Remark (Strategy). Find the dimension of the extension and write
out a basis. The order of the Galois group is bounded from above by
the extensions dimension. Once in possession of this estimate, look
at the action of any -automorphism on a general element (using the
basis). Count.
Proof. First observe that, by V.1.2,
[(
2,
3,
5) : Q] = [(
2,
3,
5) : (
2,
3)][(
2,
3) : Q].
Recall from previous work that [(
2,
3) : Q] = 4. Tedious and
elementary algebra shows that
5 / (
2,
3). Now
5 is algebraic
of degree 2 (p(x) = x
2
5) over . And since the only roots of p are
5 / (
2,
3), it follows that

[(
2,
3,
5) : (
2,
3)] = 2.
So [(
2,
3,
5) : Q] = 8 and a basis for (
2,
3,
5) over is
1,
2,
3,
5,
6,
10,
15,
30
Consequently,
Aut
2,
3,
5)
8.
ASSIGNMENT 3 205
Any f Aut
2,
3,
5) is completely and uniquely deter-

mined by the images f(
2), f(
3), f(
5), but
2,
3,
5 must be
sent by f to other roots of x
2
2, x
2
3, x
2
5 respectively. So
f(
2) =
2, f(
3) =
3, f(
5) =
5
dene eight -automorphisms of (
2,
3,
5). A tedious and rou-

tine verication shows that these eight maps are distinct. Thus
Aut
2,
3,
5)
= 8.
Accordingly, (
2,
3,
5) is isomorphic to one of Z
8
, Z
4
Z
2
, or
Z
2
Z
2
Z
2
. Observing that each automorphism described above has
order 2, it is clear that Aut
2,
3,
5) Z
2
Z
2
Z
2
.
In the additional exercises F M L K are elds, G = Aut
K
F
and H is a subgroup of G. The prime operation is dened by
L
= Aut
K
F : u L (u) = u
H
= v F : H (v) = v.
Show that L
= L
and H
= H
.
Lemma. L L
and H < H
.
Proof. L L
because given u L and L
it follows that
(u) = u. H < H
because given H and v H
it follows that
(v) = v.
Definition.
L
=
_
Aut
K
F : v L
(v) = v
_
H
=
_
v F : H
(v) = v
_
Corollary. L
< L
and H
.
Proof of Additional Exercise 1. L
< L because given

L
and u L the Lemma gives u L
. Now it follows that (u) = u.

H
H because given v H
and H the Lemma gives

H
. Now it follows that (v) = v.

Show that if [M : L] < then so is [L
: M
] and
[L
: M
] [M : L].
In particular, if F is a nite-dimensional extension of K then G is
nite and
#G [F : K].
.
Proof. Proceed by strong induction on [M : L] = m.
Base case When [M : L] = 1 we have M = L and the result
is trivial.
Inductive case Assume the result holds for all i < n, meaning
that if [M : L] < n then [L
: M
] is nite and [L
: M
] [M :
L]. Given u M L, if [M : L] < then M is an algebraic
extension of L; in particular u is algebraic over L. Let
f(x) = a
k
x
k
+a
k1
x
k1
+ + a
1
x +a
0
L[x]
(with a
k
,= 0, k 2) be us minimal polynomial over L. By
V.1.6, [L(u) : L] = deg f and by V.1.1, [M : L(u)] =
n
degf
.
There are two cases:
deg f < n Here, we have 1 <
n
deg f
< n. So by the
induction hypothesis [L(u) : L] = n and
[L
: L(u)
] = n = [M : L]
which completes the proof.
deg f = n Here, [M : L(u)] = 1 implies M = L(u). To
reach a conclusion in this case we must dene an injection
fromS = L
_
M
(left cosets) to T = distinct roots of f in F.

It will then follow that #S #T. Finally since #T n
III.6.7 and #S = [L
: M
] (by denition) we will have

[L
: M
] #T n = [M : L]
and in particular
#Aut
K
F = [Aut
K
F : 1] = [K
: F
] [F : K]
as desired.
Select M
_
M
. If M
, then since M
= Aut
M
F
for any u M we know (u) = (u). This tells us that
every element of M
has the same eect on u: mapping

u (u). Because L
= Aut
L
F and L
, since u is a
root of f in L it follows from V.2.2 that (u) is another
root of f in L. Accordingly, the map
S M
(u) T
ASSIGNMENT 3 207
is well-dened.
To verify that this map is injective suppose ,
sat-
isfy (u) =
(u). Then, since
is invertible, (
)
1
(u) =
u. Thus (
)
1
xes u, and since 1, u, . . . , u
deg f1
is
a basis for L(u) over L it follows that (
)
1
xes L(u)
element-wise. From this it follows that
M = M, which
means that our map is injective.
Show that for every eld extension F K nite subgroups H of
G = Aut
K
F are closed.
Lemma (Hungerford V 2.8). Let K L M F. If [M : L] <
, then [L
: M
] [M : L].
Lemma (Hungerford V 2.9). Let G > J > H. If [J : H] < , then
[H
: J
] [J : H].
Lemma. If A < B < G with A closed and [B : A] < , then B is
closed and
[A
: B
] = [B : A].
Proof. Recall B < B
. Since A < B, it follows that [B : A]

[B
: A]. By hypothesis A = A
, so [B
: A] = [B
: A
]. Now, by
Lemmas V 2.8 and V 2.9
[B
: A
] [A
: B
] [B : A],
and so [B : A] = [A
: B
] (equality actually holds throughout the

above estimation). In particular
[B
: A] = [B
: B][B : A]
implies [B
: B] = 1, i.e. B
= B. In conclusion, B is closed.
Now use this with A = e and B = H to conclude that every nite
subgroup of G is closed.
Proof of Additional Exercise 3. The setup here is G > H >
e. The statement that H is nite means that [H : e] < . By
lemma 2, H is closed and [F : H
] [H : e]. There I used the fact

that e
= F.
Let F E K be elds. If E is algebraic and Galois over K then
E is stable.
Proof. Given u E it must be shown that (u) E for any
Aut
K
F.
Let f K[x] denote the irreducible polynomial of u, and let u =
v
1
, . . . , v
n
denote the distinct roots of f in E. Note n deg f. Every
Aut
K
E sends roots of f to roots of f, so it follows that permutes
v
i
: i = 1, . . . , n. The coecients of
g(x) =
i
(x v
i
)
are given by
g(x) = x
n
+
n
k=1
(1)
k
_

1j
1
<<j
k
n
v
j
1
v
j
2
. . . v
j
k
_
x
nk
.
Thus all coecients of g are xed by any , so g (Aut
K
E)
[x]. Be-
cause E K is Galois, (Aut
K
E)
= K. Hence g K[x].
Since g(u) = 0, f[g; but n deg f implies g[f. As g is monic,
it follows that f = g and hence fs roots are distinct elements of E.
Any Aut
K
F sends roots to roots, in particular (u) = v
i
E. In
conclusion, E is stable.
Remark. Write down us irreducible polynomial f. Show that
roots of f E (see below). Use the fact that Aut
K
F sends
roots of f to roots of f.
For the second part, Make a polynomial g out of distinct roots of f in E.
Use (Aut
K
E)
= K to show g K[x]; deduce f = g.

Final Exam
201 C Final Exam Problem 1.
Show that the extension is innite-dimensional. Deduce that
is innite.
Proof. For each prime p the polynomial x
p
1 splits over . In
particular contains a primitive pth root of unity for all primes p.
Certainly distinct primes give rise to distinct primitive pth roots of
unity. If is a primitive pth root of unity then () .
By way of contradiction assume d = [ : ] is nite. In particular,
then, p 1 = [() : ] d. There exist arbitrarily large primes, so
choosing p such that p > d + 1 leads to a contradiction. Accordingly
the dimension d must not be nite.
FINAL EXAM 209
Since is perfect, is Galois. Now
[ : ] = [1
: Aut
] [Aut
: 1
] =
Aut
shows that the Galois group must be innite.

Determine the Galois group, three of its subgroups, and the corre-
sponding intermediate elds of the splitting eld (over ) of the poly-
nomial f(x) = (x
3
2)(x
2
3) [x].
Theorem. If : K L, S = f
i
K[x] (positive degrees),
S
= f
i
L[x] (positive degrees), and M, N are splitting elds of
S, S
over K, L respectively then extends to : M N.

Because x
3
2 has two complex roots its Galois group
H = Aut
(
3
2, ) S
3
or A
3
( is a primitive third root of unity) contains the -automorphism
induced by complex conjugation. This element corresponds to an odd
permutation, and so H S
3
.
A splitting eld of f over the rationals is
F = (
3
2, ,
3).
Any H extends to an element of G, the Galois group of f over the
rationals, in one of two ways (by the Theorem) according to whether
3 or
3. Labeling roots and technical mumbling

shows that there is a group homomorphism G S
3
Z
2
. But F is
Galois over since the latter is perfect, and so [G[ = [S
3
Z
2
[ which
means that G is isomorphic to the product S
3
Z
2
.
Three intermediate elds are given by
F
(
3)
9
t
t
t
t
t
t
t
t
t
t
(
3
2)
O
(
3
2, )
fL
L
L
L
L
L
L
L
L
L
L
dJ
J
J
J
J
J
J
J
J
J
O 9
r
r
r
r
r
r
r
r
r
r
r
The corresponding subgroup diagram is as follows
S
3
Z
2
S
3
9
s
s
s
s
s
s
s
s
s
s
s
(23) 0
O
Z
2
eK
K
K
K
K
K
K
K
K
K
K
((),
0)
eK
K
K
K
K
K
K
K
K
K
K
O
9
s
s
s
s
s
s
s
s
s
s
s
Let K be a nite eld with [K[ = q (a prime power). Let K be
an algebraic closure of K. The Frobenius automorphism Fr is the K-
automorphism of K which is dened by the formula Fr(u) = u
q
. Let E
be an intermediate eld such that [E : K] < . Show that Fr generates
K
E.
Proof. I have gone through arguments like this one a thousand
times over.
Let d = [E : K], so that [E[ = q
d
. The extension E K is
Galois by an important theorem which implies that [Aut
K
E[ = d.
Note also that K and E are, by the theorem characterizing nite elds
are splitting elds of x
q
x and x
q
d
x over Z
p
(the prime subeld of
K) respectively. Here is what to do:
(1) Show that Fr Aut
K
E. Injective is almost too easy to men-
tion, surjecive comes from injective, and operation preserving
comes from the Freshmans dream.
(2) Show Fr
d
= 1
E
. This is true because E is a splitting eld
as mentioned above. Every element of E is a root of that
polynomial.
(3) Show Fr
i
,= 1
E
for any i < d. Use contradiction, it equality
holds then a polynomial of degree less than q
d
has q
d
roots,
no.

Show that every eld of characteristic zero is perfect.
Definition. A eld K is perfect when every irreducible f K[x]
is separable, i.e. has no multiple roots.
Proof. Suppose
f(x) =
n
i=0
a
i
x
i
K[x]
FINAL EXAM 211
is irreducible of degree n. If f has a multiple root u (say with multi-
plicity m), then there exists g K(u)[x] with g(u) ,= 0 such that
f(x) = (x u)
m
g(x) K(u)[x]
The derivative of f is then
f
(x) = (x u)[m(x u)
m2
g(x) + (x u)
m1
g
(x)] K(u)[x]
which shows that u is a root of f
as well. Since f is irreducible and

deg f
< f, either f
= 0 or (f, f
) = 1 in K[x]. It will be shown that

neither alternative is possible, whence f has no multiple root.
If f and f
are coprime in K[x] then there exist h, k K[x] such

that fh +f
k = 1. But then
0 = f(u)h(u) + f
(u)k(u) = 1
which is false. Consequently f and f
are not coprime. If f
= 0 then
examining the formula
0 = f
(x) =
n
i=1
ia
i
x
i1
shows that ia
i
= 0 for all i = 1, . . . , n. Since K has characteristic zero
it follows that a
i
= 0 for all i = 1, . . . , n. But then f(x) = a
0
which
implies that a
0
= f(u) = 0, i.e. f is the zero polynomial. This is
illogical because f is supposed to be irreducible.
201 C Final Exam Problem 5. Let K be a eld of positive
characteristic p. Let K
(p
n
)
be the set of all elements of the form u
p
n
where u K. Show that K
(p
n
)
is a subeld of K. Prove that K is a
purely inseparable extension of K
(p
n
)
.
Proof of Problem 5. It is assumed that n is xed.
To prove that K
(p
n
)
is a subeld of K it will be shown that K
(p
n
)
and K
(p
n
)
are subgroups of (K, +) and (K
, ) respectively.
Evidently 0 = 0
p
n
K
(p
n
)
. Suppose a, b K
(p
n
)
; then there exist
u, v K such that a = u
p
n
and b = v
p
n
. Since (K, +) is a group,
u v K by Theorem I.2.5. Thus
a b = u
p
n
v
p
n
= (u v)
p
n
K
(p
n
)
by the Freshmans Dream (III.1 Exercice 11). Another appeal to The-
orem I.2.5 lets us conclude that (K
(p
n
)
, +) is a subgroup of (K, +).
Certainly 1 = 1
p
n
K
(p
n
)
. Assume instead that a, b K
(p
n
)
; then
there exist u, v K
such that a = u
p
n
and b = v
p
n
. Since (K
, ) is a
group, uv
1
K
by Theorem I.2.5. Thus

ab
1
= u
p
n
(v
p
n
)
1
= u
p
n
(v
1
)
p
n
= (uv
1
)
p
n
K
(p
n
)
.
One last appeal to Theorem I.2.5 gives (K
(p
n
)
, ) is a subgroup of (K, ).
To see that K K
(p
n
)
is purely inseparable, rst observe that K
is algebraic over K
(p
n
)
. This is true because for all a K there exists
a polynomial, namely
f
a
(x) = x
p
n
a
p
n
K
(p
n
)
[x],
of which a is a root. By Theorem V.6.4, it suces to show that for all
u K there exists r 0 such that u
p
r
K
(p
n
)
. But this is obvious:
r = n. Hence the extension is purely inseparable.
201 C Final Exam Problem 6. Let K be an algebraic closure
of the eld K and Aut
K
K. Consider the eld
F = u K : (u) = u.
Prove that every nite dimensional extension E of F is cyclic.
Remark. The outline for this proof is as follows. The setup is this
K F E K
where the last containment comes from the fact that E is algebraic
over K and thus injects into K (identify E with its image).
(1) Pass to a normal closure N of E over F, which adds F E
N to the tower.
(a) Prove that N is Galois over F
(b) Prove that Aut
F
N is cyclic
(2) Observe that Aut
F
E is a quotient of Aut
F
N.
The Fundamental Theorem of Finite-Dimensional Galois Theory plays
a central role throughout. The normality of N over F is what makes
N Galois over F, and the second part follows in part from Lagranges
theorem.
We proved in class that if

K is a Galois radical extension of K
then Aut
K

K is solvable. Using this result, prove that if F is a radical
extension of K and E is an intermediate eld of the extension F K
then Aut
K
E is solvable.
Remark. Here is an outline. The tower is K E F.
(1) Move the ground eld K up to K
0
= Aut
K
E
.
(a) Show that E K
0
is Galois by denition and has the
same Galois group as the original.
(b) Show that F remains radical over K
0
.
So it can be assumed from the outset that E is Galois over K.
(2) Pass to a normal closure N of F over K, K E F N.
FINAL EXAM 213
(a) Note that E is a stable intermediate eld and obtain a
homomorphism Aut
K
N Aut
K
E (given by restriction).
(b) Use the characterization of normal extension as splitting
elds and the standard extension theorem to show that
the restriction homomorphism is surjective. Observe that
it suces to show that Aut
K
N is solvable.
(c) Show that N K is radical Galois by increasing the
ground eld K K
1
= Aut
K
N
, checking the Galois

group, and then showing that the extension stays radical
using a similar argument to the one for F K
0
.
(3) Use the result proven in class on N K, conclude that Aut
K
E
is solvable.
Proof. The setup is as follows:
K E F
where
(1) F = K(u
1
, . . . , u
n
),
(2) for each i there exists n
i
N such that u
n
i
i
K,
(3) for each i 2 there exists m
i
N such that u
m
i
i
K(u
1
, . . . , u
i1
).
Note that each extension is algebraic, since K F is algebraic by
denition.
Let K
0
= Aut
K
E
; by denition E K
0
is Galois, and Aut
K
E =
Aut
K
0
E. The local goal is to show that it is permissible to assume
that K E is Galois; but in order to do this it must be proven that
K
0
F is radical.
Three items must be veried
(1
) F = K
0
(v
1
, . . . , v
n
)
(2
) For each j = 1, . . . , n
there exists n
j
N such that v
n
j
j
K
0
.
(3
) For each j 2 there exists m
j
N such that v
m
j
j
K
0
(v
1
, . . . , v
j1
).
(1
) is immediate from (1): The tower is

K K
0
K(u
1
, . . . , u
n
).
Certainly there exists U u
i
such that K
0
= K(U). Dene v
j
as
U u
i
so that
F = K
0
(v
1
, . . . , v
n
) = K(U)(v
1
, . . . , v
n
).
Insist that the order is preserved, i.e. if v
j
1
= u
i
1
and v
j
2
= u
i
2
then
j
1
< j
2
u
1
< u
2
.
Now for each j there exists i such that v
j
= u
i
. Using (2), there
exists n
i
such that
v
n
i
j
= u
n
i
i
K K
0
.
Declaring n
j
= n
i
shows that (2
) holds. If j 2 then i 2 and,

using (3), there exists m
i
such that v
m
i
j
= n
m
i
i
K(u
1
, . . . , u
i1
). The
ordering dened above ensures that
n
1
, . . . , n
i1
= U

v
1
, . . . , v
j1
,
and since K
0
= K(U) it is clear that
K(u
1
, . . . , u
i1
) = K(U)(v
1
, . . . , v
j1
) = K
0
(v
1
, . . . , v
j1
).
So, dening m
j
= m
i
shows that (3
) holds. In conclusion F K
0
is a
radical extension.
For these reasons the extension E K may be regarded as Galois
from the outset: the ground eld can always be enlarged to make the
extension Galois in a way that does not aect the Galois group (the
object of concern in this argument). Furthermore, the new extension
remains radical. In what follows it is assumed that E K is algebraic
and Galois.
Choose a normal closure N of F over K, which gives
N F E K.
Since E K is algebraic and Galois, it is a stable intermediate eld
of the extension N E K. This means that given Aut
K
N, the
restriction [
E
Aut
K
E. As a consequence, there is a homomorphism
of groups
Aut
K
N
f
[
E
Aut
K
E.
The characterization theorem on normal extensions indicates that N is
a splitting eld of some set of polynomials f
i
: i I K[x]. Given
Aut
K
E, by denition : E E is an isomorphism. But N is also
a splitting eld of f
i
over K, and so extends to an isomorphism
: N N. Since xes K, it is certain that Aut
K
N. By
design [
E
= , which shows that f is an epimorphism. The First
Isomoprhism Theorem (for groups) indicates that
Aut
K
N
_
ker f Aut
K
E.
The quotient of a solvable group is solvable, so it suces to prove that
Aut
K
N is solvable.
Dene K
1
= Aut
K
N
; observe that N K
1
is Galois by denition,
and Aut
K
1
N = Aut
K
N. Similar considerations to the case showing
that F K
0
is radical can be applied here, yielding the conclusion
that N K
1
is radical and Galois. Hence it may be assumed without
loss of generality that N K is radical and Galois.
By the result proven in class, Aut
K
N is solvable and the proof is
complete.
FINAL EXAM 215
201 C Final Exam Problem 8. Let F = K(x
1
, . . . , x
n
) be the
eld of rational functions in n indeterminates, and E F be the
subeld of all symmetric rational functions in K(x
1
, . . . , x
n
).
(1) Show that x
1
, . . . , x
n
is a transcendence basis of F over K.
(2) Show that the elementary symmetric functions f
1
, . . . , f
n
form
a transcendence basis of E over K.
(3) Do the functions f
1
, . . . , f
n
also form a transcendence basis of
F over K?
Remark. A couple properties from the Symmetric Rational Func-
tions appendix are used in this problem.
(a) E = K(f
1
, . . . , f
n
)
(b) F is algebraic over E of dimension n!
Proof of (1). It is clear that if f F[y
1
, . . . , y
n
] satises f(x
1
, . . . , x
n
) =
0 then f = 0, so x
1
, . . . , x
n
is algebraically independent in F over K.
To see that x
1
, . . . , x
n
is -maximal with this property, suppose that
/ K(x
1
, . . . , x
n
) and consider the larger set x
1
, . . . , x
n
, / F.
The polynomial f F[y
1
, . . . , y
n
, y
n+1
] given by
f(y
1
, . . . , y
n
, y
n+1
) = y
n+1
(y
1
, . . . , y
n
) (y
1
, . . . , y
n
)
is such that
f(x
1
, . . . , x
n
, /) =
= 0.
So x
1
, . . . , x
n
is a -maximal algebraically independent set in F over
K.
Proof of (2). The tower K E F implies the equation
tr.d.
K
F = tr.d.
E
F + tr.d.
K
E.
But since F is algebraic over E (b) it follows that tr.d.
K
E = n. Now be-
cause E = K(f
1
, . . . , f
n
), as indicated in (a), it follows that f
1
, . . . , f
n
contains a transcendence basis of E over K. But f

1
, . . . , f
n
is an n-
ary set, so it must actually be a transcendence basis of E over K.
In particular f
1
, . . . , f
n
is algebraically independent in E over K.
Since algebraic independence depends only on the ground eld, it is
clear that f
1
, . . . , f
n
is algebraically independent in F over K. As
observed in the proof of (2), an algebraically independent subset of F
over K is a transcendence basis of F over K if and only if it is an n-ary
set. Consequently, the answer to (3) is yes.
Bibliography
[1] David S. Dummit and Richard M. Foote. Abstract Algebra. John Wiley & Sons,
3rd edition, 2004.
[2] T. W. Hungerford. Algebra. Graduate Texts in Mathematics. Springer, 1974.
Department of Mathematics, University of California, Riverside,
California 92521.
jmd@math.ucr.edu
http://math.ucr.edu/~jmd
217

Algebra Qual

Uploaded by

Copyright:

Available Formats

Algebra Qual

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Algebra Qual

Uploaded by

Copyright:

Available Formats

Algebra Qualifying Exam

belongs to the orbit of s or that the orbit sets of s and

Hs: there exists h H such that

= h.s. The denition of group-action gives

. Then the denition

2). Any element of the latter has

G is closed under the binary operation

. Prove the following statements.

X then there exists

Gx. In particular, by [Hungerford II-4.2]

k!. Since kernels are normal, G

, f maps the element

are bases for V, W respectively nd a natural basis for Hom

just observe that f(v) =

)) vs. ((a, g)(a

K with as in the previous Lemma.

of Sylow 5-sybgroups of G. This provides a group homomorphism

() = f. Its easy to verify that f

injective) Hypothesis: ker f

injective) Hypothesis: im f = W. Assume

(). For all w W there exists (by

()) (v) = f(v) = (w),

K: The algebraic closure

= F 0, a nite abelian group under multiplication.

, which is to say that each u F

3). I can do this with a, b 1

b). Thus the reverse inclusion holds.

F is completely determined by its action

2, meaning b = 1. But then a

in r of degree at most n. Denote f =

. Recall that the invariant

. Since K has no zero-divisors,

(x) = 1 is relatively prime to

and it follows that (h) = R[x]. Hence (x) is maximal in

(). Note that G is also the Galois

B, and r R. The quotient group F/K is

B and all r R the identities

10], N(uv) = N(u)N(v) and u Z[

10 is a unit with inverse a b

10]. Note that R is Noetherian, so each

10 are irreducible. If 2 = uv then

but not conversely.

but not conversely.

. But then (v) = v by the denition of H

which shows that H

st. Then tzt = tyt

and hence t(yt

For the proof that [G[ = p

is a conjugate of S contained in a subgroup

is a Sylow subgroup of K as well. In particular, we are

2003 B6. Let K be a eld and let M and N be nitely generated

over L, then extends to an isomorphism M L.

= the irreducible elements of K

< deg f these two polynomials must

k = 1 for some h, k K[x]. But then 0 = 1,

is the set of u F which are the

C(G) such that gc = a

. A similar statement applies to h.

C(G) such that