You are on page 1of 17

ALGEBRA QUALIFYING EXAM PROBLEMS

SOLUTIONS BY DAVE GAEBLER


Contents
Fall 1998 1
Groups 1
Rings 2
Spring 1999 4
Groups 4
Rings 5
Fall 1999 6
Groups 6
Rings 7
Spring 2000 8
Groups 8
Rings 9
Fall 2000 9
Groups 9
Spring 2001 10
Groups 10
Fall 2001 12
Groups 12
Winter 2005 12
Groups 12
Fall 2006 13
Groups 13
Rings 15
Spring 2007 16
Rings 16
Fall 1998
Groups.
G1: Let Q be the additive group of rational numbers and let H be a subgroup
of Q such that H +Z = Q. Show that H = Q.
Solution thanks to Tim Eller. Clearly H is nontrivial since Z ,= Q. Let
x =
a
b
H with a ,= 0. By repeated addition, a H. Now
1
a
= x + n for
some x h and n Z; then 1 = ax + an H since x H and a H.
Since 1 H, Z H so Q = Z +H = H.
1
2 ALGEBRA QUALS
G2: Let G be a nite group, H a normal subgroup of G, and P a Sylow
p-subgroup of G. Let N be the normalizer of P in G. Show that if H P,
then G = NH.
Solution from D/F 195. Note that P is a Sylow p-subgroup of H. Let
g G. By the normality of H, gPg
1
is also a Sylow p-subgroup of H. By
the Second Sylow Theorem, P and gPg
1
are conjugate in H, i.e. h H
with gPg
1
= hPh
1
. Thus (h
1
g)P(h
1
g)
1
= P, so h
1
g N. Thus
g Nh NH. This is true for all g G, so G = NH.
G3: Let G be a group and X a nontrivial normal abelian subgroup of G. Set
X
G
= x X [ gx = xg g G.
(a) Show that if G has order p
n
then X
G
,= e.
(b) Prduce a counterexample to show that this conclusion need not hold
if G is not a p-group.
Solution.
(a) (From D/F 191) Note that X
G
is the same as X Z(G). Partition
X into its conjugacy classes in G; since the size of a conjugacy class
must divide the order of a group, these classes all have sizes which are
powers of p. Since they add up to X which is also a power of p, the
number of classes of size 1 (i.e. the number of elements in X Z(G))
must be a multiple of p; in particular, it is not 1.
(b) S
3
has the nontrivial normal abelian subgroup 1, (123), (132), but
Z(S
3
) is trivial.

Rings.
R1: Let A be a nite commutative ring with identity.
(a) Show that A contains a maximal ideal.
(b) Show that there exists a surjective ring homomorphism : A k
where k is a eld.
Solution.
(a) Because all proper ideals of A are nite, we can simply take one of
largest size. (It is possible that 0 is the only proper ideal, in which
case its maximal.)
(b) Let M be a maximal ideal of A, which exists by part (a). Then the
natural homomorphism : A A/M is surjective, and A/M is a eld
because M is maximal.

R2: Let R be a Noetherian ring, I R an ideal, and K the radical of I. That


is, K = x R [ n such that x
n
I. [Editorial note: Commutativity is
probably meant to be assumed as well.]
(a) Show that there exists an integer N 1 such that x
N
I for all
x K.
(b) Show by counterexample that the conclusion of (a) is false if R is not
assumed to be Noetherian.
Solution of (a) thanks to Tim Eller.
ALGEBRA QUALS 3
(a) Suppose this is not the case. Let x
1
K and choose n
1
such that
x
n1
1
I. Let K
1
= x
1
). If K
1
= K then y
n1
I for all y K,
contrary to hypothesis. So x
2
inKK
1
. Choose n
2
such that x
n2
2
I.
Let K
2
= x
1
, x
2
). Note that for y K
2
, we can write y = a
1
x
1
+a
2
x
2
,
so that y
n1+n2
I. (Binomial expand and note that each term has
either x
n1
1
or x
n2
2
.) This implies that K
2
,= K, or we could take
N = n
1
+ n
2
. So x
3
K K
2
. Continuing, we obtain an innite
properly ascending chain K
1
K
2
. . . . But this is impossible by
the Noetherian property.
(b) Let R = Z[
k

n : n, k = 2, 3, . . . ] be the ring generated over Z by


all roots of positive integers. Note that this is not Noetherian be-
cause

2)

2,

3)

2,

3,

5) . . . properly. (This follows


from the linear independence over Q of square roots of primes; the
straightforward proof of this is to just induct on the number of primes.
See http://mathforum.org/library/drmath/view/51638.html). (Note
that, strictly speaking, part (b) of this problem does not require us to
show that R is not Noetherian; by part (a) this will follow from the
fact that it is a counterexample.) Consider the ideal I =

2R. The
elements
2n

2 are in rad(I), but must be raised to the nth power in


order to be in I.

R3: Let A be a commutative ring and M an A-module with nitely many


elements. For A, set
M[] = m M [ m = 0.
(a) Show that [M/M[ = [M[][.
(b) Show that if A is a PID, then M/M and M[] are isomorphic as
A-modules.
(c) Find an example in which M/M and M[] are not isomorphic as
A-modules.
Solution.
(a) The map x x is a homomorphism from M to M with kernel M[]
and image M; by FIT, M/M[] M. Since all orders involved are
nite,
[M[/[M[][ = [M[ [M[/[M[ = [M[][ [M/M][][ = [M[.
(b) Basic factorizational ugliness: Write M in elementary divisor form
as R/p
11
1
R R/p

km
k
R, where the p
i
be distinct. Write =
p
1
1
. . . p

k
k
r for some r not divisible by any p
i
. Note that R/p
s
k

R/p
s
k
k
so that M/M

i,j
R/p
iji
i
. It is also easy to show
directly from the factorizations that the latter product is isomorphic
to M[].
(c) Dont know; seems that you need to nd a nite non-PID since all
module actions factor through End(M), but I dont know any nite
non-PIDs.

4 ALGEBRA QUALS
Spring 1999
Groups.
G1: List up to isomorphism all nite abelian groups having at most four
elements of any given order.
Solution. By FTFGAG, any nite abelian group is a direct product of cyclic
groups of prime power order. No cyclic group of order p
n
can be a factor
if p > 5, if n > 2, or if n = 2 and p > 2, because any of those possibilities
will cause (p
n
) > 4 generators. From this it follows that the only possible
factors of prime power order are Z
2
, Z
2
Z
2
, Z
4
, Z
2
Z
4
, Z
8
, Z
3
, and
Z
5
. These can be combined in 6 2 2 = 20 ways. Of these, any of the 3
possibilities with Z
8
times a nontrivial subgroup of Z
15
can be ruled out
because Z
8
has 4 elements of the same order and any subgroup of Z
15
has
multiple elements with the same order. Next, the 5 possibilities with Z
3
Z
5
as a subgroup is ruled out because Z
3
Z
5
has 8 elements of order 15. Of
the others, we can rule out Z
2
Z
2
Z
3
because it has 6 elements of order
6, Z
2
Z
4
Z
3
because it contains a subgroup isomorphic to Z
2
Z
2
Z
3
,
Z
2
Z
2
Z
5
because it has 12 elements of order 10, Z
2
Z
4
Z
5
because
it contains a subgroup isomorphic to Z
2
Z
2
Z
5
, and Z
4
Z
5
because it
has 8 elements of order 10. It is tedious but straightforward to verify that
the remaining 11 possibilities all work, namely:
0
Z
2
Z
2
Z
2
Z
4
Z
2
Z
4
Z
8
Z
3
Z
2
Z
3
Z
4
Z
3
Z
5
Z
2
Z
5

G2: Determine all groups of order 1445 up to isomorphism.


Solution. Since 1445 = 5 17
2
, the third Sylow Theorem implies that n
5
=
n
17
= 1. Since all the Sylow subgroups are unique and therefore normal by
Sylow 2, G must be the product of its Sylow subgroups (see D/F 193). But
these are both abelian because groups of order p
2
are always abelian (this
is a well known problem; Z(G) is nontrivial by D/F 190, and if its order
is p then G/Z(G) is cyclic which implies G is abelian). By FTFGAG, the
possibilities are Z
5
Z
2
89 and Z
5
Z
1
7 Z
1
7.
Note: The hint to this problem says to compute the order of GL
2
(F
17
).
WTF??
G3: Let G be a p-group and suppose that G acts linearly on a vector space V
over the nite eld Z/pZ. Prove that there exists a nonzero vector v V
such that gv = v for all G.
ALGEBRA QUALS 5
Solution. If V is nite-dimensional, then it has a nite number of elements
p
m
. The orbits of V under G all have orders that are powers of p, and since
they add up to p
m
, the number of orbits of size 1 must be a multiple of p;
in particular, it is greater than 1. If V is innite-dimensional I dont know
how to solve the problem, or even whether the given statement is true.
Rings.
R1: List up to isomorphism all rings with identity with pqr elements where
p, q, r are distinct primes. Prove that your list is complete.
Solution. In fact there is only one such ring. As an abelian group, any
such ring R must be cyclic, by FTFGAG. Now if 1
R
has order k, then
kx = (1 + + 1)x = 0x = 0 for all x R, so 1
R
has order at least
as great as any other element; hence it must be a generator of R as an
additive group. But then the multiplicative structure is determined by the
distributive law, since everything is a repeated sum of 1s. So Z Z/pqrZ
is the only possibility.
R2: Let I be the ideal in Z[X] consisting of all polynomials f(X) such that
f(3) is divisible by 2. Show that I is generated by two elements but that I
is not a principal ideal.
Solution. Every polynomial f(X) can be rewritten as a polynomial

f(X
1); we see that f(3) is divisible by 2 i the constant term of

f is. Now the
non-constant part of

f(X 1) is divisible by X 1, hence is in X 1,
and an even constant term is in 2). So all polynomials with 2 [ f(3) are
in X 1, 2). However, this ideal is not principal; it contains constants,
so any generator would have to be an even constant, but then it could not
generate X 1.
R3: A commutative ring R is called Artinian if every nonempty family of
ideals has a minimal element. Let R be an Artinian ring.
(a) Show that R contains at most nitely many maximal ideals.
(b) Show that if R is an Artinian integral domain, then R is a eld, and
conclude that in an Artinian ring, every prime ideal is maximal.
Solution thanks to Tim Eller.
(a) Let F be the family of all intersections of nitely many maximal ideals
of R. Let M
1
M
n
be a minimal element of F, If R has in-
initely many maximal ideals, there exists M
n+1
not equal to any of
M
1
, . . . , M
n
. Since no M
i
is contained in another, x
j
M
j
M
n+1
for
j = 1, . . . , n. Then x
1
. . . x
n
M
1
M
n
, but if it were in M
n+1
,
which is prime because its maximal, we would have some x
i
M
n+1
,
a contradiction. This proves that M
1
M
n+1
is properly contained
in M
1
M
n
, contradicting minimality.
(b) It is easy to show the given minimality denition of Artinian is equiva-
lent to the usual DCC; given this, for any x R, R xR x
2
R . . .
stabilizes, so mx
n+1
= x
n
for some n Z and m R, implying
x
n
(mx 1) = 0 mx = 1 so that x is invertible. Hence R is a eld.
If we also use the fact that quotients of Artinian rings are Artinian
6 ALGEBRA QUALS
(can just check from the denitions), then for any prime ideal P, R/P
is an Artinian integral domain, hence a eld, hence P is maximal.

Fall 1999
Groups.
G1: Let G be an additive group. We call G divisible if for each nonzero
integer n and each g G there exists h G such that nh = g
(a) A direct sum of divisible groups is divisible.
(b) Every additive group can be homomorphically embedded into a divis-
ible group.
Solution.
(a) For the heck of it, lets consider the general case where the direct sum
may be over innitely many groups. We thus write

x
i
for an element
in

G
i
where all but nitely many x
i
are zero. Suppose all the G
i
are
divisible, and consider an element x =

x
i


G
i
. Let x
i1
, . . . , x
ij
be the nonzero components, if there are any. For any nonzero integer
n, choose y
i
k
G
i
k
with ny
i
k
= x
i
k
. Let z

G
i
be dened by
z
i
k
= y
i
k
and z
i
= 0 for i ,= i
1
, . . . , i
j
. Then nz = x. Thus

G
i
is
divisible.
(b) (Thanks to Tim Eller) The Z-module G = ZG is embedded in the
module QG, which is a divisible group.

G2: Classify all groups of order 21 up to isomorphism.


Solution from D/F 183-4. Let G be a group of order 21, and let x, y G
with [x[ = 3 and [y[ = 7; such x and y exist by Cauchys theorem, or Sylow
1 if you prefer. Let H = y) and K = x). By Sylow 3, n
7
= 1 so that
H G. Now K acts on H by automorphisms, viz. h khk
1
for any xed
h; this action corresponds to a homomorphism f : K Aut(H). Since
Aut(H) Z
6
and [K[ = 3, the action is either trivial (in which case x and
y commute, so that xy has order 21 and G is a cyclic group isomorphic to
Z
21
), or maps K onto the unique subgroup 1,
2
,
4
Aut(H) of order 3,
where : y y
2
is a generator of Aut(H). In the latter case we must have
either x
2
or x
1

2
. Since x and x
1
can be arbitrarily relabeled,
these result in isomorphic groups; WLOG x
2
. This yields the group
x, y : x
3
= y
7
= 1, xyx
1
= y
4
), which can also be written as Z
7
Z
3
where
the semidirect product is given by the homomorphism x
2
indicated
above.
Second solution thanks to Tim Eller. As above, choose [x[ = 3 and [y[ = 7.
By normality of y), xyx
1
= y
i
. Raising both sides to the ith power
yields y
i
2
= xy
i
x
1
= x(xyx
1
)x
1
= x
2
yx
2
; doing it again yields y
i
3
=
x
3
yx
3
= y. So i
3
1 mod 7; we can take i = 1, 2, or 4. In the case i = 1
we get that G is commutative and therefore equals Z
21
. The other two
cases are isomorphic because we can switch x with x
1
(or y with y
2
since
both generate y)); they yield the group mentioned in the rst solution.
ALGEBRA QUALS 7
Rings.
R1: Let R be a domain. Show that R is a UFD if and only if any nonzero
non-unit in R can be written as a nite product of prime elements.
Solution. Suppose R is a UFD. We need only show that irreducibles are
prime. Let r be an irreducible, and suppose r divides ab for some nonzero
non-units a and b. The unique factorization of ab into irreducibles is the
product of the factorization of a and the factorization of b, so one of these
must contain a factor of r. Now suppose R is a domain where every element
has a prime factorization. Since primes are irreducible, we need to show
there is no other factorization into irreducibles. Let p
1
. . . p
k
be a nonzero
non-unit in R. (Here the p
i
are not necessarily distinct.) Suppose there
is another factorization r
1
. . . r
s
into irreducibles. Since p
1
is prime, one of
the r
i
is divisible by p
1
; but since this r
i
is irreducible, it must in fact be p
1
times a unit c
1
. Since R is a domain, the cancellation property allows us to
divide both sides by p
1
. Repeating for all the p
1
, we get 1 on the left and
a product of r
i
s times units on the right. This implies that there are no r
i
left, so the r
i
were just the p
i
rearranged. (If there were more p
i
we would
get a product of primes equaling a unit, which is also impossible.)
R2: Let R be a domain and K its quotient eld. If M R is a maximal
ideal, let
R
M
= a/b [ a, b R and b / M
be the localization of R at M. Prove that
R =

M max ideal
R
M
.
Solution from D/F 687. Clearly R R
M
for any M, so one inclusion is
obvious. For the other direction, suppose x R
M
. We will show x R.
Let I
x
= d R : dx R. This is basically the set of all possible
denominators in the dierent equivalent ways of writing x. It is easy to see
that I
x
is an ideal of R. If x / R, then 1 / I
x
, so I
x
is a proper ideal and
is therefore contained in some maximal ideal M. Now since x R
M
by
assumption, we can write x = a/b for some a, R with b / R
M
. But then
b I
x
R
M
, a contradiction. Hence x must be in R.
R3: Let K be a eld.
(a) Prove that the following are equivalent:
(i) Every maximal ideal in the polynomial ring K[X
1
, . . . , X
n
] has
the form (X
1
a
1
, . . . , X
n
a
n
) for some a
i
K.
(ii) If I is a proper ideal of K[X
1
, . . . , X
n
], then there is a point
a = (a
1
, . . . , a
n
) K
n
such that f(a) = 0 for all f I.
(iii) If A is the homomorphic image of K[X
1
, . . . , X
n
] and also a eld
then the natural map K A is an isomorphism.
(b) Prove that if K satises (i), (ii), and (iii) for some n 1, then K is
algebraically closed.
8 ALGEBRA QUALS
Spring 2000
Groups.
Problem A1: (10 points) State a theorem which classies (i.e. lists) all nite
abelian groups up to isomorphism. This means that each nite abelian
group should be isomorphic to exactly one group of your list. Use your
classication to list abelian groups of order 24.
Solution. The Fundamental Theorem of Finitely Generated Abelian Groups
states that every abelian group is a direct product of cyclic groups. This
representation can be made unique by requiring the cyclic groups to have
prime power order. Using this, the abelian groups of order 24 are Z
3
Z
8
,
Z
3
Z
4
Z
2
, and Z
3
Z
2
Z
2
Z
2
.
Problem A2: (15 points) Let S
5
be the symmetric group on 5 letters. For
each positive integer n, list the number of elements of S
5
of order n. Justify
your answer.
Solution. Consider the cycle representation of an element S
5
. The
order of is simply the lcm of the lengths of these cycles. We use the
notation [i
1
. . . i

] to denote a cycle of type i


1
. . . i

, i.e. with cycles of


length i
1
, . . . , i

. By basic combinatorics, S
5
has 1 element of type [11111],
_
5
2
_
= 10 elements of type [1112], 53 = 15 elements of type [122],
_
5
3
_
2! = 20
elements of type [113],
_
5
2
_
2! = 20 elements of type [23], 5 3! = 30 elements
of type [14], and 4! = 24 elements of type [5]. Since such elements have order
1, 2, 2, 3, 6, 4, and 5 respectively, S
5
has 1 element of order 1, 25 elements of
order 2, 20 of order 3, 30 of order 4, 24 of order 5, and 20 of order 6. (Note:
It is worth checking that 1+25+20+30+24+20=120. It is also worth noting
that the number of elements of each type divides 120 in accordance with
the class equation.)
Problem A3: (20 points) Let F
4
be the eld with 4 elements. Let G =
SL(2, F
4
) be the group of 2 by 2 invertible matrices with entries in F
4
.
What is the order of G? Show, by analysing the action of G on the lines
containing the origin in (F
4
)
2
, that G is a simple group.
Solution. First, note the typo: SL is not the group of invertible matrices.
Thats GL. SL is the matrices of determinant 1. Now, the number of
pairs (a, b) F
2
4
with ab = 0, 1, i, and 1 + i are 7, 3, 3, 3 respectively. (One
way to see this is that for x ,= 0, there are three such pairs because any
a ,= 0 corresponds to a unique b = x/a.) Therefore, the number of pairs
of pairs (a, b), (c, d) with ab = cd + 1 is 7 3 + 3 7 + 3 3 + 3 3 =
60. Since a matrix
_
a b
c d
_
has determinant 1 i ab = cd + 1, [G[ = 60.
(Alternatively, [SL(2, F
4
)[ = (4
2
1)(4
2
4) because this is the number of
ways to choose a basis to which to send the standard basis; one can then
show that GL/SL D where D is the scalar matrices, a group isomorphic
to F

4
.) Now consider the action of G on the set S of lines through the
origin in F
2
4
. (This action can be performed by taking a representative
vector from the line, performing a matrix multiplication, and taking the line
corresponding to the result.) This action is faithful because the elements of
ALGEBRA QUALS 9
a matrix are uniquely determined by its action on the unit vectors. Hence
it corresponds to an injection G S
5
. Thus, G is isomorphic to a subgroup
of S
5
of order 60. But the only such subgroup is A
5
, which is simple. So G
is simple.
Rings.
R1: List, up to isomorphism, all commutative rings with four elements. Prove
your answer.
R2: Let p be a prime number. Show that a free Z-module of rank 2 has p +1
submodules of index p.
R3: Let R be a commutative noetherian ring in which each ideal I is principal
and satises I
2
= I. Show that R is isomorphic to a nite product of elds.
Fall 2000
Groups.
A1: (10 points) Let D
2n
be the dihedral group of order 2n with n > 1.
Determine the number of subgroups of D
2n
of index 2, and justify your
answer.
Solution. If n is odd then there is only one such subgroup, whereas if n is
even then there are three. Consider the usual presentation D
2n
= , [

n
=
2
= 1, =
1
). Let R = ) D
2n
. Clearly R has index 2, so it
is one possibility. Now suppose H is any other subgroup of index 2 in D
2n
.
Then HR = G, so [R H[ = [R[[H[/[HR[ = [R[/2 = n/2. If n is odd this
impossible, so there are no other subgroups of index 2. If n is even then
R H is the unique subgroup R

of index 2 in R, namely
2
). Now since
[H[ > [H R[, H must contain some element not in R

; this element can


be expressed in the form
i
. If i is even then H contains all elements
i
with i even, as well as R

; in other words, H = R

. If i is odd then H
must similarly contain all
i
with i odd, so H = R

. This covers
all the possibilities.
A2: (15 points) A group of order a power of a prime p is called a p-group.
Let G be a nite group. Prove that for any given prime p, there exists a
unique normal subgroup N of G such that (i) G/N is a p-group, and (ii) any
homomorphism of G into a p-group is trivial on N (that is, (N) = 1).
Solution thanks to Tim Eller. Let [G[ = p
n
m where p m. Let
N =
_
N

G
m||N

|
N

.
This intersection is nonempty because it contains the term N

= G. Obvi-
ously N is normal because it is an intersection of normal subgroups. Let
f : G P be any homomorphism to a p-group. Then [imf[ = p
k
for some
k, so [ ker f[ = [imf[/[G[ = mp
s
for some s. Since m [ [ ker f[, we have
N ker f by construction, so f(N) = 1. Finally, to prove uniqueness,
suppose H is any other group with the desired properties. If H ,= N then
there is a subgroup N

G with m [ [N

[ and H , N

. Let be the projec-


tion G G/N

. Now G/N

is a p-group because m [ [N

[, but does not


10 ALGEBRA QUALS
vanish on H because H , N

. Hence N is the only group with the desired


properties.
A3: (20 points) Let G be a nite group of order n. Suppose that G has a
unique subgroup of order d for each positive divisor d of n. Prove that G
is a cyclic group.
Solution. We rst prove that for each d[n there are at most (d) elements of
G having order d. Suppose there is at least one element of order d, and call
it x. Then x) is the unique subgroup of G of order d, so all the elements
of order d must be contained in it; but its cyclic, so it contains exactly
(d) generators. Thus, the number of elements of G of order d is either 0
or (d). This implies that the number of elements of order less than n is
at most

d|n
d<n
(d) = n (n).
Thus, there are at least (n) elements of G of order n; in particular, there
is at least 1, so G is cyclic.
Second Solution thanks to Tim Eller. Since the Sylow subgroups of G are
unique, G is the product of its Sylow subgroups (D/F 193). Thus, we need
only show that each of these is abelian. Now a group of order p
m
with at
most one subgroup of each order has at most 1+p+ +p
m1
=
p
m
1
p1
< p
m
elements of order less than p
m
. Hence there are elements of order p
m
, so
the Sylow subgroups are all cyclic and hence abelian.
Spring 2001
Groups.
A1: (10 points) Determine a complete set of groups of order eight up to
isomorphism and show that every group of order eight is isomorphic to one
of these.
Solution. By the FTFGAG, the abelian groups of order 8 are Z
2
Z
2
Z
2
,
Z
2
Z
4
, and Z
8
. We will show that the non-abelian groups are D
4
and the
quaternion group Q
8
.
If [G[ = 8 and G is not abelian, then every non-identity element has
order 2 or 4. Suppose all elements of G have order 2. In this case ab =
bbab = baabab = ba(ab)
2
= ba for all a, b G, so G is abelian after all.
So in the non-abelian case, there must be an element x of order 4. Let
H = x). Suppose G H has an element y G H of order 2, and let
K = 1, y. Then H is normal (because it has index 2), H K = 1, and
[H[[K[ = [G[, so G = H K. Since [K[ = 2 there are only two choices for
the automorphism in this semidirect product, the nontrivial one of which
produces D
4
.
The remaining case is where G H consists only of elements of or-
der 4. Let y be one of these elements, and K = y). Now [H K[ =
[H[[K[/[HK[ = 2, so x
2
= y
2
since these are the only elements of order 2
in H and K respectively. Now by the normality of H, yxy
1
= x
i
for some
i. We can rule out i = 0 because x ,= 1, i = 1 because if x and y commuted
ALGEBRA QUALS 11
G would be abelian, and i = 2 because then we could square both sides
and get 1 = x
4
= yx
2
y
1
= y
3
y
1
= y
2
. So yxy
1
= x
3
. We thus have
the presentation G = x, y [ x
4
= y
4
= 1, yxy
1
= x
3
, x
2
= y
2
) which we
recognize as the quaternion group Q
8
.
A2: (15 points) A nite group G acts on itself by conjugation. Determine all
possible G if this action yields precisely three orbits. Prove your result.
Solution. If all three of the orbits have size one, the group is Z
3
. Suppose
exactly two of the orbits have size one, and the third has size m. By the
class equation, m must divide m+2, so m = 2 and [G[ = 4. But all groups
of order 4 are abelian, so this case is impossible. Finally, suppose the center
of G is trivial. Then 1 + d
1
+ d
2
= n where n = [G[ and d
1
and d
2
are
divisors of n. These divisors are each at most n/2, so the only way the sum
can be attained is if one of them is
n
2
and the other
n
2
1. But
n
2
1 must
also be a divisor of n; it is easy to check that for n > 6 we have
n
3
<
n
2
1,
so
n
2
1 cannot be a divisor. Hence the only remaining possibility is n = 6,
leaving us with the only non-abelian group of order 6, namely S
3
.
A3: (20 points) Let G be a nitely generated group.
(a) Show for each integer n there exist nitely many subgroups of index
n.
(b) Suppose that there exists a subgroup of nite index in G. Prove that
there exists a characteristic subgroup of nite index.
Solution.
(a) Let H G with [G : H] = n. Let G act on the cosets of H by left
translation. This action corresponds to a homomorphism G S
n
with H = stab
G
(1) (i.e. H consists of precisely those elements x G
with xH = H). There are nitely many homomorphisms G S
n
(since each is determined by where it sends the generators), so nitely
many possibilities for the stabilizer of 1.
(b) One assumes, of course, that the trivial case [G : G] = 1 is meant
to be excluded. Let H = H
1
be a subgroup of G of index n, and
let H
1
, . . . , H
k
be the nitely many subgroups of index n. Dene
H

= H
i
. To show this has nite index, we proceed inductively; let
f
i
: G S
n
be the homomorphisms corresponding to H
i
, as in part
(a). Note that H
1
H
2
consists of those elements of G whose images
under both f
1
and f
2
x 1; its index in H
1
is given by [H
1
: H
1
H
2
] =
[orb
f2,H1
(1)[ n, which is nite. Inductively, we have [G : H

] < .
To show H

is characteristic, we rst note that [G : (H)] = [G : H]


for any subgroup G and automorphism (checking this is elementary,
my dear Watson!) Thus,
(H

) =
_

H
i
_
=

H
i
H

where H
i
= (H
i
). On the other hand, replacing by
1
yields H

1
(H

) (H

) H

. Hence (H

) = H

, so H

is characteristic.

12 ALGEBRA QUALS
Fall 2001
Groups.
G1: (10 points) Let G be a nite group whose center has index n. Show that
every conjugacy class in G has at most n elements.
G2: (15 points) Let G be a subgroup of S
n
that acts transitively on the set
1, 2, . . . , n. Let H be the stabilizer in G of an element x 1, 2, . . . , n.
Prove that

gG
gHg
1
= e.
G3: (20 points) Let G be the group of matrices of the form
_
a b
0 1
_
where a (Z/pZ)

and b Z/pZ. Describe all normal subgroups of G.


Solution. Consider the subgroup S of matrices with a = 1. Clearly it has
order p; by Sylow 3 and 2, it is the unique, therefore normal, p-subgroup.
Now suppose H is any other normal subgroup of G...
Winter 2005
Groups.
A1: (a) If G is a simple group that has a subgroup of index n, prove that
the order of G is a factor of n!.
(b) Prove that there is no simple nonabelian group of order p
e
m with e > 0
for a prime p > m.
Solution.
(a) Let H be a subgroup of G with index n, and let G act by left translation
on the cosets of H. This action corresponds to a homomorphism f :
G S
n
. Because G is simple, ker f must be trivial, so f is injective.
Therefore [imf[ = [G[, and since imf is a subgroup of S
n
, its order
divides [S
n
[ = n!.
(b) If [G[ = p
e
m, the Sylow p-subgroups of G have index m. By part (a),
this implies that [G[ divides m!. But this is impossible because p > m
so p does not divide m!.

A2: An additive abelian group is called divisible if multiplication by n for


every positive integer n is a surjective endomorphism.
(a) Show that if G is divisible, G/H is divisible for any subgroup H of G.
(b) Give an example (with a proof) of a divisible group for which the
multiplication by n is not an automorphism for every positive integer
n.
(c) Prove or disprove that there is only one isomorphism class of nitely
generated divisible groups.
Solution.
ALGEBRA QUALS 13
(a) Let H G. For x + H G/H, dene n(x + H) = nx + H. This is
clearly an endomorphism from G/H to itself, and it is straightforward
to check its surjectivity: Let y +H G/H, and choose x with nx = y.
Then n(x +H) = y +H.
(b) The group Q/Z is divisible: The endomorphism x nx is surjective
because given
p
q
+ Z, and positive integer n, n(
p
nq
+ Z) =
p
q
+ Z.
However, multiplication by n is not injective and therefore is not an
automorphism; in particular, n
1
n
= 0.
(c) By the Fundamental Theorem of Finitely Generated Abelian Groups,
every nitely generated abelian group is a direct product of cyclic
groups. But (as can be very easily veried) a direct product of groups
is divisible i each of the groups is, and no cyclic groups are divisible;
Z is not divisible because 1 is not in the image of the map x 2x, and
Z/nZ is not divisible because 1 is not in the image of the map x
nx, which is in fact the zero map. Thus, the only nitely generated
divisible groups are the trivial groups, which are a single isomorphism
class.
Note to self: It might be desirable to prove this without appealing to
the FTFGAG. This always seems to lead into module theory territory.
For example, we might take a (minimally large) generating set. Divid-
ing one of its members by 2 and writing the result in the basis yields
a dependence relation over Z, but since Z is not a eld, this doesnt
mean that you can write one generator as a linear combination of the
others (i.e. that you can nd a smaller basis and hence a contradic-
tion). Stirs up dim memories of some basic problems in module theory,
doesnt it?

Fall 2006
Groups.
G1: List all nite groups G whose automorphism group has prime power
order. Justify your answer.
Solution. We shall use the general fact that
G/Z(G) Inn(G) Aut(G),
which comes from applying the First Isomorphism Theorem to the action
of G on itself by conjugation. If Aut(G) has prime power order, this implies
that [G/Z(G)[ is either 1 or prime. In the former case we have Z(G) = G so
that G is abelian. In the latter case we have G/Z(G) is cyclic, which implies
that G is abelian. So G is abelian. By the Fundamental Theorem of Finitely
Generated Abelian Groups, G is isomorphic to a direct product of cyclic
groups of prime power order. But in general Aut(GH) contains copies of
Aut(G) and Aut(H) as subgroups, so each factor in this direct product must
have trivial automorphism group except one, which has an automorphism
group of prime order. Now it is well known that [Aut(Z/nZ)[ = (n),
and (n) = 1 only for n = 2. So all but one factor of G must be Z/2Z;
moreover, there cannot be more than one such factor because Aut(Z
2
Z
2
)
has order 6 (choose two generators and map the two of them to any distinct
14 ALGEBRA QUALS
elements of order 2). Moreover, the only prime powers n such that (n)
is prime are n = 3 and 4. So G contains one factor Z
3
or Z
4
and at
most one factor Z
2
, i.e. G = Z
3
, Z
4
, Z
6
, or Z
4
Z
2
. One can easily check
that these all have automorphism groups of prime order, i.e. that there
arent any extra automorphisms introduced by the product; for Z
3
and
Z
3
Z
2
, order considerations dictate that Z
2
and Z
3
are both mapped into
themselves; for Z
4
and Z
2
this follows as well because the generator of Z
4
must be mapped to an element of order 4.
G2: Let G be a nite group and H be a non-normal subgorup of G of index
n > 1.
(a) Show that if [H[ is divisible by a prime p n, then H cannot be a
simple group.
(b) Show that there is no simple group of order 504 = 2
3
3
2
7.
Solution.
(a) Let H act by left translation on G/H. This provides a homomorphism
f : H S
n
. If H is simple, ker f is either H (which is impossible
because the action is nontrivial, in fact, transitive) or trivial. In the
latter case, f is an injection, so [H[ = [imf[ which is a subgroup of
S
n
. Hence [H[ divides [S
n
[ = n!. But this is impossible because p [ [H[
and p n!.
(b) If this problem seems hard, its because its false. The simple group
PSL(2, 8) has order 504. Heres what I think you were supposed to do:
By Sylows theorem, n
3
= 1, 4, 7, or 28. We can rule out 1, obviously,
and if its 4 we can map G into S
4
(action by conjugation on the Sylow
3-subgroups) which has nontrivial kernel. So it must be 7 or 28. If
we assume its 7, we have a map into S
7
. Now because G has no
subgroup of index 2, it must in fact be contained in A
7
(see D&F page
206 Prop 12), which has order 2520. Then G is a subgroup of A
7
of
index 5, but whose order is divisible by 7, so part (a) implies it cannot
be simple. So far so good. But theres no way to rule out 28, and
evidently PSL(2, 8) does in fact have 28 Sylow 3-subgroups. Oops.

G3: Let
SL
2
(Z/pZ) =
__
a b
c d
_
[ a, b, c, d Z/pZ, ad bc = 1
_
,
where p is an odd prime.
(a) Prove that any subgroup of a cyclic group is cyclic.
(b) Compute the order of G.
(c) Prove that for any odd prime , the Sylow -subgroup of G is cyclic.
Solution.
(a) Let H be any cyclic group with generator x, and let K H be a
subgroup. Because x generates H, every y H can be written in
the form x
k
for some integer k; this integer can be made unique by
requiring it to have minimal possible absolute value, and be positive if
a tiebreaker is needed. Using this scheme, let z = x
j
be an element of
K with the smallest possible exponent. If z does not generate K then
ALGEBRA QUALS 15
there exists w K where w = x
sj+r
where 0 < r < j, by the Euclidean
Algorithm. But then x
r
K, a contradiction. So z generates K.
(b) For x Z/pZ, the number of pairs (n
1
, n
2
) (Z/pZ)
2
with n
1
n
2
= x
is 2p 1 if x = 0, and p 1 otherwise. The number of quadruplets
(a, b, c, d) with ad bc = 1 can then be computed by breaking into p
cases depending on the value of bc; of the resulting p products, there
are 2 of the form (2p 1)(p 1) and (p 2) of the form (p 1)
2
, for
a total of 2(2p 1)(p 1) + (p 2)(p 1)
2
= p(p
2
1). Another way
to derive the same result is to note that if a ,= 0 (which can happen in
p 1 ways), then b and c can be chosen arbitrarily (in p
2
ways), and
d will be uniquely determined. On the other hand, if a = 0 then d is
arbitrary, b can be anything nonzero, and c will be determined. The
total number is thus p
2
(p 1) +p(p 1) = p(p
2
1).
(c) This one stumped me. Seems like the idea is to show that theres a
unique Sylow -subgroup (dont know how), then show that G contains
a copy of F

p
2
. For the latter case, we might proceed by analogy with
the correspondence between complex numbers and 2 2 matrices.

Rings.
R1: Determine all prime ideals in the polynomial ring Z[X]. Justify your
work.
R2: Let R be a noetherian domain. A nonzero element x in R is called a
prime element if (x) is a prime ideal. Prove all of the following:
(a) Every nonzero non-unit in R is a product of irreducible elements.
(b) Every nonzero ideal I ,= R in R contains a (nite) product of nonzero
prime ideals.
(c) If every nonzero prime ideal in R contains a prime element then every
irreducible element in R is a prime element.
Solution.
(a) We rst prove the following lemma:
Lemma 1. Let x R with x ,= 0 and x / R

. Then x is divisible by
some irreducible element r R.
Proof. If x is irreducible were done. If not, x = a
(1)
1
b where neither a
nor b is a unit. If a
(1)
1
is irreducible were done. If not, a
(1)
1
= a
(2)
1
a
(2)
2
where neither factor is a unit. Continuing, we have ...[a
(3)
1
[a
(2)
1
[a
(1)
1
[x
where a
(i)
1
is not a unit. But then xR a
(1)
1
R a
(2)
1
R . . . and by
the Noetherian property, this chain stabilizes, so a
(i)
1
R = a
(i+1)
1
R for
some i. But this implies that a
(i+1)
1
is a unit, a contradiction. So one
of the a
(i)
1
must be irreducible, and they all divide x, so were done.
Now let x R be any nonzero non-unit. By the lemma, x is divisible
by some irreducible r
1
, say x = r
1
m
1
. If m
1
is irreducible were done; if
not, m
1
= r
2
m
2
with r
2
irreducible, by the lemma again. Continuing,
we have x = r
1
. . . r
i
m
i
for i = 1, 2, . . . . If this process never terminates
we have xR m
1
R m
2
R . . . where the containments are proper
because m
i
is not a unit; but this contradicts the Noetherian property,
16 ALGEBRA QUALS
so the process terminates and we have x as a product of nitely many
irreducibles.
(b) Let I ,= R be a nonzero ideal. Then I contains a nonzero element
a and therefore contains the ideal aR. Since I ,= R, a cannot be a
unit, so a = r
1
. . . r
n
for some irreducible r
1
, . . . , r
n
by part (a). Then
(r
1
R) . . . (r
n
R) = r
1
. . . r
n
R = aR I and since r
i
R is prime for
irreducible r
i
, we have a nite product of prime ideals inside I. HOW
NOW IRREDS PRIME? NOT ALWAYS!
(c) Let r be irreducible. Then rR is prime

R3: Let R be a commutative ring and M a nitely generated R-module.


Suppose there exists a positive integer n and a surjective R-module ho-
momorphism : M R
n
. Show that ker is also a nitely generated
R-module.
Spring 2007
Rings.
R1: Let D be a division ring (a ring with identity in which every nonzero
element is invertible). Let R = Mat
n
(D) be the ring of nn matrices with
entries from D. Prove that R has no two-sided ideals other than R itself
and 0.
Solution. It is easy to verify that E
ri
Ae
jr
has the element a
ij
in the r, r
position and 0s elsewhere, where E
st
is the matrix with a 1 in the s, t
position and 0s elsewhere. If I is a nonzero ideal in R, it contains a nonzero
matrix A with some nonzero entry a
ij
; then

n
r=1
E
ri
AE
jr
I because
it is an ideal, but as just noted this sum equals a
ij
times the identity.
Multiplying by a
1
ij
, we see that the identity matrix is in I, so I = R.
R2: Let R = End(V ) be the ring of all linear endomorphisms of an innite
dimensional complex vector space V with countable basis e
1
, . . . . Prove
that R and R R are isomorphic as left R-modules.
Solution. Dene a map : R R R as follows: Given (f, g) R R,
i.e. f, g : V V linear, let (f, g) = f g : V V be the linear map
dened on the basis by
f g(v
2n1
) = f(n)
f g(v
2n
) = g(n)
and extended linearly to the rest of V . It is trivial to check that is
bijective, that its linear in f and g, and that h(f g) = (hf) (hg), so that
is an R-module isomorphism.
R3:
(a) Give a description of all maximal ideals of the ring C[X, Y ]. Justify
your description. You may use the Nullstellensatz.
(b) Let M = (x
2
y, y
2
5) be an ideal in R = Q[X, Y ]. Prove that M
is a maximal ideal.
Solution.
ALGEBRA QUALS 17
(a) Let M C[X, Y ] be a maximal ideal. Since 1 / M, the Nullstellensatz
guarantees that x, y C with g(x, y) = 0 for all g M. Now
h C[X, Y ] : h(x, y) = 0
is an ideal containing M, and it is not equal to C[X, Y ] since it doesnt
contain any nonzero constants. Since M is maximal, it equals this
ideals, which can also be written as X x, Y y). So all maximal
ideals are of this form.
(b) Let f = X
2
Y and g = Y
2
5. Now
Q[X, Y ]/f, g) (Q[Y ]g))/f);
one can check this directly by writing an element on either side as
q(X, Y ) + af(X, Y ) + bg(Y ) and checking the obvious isomorphism.
Now Q[Y ]/g) Q(

5) because g is irreducible by Eisenstein, so


Q[X, Y ]/f, g) Q[

5][X]/X
2

5) Q[
4

5]
because X
2

5 is irreducible in Q[

5] (else it would have a root,


which would be degree 4 over Q because X
4
5 is irreducible by
Eisenstein; but everything in Q[

5] is degree at most 2 over Q).

You might also like