Você está na página 1de 11

1 LINEAR ALGEBRA

PH.D. COMPREHENSIVE EXAM


November 21, 2008
(Corrections and Additions: December 1, 2008)
1 Linear Algebra
3. Let P be a projection operator on the nite dimensional vector space V . (This means P
2
= P.) Let X = ker P =
nullspace P. Let Y = imP = image P. Dene an isomorphism : V

= X Y .
Solution 1: Let F be the eld over which V is dened. Dene (v) = (v Pv, Pv), for v V . Note that
v Pv ker P, since P(v vP) = Pv P
2
v = Pv Pv = 0. Therefore, is a map from V into X Y .
Also, is a homomorphism: ( F)(v
1
, v
2
V )
(v
1
+v
2
) = ((v
1
+v
2
) P(v
1
+v
2
), P(v
1
+v
2
))
= (v
1
Pv
1
, Pv
1
) + (v
2
Pv
2
, Pv
2
)
= (v
1
) +(v
2
).
Thus, Hom(V, X Y ). Next, dene : X Y V by (x, y) = x +y. Clearly Hom(X Y, V ), and
(v) = v Pv +Pv = v holds for all v V . Thus, = 1
V
. Also, for x ker P and y imP,
(x, y) = (x +y) = (x +y P(x +y), P(x +y))
= (x +y Px Py, Px +Py) = (x, y).
The last equality holds since x ker P, and since y imP implies Py = y. (Proof: y imP y = Pv for
some v V Py = P
2
v = Pv = y.) Therefore, = 1
XY
. This proves that is an isomorphism. .
Solution 2: As above, dene (v) = (v Pv, Pv), for v V , and note that Hom(V, X Y ). Let F be the
eld over which V is dened, and consider the following diagram of F[]-module homomorphisms:
0 X

V
P
Y 0

1
X

1
Y
0 X
1
X Y
2
Y 0
Obviously, the rows are exact. Also, if x X, then (x) = (x, 0) =
1
1
X
(x), so the rst box is commutative. If
v V , then
2
(v) =
2
(v Pv, Pv) = Pv = 1
Y
Pv, so the second box (hence the diagram) is commutative.
By the short-5 lemma, is an isomorphism. .
1
1 LINEAR ALGEBRA
4. Let V be a nite dimensional vector space. Let T be a linear transformation from V to V .
(a) Show that
V TV T
2
V
and that
0 ker T ker T
2
.
(b) Let V
1
=

i=1
T
i
V , and let V
2
=

i=1
ker T
i
. Show TV
1
V
1
and TV
2
V
2
.
(c) Show V = V
1
+V
2
and V
1
V
2
= 0.
(d) Let T
i
be the restriction of T to V
i
. Show that T
1
is invertible and T
2
is nilpotent.
Solution: (a) First note that T End(V ) implies TV V. Suppose T
n
V T
n1
V, and let v T
n+1
V. Then
v = T
n+1
u for some u V, so v = T
n
(Tu) = T
n
u

, where u

= Tu V (since TV V ). Therefore,
v T
n
V. This proves that T
n+1
V T
n
V and, by induction, V TV T
2
V .
Since T is linear, 0 ker T. (Proof: T(v) = T(v + 0) = T(v) + T(0) T(0) = 0.) Suppose ker T
n1

ker T
n
, and let v ker T
n
. Then v ker T
n+1
, since T
n+1
(v) = TT
n
(v) = T(0) = 0. This proves that
ker T
n
ker T
n+1
and, by induction, 0 ker T ker T
2
.
(b) Let v V
1
=

i=1
T
i
V . Fix n N. I claim that Tv T
n
V . Indeed, v V
1
implies v = T
n1
u for some
u V . Therefore, Tv = T
n
u T
n
V . Since n N was arbitrary, this shows that TV
1
V
1
.
Let v V
2
=

i=1
ker T
i
. If v ker T, then Tv = 0 V
2
. If v ker T
n
for some n > 1, then
0 = T
n
v = T
n1
Tv. Thus, Tv ker T
n1
V
2
. Therefore, TV
2
V
2
.
(c) By part (a), 0 ker T ker T
2
. I claim that, if ker T
n
= ker T
n+1
for some n N, then ker T
n
=
ker T
n+j
for all j N. For, suppose ker T
n
= ker T
n+1
for some n N, and v ker T
n+j
. Then
0 = T
n+j
v = T
n+1
T
j1
v implies T
j1
v ker T
n+1
= ker T
n
. This holds if and only if
T
n
T
j1
v = 0 T
n+1
T
j2
v = 0 T
j2
v ker T
n+1
= ker T
n
.
Continuing this way, after a nite number of steps, we reach v = T
jj
v ker T
n
, which proves ker T
n+j

ker T
n
. Combining this with part (a) gives ker T
n+j
= ker T
n
, as claimed.
Now, since V is nite dimensional and 0 ker T ker T
2
, it is clear that ker T
n
= ker T
n+1
must
hold for some n N, and then, by the foregoing, ker T
n
= ker T
n+1
= ker T
n+2
= . Let N be the
smallest positive integer such that ker T
N
= ker T
N+1
. Then V
2
=

i=1
ker T
i
= ker T
N
, and therefore,
V
1
=

i=1
T
i
V = T
N
V .
Two ways to complete the proof are as follows:
(i) If v T
N
V ker T
N
, then T
N
v = 0 and v = T
N
u for some u V . Therefore, 0 = T
N
v = T
2N
u,
which implies u ker T
2N
= ker T
N
, so v = T
N
u = 0. This shows T
N
V +ker T
N
= T
N
V ker T
N
.
It remains to show V = T
N
V + ker T
N
. But this follows from dim(V ) = dim(imf) + dim(ker f),
which holds for any f End(V ), in particular, f = T
N
.
(ii) Let F be the eld over which V is dened. Then V (as well as any subspace of V ) is a free (hence
projective) F[x]-module. In particular, T
N
V is projective, so the short exact sequence
0 ker T
N

V
T
N
T
N
V 0
is split exact. In particular, V

= ker T
N
T
N
V = V
2
V
1
.
(d) Note that ker T
1
= v V
1
: Tv = 0 = V
1
ker T, and ker T V
2
. Then, since V
1
V
2
= 0, ker T
1
= 0.
Therefore, T
1
is invertible.
Let N be the positive integer found in part (c), so that V
2
= ker T
N
. For v V
2
, T
2
v = Tv. Thus, TV
2
V
2
implies T
2
V
2
= TV
2
V
2
, so T
2
2
v = T
2
v for all v V
2
. That is, T
2
2
V
2
= T
2
V
2
. Inducting on k,
T
k
2
V
2
= T
k
V
2
, for all k N. In particular, T
N
2
V
2
= T
N
V
2
= 0. Therefore, T
2
is nilpotent. .
2
2 RING THEORY
PH.D. COMPREHENSIVE EXAM
November 21, 2008
(Corrections and Additions: December 1, 2008)
2 Ring Theory
All rings are unitary, i.e., have an identity.
1. List the ideals of the rings:
(a) Z/12Z
(b) M
2
(R) (the ring of 2 2 real matrices)
(c) Z/12Z M
2
(R).
Solution: (a) An ideal is, in particular, a subgroup of the additive group. Since 'Z
12
; +, 0` is cyclic, all
subgroups are cyclic. Thus, the subgroups of 'Z
12
; +, 0` are as follows:
(0) = 0
(1) = 0, 1, . . . , 11
(2) = 0, 2, 4, 6, 8, 10
(3) = 0, 3, 6, 9
(4) = 0, 4, 8
(6) = 0, 6
Of course, 5, 7, and 11 are relatively prime to 12, so (1) = (5) = (7) = (11) = Z
12
. The remaining
generators, 8, 4, and 9, are also redundant: (8) = (4), (9) = (3), and (10) = (2). It is easy to verify that
each of the 6 distinct subgroups listed above call them G
i
: 1 i 6 have the property that, if
r Z
12
and a G
i
, then ra G
i
. Thus, each G
i
, 1 i 6, is an ideal of Z
12
.
(Remark: This also shows Z
12
is a PID, hence the proper prime ideals are the maximal ideals, (2) and (3).)
(b) By the following lemma, since R is a eld (and thus has no non-trivial proper ideals), the only ideals of
M
2
(R) are M
2
(0) and M
2
(R).
Lemma 2.1 Let R be a commutative ring with 1
R
= 0. Then the ideals of M
2
(R) are precisely the subsets
M
2
(J) M
2
(R), where J is an ideal of R.
Proof: Suppose J is an ideal and A1, A2 M2(J). Then A1 A2 has all elements in J, since J is, in particular, a
subgroup. Therefore M2(J) is a subgroup of M2(R); +, 0. If M = (mij) M2(R) and A = (aij) M2(J), then
MA = (
P
2
k=1
m
ik
a
kj
). Clearly, all the elements of MA are in J, since J is an ideal and aij J. Similarly, all the
elements of AM are in J. This proves that M2(J) is an ideal of M2(R) whenever J is an ideal of R.
Now, let I M2(R) be an ideal of M2(R), and let J be the set of all a R such that a is an element of some
A I. Clearly 0 J, since A I implies 0 = A A I. Let a, b J. Suppose a is the ij
th
entry of
A I and b is the kl
th
entry of B I. Let Mij be a matrix with 1 in the ij
th
position and 0 elsewhere. Then
M1iAMj1 =

a 0
0 0

I and M
1k
BM
l1
=

b 0
0 0

I. Therefore,

a b 0
0 0

I, so a b J, whence,
J is a subgroup. Fix r R. Then

r 0
0 0

a 0
0 0

ra 0
0 0

I, so ra J, and J is an ideal.
3
2 RING THEORY
(c) By the following lemma, and parts (a) and (b), the ideals of Z/12ZM
2
(R) are all sets of the form I J,
where I (0), (1), (2), (3), (4), (6) and J M
2
(0), M
2
(R).
Lemma 2.2 Let R and R

rings. Then the ideals of R R

have the form I J, where I is an ideal of R


and J is an ideal of R

.
Proof: Let A R R

be an ideal. I will show (i) A = A1 A2, for some A1 R and A2 R

, and (ii) A1 and


A2 are ideals of R and R

, respectively.
(i) Dene
A1 = {a R : (a, a

) A for some a

} and A2 = {a

: (a, a

) A for some a R}.


That is, A1 (resp. A2) is the set of all rst (resp. second) coordinates of elements in A. I claim that A = A1 A2.
1
Fix a A1 and b

A2. We show (a, b

) A. Since a A1, there is some a

A2 such that (a, a

) A. Similarly,
(b, b

) A, for some b A1. Since A is an ideal, (1, 0) (a, a

) = (a, 0) A, and (0, 1) (b, b

) = (0, b

) A.
Therefore, (a, 0) + (0, b

) = (a, b

) A, as claimed. This proves A1 A2 A. The reverse inclusion is obvious.


(ii) Fix a, b A1. Then, there exist a

, b

A2 such that (a, a

) A and (b, b

) A. Now, since A is, in particular, a


subgroup of the additive group of R R

, we have (a, a

) (b, b

) = (a b, a

) A. Therefore, a b A1,
so A1 is a subgroup of R, +, 0. Fix r R. Then (ra, a

) = (ra, 1a

) = (r, 1) (a, a

) A, since A is an ideal of
RR

. Therefore, ra A1, which proves that A1 is an ideal of R. The same argument, mutatis mutandis, proves that
A2 is an ideal of R

. This completes the proof of (ii) and establishes the lemma.


1
Although this may seem trivial, and to some extent it is trivial, there is something to verify here, since we could have (a, a

) A, (b, b

) A,
so that (a, b

) A
1
A
2
, and yet (a, b

) / A. That is, in general, A need not be equal to the product of all rst and second coordinates of
elements of A. (Consider, for example, the diagonal {(0, 0), (1, 1), (2, 2)}, which cannot be written as A
1
A
2
.)
4
2 RING THEORY
3. Let R be a commutative ring and let A, B, and C be R-modules with A a submodule of B. Which (if any) of the
following conditions guarantee that the natural map A
R
C B
R
C is injective?
(a) C is free,
(b) C is projective.
Solution: I will prove a slightly more general result (as the increased generality comes at no additional cost).
That is, if : A B is an R-module monomorphism, then, under conditions (a) or (b), the natural map 1
C
:
A
R
C B
R
C in injective. (To answer the question above, take to be the inclusion map.)
We begin with the simplest case, where C = R.
Claim 1: If : A B is an R-module monomorphism, then the natural map 1
R
: A
R
R B
R
R is
injective.
Proof 1: Dene
A
: A
R
R

= A by
A
(a r) = ar, and
B
: B
R
R

= B by
B
(b r) = br. (See
lemma 2.4 below for proof that these are, indeed, isomorphisms.) I claim that the following diagram of R-module
homomorphisms is commutative:
A
R
R

A
A
1
R

B
R
R

B
B
Fix a A and r R. By denition, ( 1
R
)(a r) = (a) 1
R
(r). Therefore, (
B
( 1
R
))(a r) =

B
((a) r) = (a)r. Following the diagram in the other direction, (
A
)(a r) = (ar) = (a)r, since is
an R-module homomorphism. Therefore, (
B
( 1
R
))(a r) = (
A
)(a r). Since a A and r R were
chosen arbitrarily,
B
( 1
R
) and
A
agree on generators, so the diagram is commutative.
Now
A
is an isomorphism, so
A
and are both injective. Therefore, by lemma 2.6 (below),
A
is injective.
Then, by commutativity of the diagram,
B
( 1
R
) is injective. Finally, lemma 2.8 implies that 1
R
is
injective, which proves claim 1.
Next, let C be a free R-module. Then C

=

iI
R, for some index set I. For ease of notation, in the sequel,

denotes

iI
.
Claim 2: If : A B is an R-module monomorphism, then the natural map 1
P
R
: A
R

R B
R

R
is injective.
Proof 2: Dene :

A

B by (a
i
) = (a
i
) for each a
i


A. Then is an R-module
monomorphism. This is seen as follows: (a
i
, a

A) (r R)
(a
i
+a

i
) = (a
i
+a

i
) = (a
i
+a

i
) = (a
i
) +(a

i
) = (a
i
)+(a

i
) = (a
i
) +(a

i
),
and (ra
i
) = (ra
i
) = (ra
i
) = r(a
i
) = r(a
i
) = r(a
i
).
If (a
i
) = 0

B, then (a
i
) = 0, which implies (a
i
) = 0 for all i I, and therefore, a
i
= 0

A. Thus, is a monomorphism.
By (the proof of) lemma 2.4, the map :

(A
R
R) given by (a
i
) = a
i
1
R
is an isomorphism.
By (the proof of) lemma 2.5, there is an isomorphism :

(A
R
R) A
R

R, dened on generators by
(a r, 0, . . . ) = a (r, 0, . . . ), (0, a r, 0, . . . ) = a (0, r, 0, . . . ), . . . , and
5
2 RING THEORY
(a
1
r
1
, a
2
r
2
, . . . ) = ((a
1
r
1
, 0, . . . ) + (0, a
2
r
2
, 0, . . . ) + )
= (a
1
r
1
, 0, . . . ) +(0, a
2
r
2
, 0, . . . ) +
= a
1
(r
1
, 0, . . . ) +a
2
(0, r
2
, 0, . . . ) + .
(The sum has nitely many non-zero terms, since a
i
is non-zero for nitely many indices i I.)
Consider the following diagram of R-module homomorphisms:


(A
R
R)

A
R

1
P
R


(B
R
R)

B
R

R
(1)
where and are the same maps as their bar-less counterparts, but dened on

B and

(B
R
R), respectively.
I claim that diagram (1) is commutative. Indeed, for any a
i

A, we have
2
(a
i
) = (a
i
1
R
) = ((a
1
1
R
, 0, . . . ) + (0, a
2
1
R
, 0, . . . ) + )
= a
1
(1
R
, 0, . . . ) +a
2
(0, 1
R
, 0, . . . ) + .
( 1
P
R
) (a
i
) = ( 1
P
R
)(a
1
(1
R
, 0, . . . ) +a
2
(0, 1
R
, 0, . . . ) + )
= ( 1
P
R
)(a
1
(1
R
, 0, . . . )) + ( 1
P
R
)(a
2
(0, 1
R
, 0, . . . )) +
= (a
1
) (1
R
, 0, . . . ) +(a
2
) (0, 1
R
, 0, . . . ) + .
In the other direction, (a
i
) = ((a
i
)) = (a
i
) 1
R
, so
(a
i
) = ((a
i
) 1
R
) = (((a
1
) 1
R
, 0, . . . ) + (0, (a
2
) 1
R
, 0, . . . ) + )
= (a
1
) (1
R
, 0, . . . ) +(a
2
) (0, 1
R
, 0, . . . ) + ,
which proves that and ( 1
P
R
) agree on generators. Therefore, diagram (1) is commutative.
Now, and are R-module isomorphisms, and is an R-module monomorphism. Therefore, by lemma 2.6
below, is an R-module monomorphism, so commutativity of (1) implies that 1
P
R
is also an R-
module monomorphism. Finally, since and are isomorphisms, so is . Whence, 1
P
R
is injective, by
lemma 2.7. This which proves claim 2 and completes part (a) of the problem.
Claim 3: If C is a projective R-module and : A B is an R-module monomorphism, then the natural map
1
C
: A
R
C B
R
C is injective.
Proof 3: Since C is projective, there exists a free R-module F and a (projective) R-module D such that F = CD
(lemma 2.3). Therefore,
A
R
F = A
R
(C D)

= (A
R
C) (A
R
D) and B
R
F = B
R
(C D)

= (B
R
C) (B
R
D),
where the isomorphisms are given by lemma 2.5.
By claim 2 above, the natural map 1
F
: A
R
F B
R
F is injective. Consider the diagram
A
R
C
1
(A
R
C) (A
R
D)

A
R
(C D)
1
C

1
F
B
R
C
1
(B
R
C) (B
R
D)

B
R
(C D)
(2)
2
Again, the sums have nitely many non-zero terms, since {a
i
} is non-zero for nitely many indices i I. The same comment applies to the
sums in the following two sets of equations.
6
2 RING THEORY
By (the proof of) lemma 2.5 the map : (A
R
C) (A
R
D) A
R
(C D) given by (a
1
c, a
2
d) =
a
1
(c, 0) +a
2
(0, d) is an R-module isomorphism. Given a c A
R
C, then,
( 1
F
)
1
(a c) = ( 1
F
) (a c, 0) = ( 1
F
)(a (c, 0)) = (a) (c, 0).
Similarly, there exists : (B
R
C)(B
R
D)

= B
R
(CD), with (b
1
c, b
2
d) = b
1
(c, 0)+b
2
(0, d),
and

1
( 1
C
)(a c) =
1
((a) c) = ((a) c, 0) = (a) (c, 0).
Therefore, ( 1
F
)
1
and
1
( 1
C
) agree on generators of A
R
C, so diagram (2) is commutative.
By lemma 2.6, ( 1
F
)
1
is injective, so, by commutativity,
1
( 1
C
) is injective. It now follows from
lemma 2.8 that 1
C
is injective, which completes the proof of claim 3 and part (b) of the problem. .
The following six lemmas are used in the answer to problem 3 given above. The rst is a standard theorem about
projective modules, the proof of which is not hard, and can be found, e.g., in Hungerford. The next two lemmas
(2.4 and 2.5) are also standard, but I havent seen them proved in detail elsewhere and, as the solution given above
makes repeated use of the maps dened in proving these lemmas, I include detailed proofs below. The last three
lemmas are trivial verications.
Lemma 2.3 Let R be a ring. The following conditions on an R-module P are equivalent:
(i) P is projective;
(ii) every short exact sequence of R-modules 0 A
f
B
g
P 0 is split exact (hence B

= AP);
(iii) there is a free R-module F and an R-module N such that F

= N P.
Lemma 2.4 If R is a commutative ring with 1
R
and A is a unitary R-module, then A
R
R

= A.
Proof: Dene : A R A by (a, r) = ar. Since is clearly bilinear, there exists a unique R-module homomorphism
: A R R A such that = , where : A R A R R is the canonical bilinear map. Dene : A A R R by
(a) = a 1R. Then it is easy to verify that is an R-module homomorphism and that
(a r) = (ar) = ar 1R = a r (a A)(r R).
Therefore, is the identity on generators of A R R. Also, (a) = (a 1R) = a1R = a, so = 1A. Therefore,
: AR R

= A.
Lemma 2.5 Let R be a commutative ring with 1
R
and let A, B, C be unitary R-modules. Then A
R
(B C)

=
(A
R
B) (A
R
C).
Proof: Recall that if 1 : A1 D and 2 : A2 D are (group) homomorphisms, then there is a unique homomorphism
: A1 A2 D such that i = i (i = 1, 2), where i : Ai A1 A2 are the canonical injections. In other words,
! Hom(A1 A2, D) such that the following diagram is commutative:
A1

1
$$
I
I
I
I
I
I
I
I
I

1
//
A1 A2

A2

2
oo

2
zzu
u
u
u
u
u
u
u
u
D
We can apply this theorem to the Abelian group (A R B) (A R C). Let 1 : A B A R (B C) be given by
1(a, b) = a (b, 0), and let 2 : AC AR (B C) be given by 2(a, c) = a (0, c). It is easily veried that i are
7
2 RING THEORY
bilinear and thus induce R-module homomorphisms 1 : A R B A R (B C) and 2 : A R C A R (B C)
such that ii = i, where 1 : A B A R B and 2 : A C A R C are the canonical bilinear maps. Therefore,
the two universal properties combine to give a unique homomorphism : (AR B) (AR C) AR (BC) such that
the following diagram is commutative:
AR B

1
((
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q
Q

1
//
(AR B) (AR C)

AR C

2
vvm
m
m
m
m
m
m
m
m
m
m
m
m

2
oo
AB

1
OO

1
//
AR (B C) AB

2
OO

2
oo
Now, for all a A and b B, 1(a b) = (a b, 0) and 1(a b) = 11(a, b) = 1(a, b) = a (b, 0), so,
(a b, 0) = a (b, 0). Similarly, (0, a c) = a (0, c), for all a A and c C. Therefore,
(a b, a

c) = (a b, 0) + (0, a

c) = a (b, 0) + a

(0, c) (a, a

A, b B, c C).
Next, dene : A(B C) (AR B) (AR C) by (a, (b, c)) = (a b, a c). Again, is bilinear, so there is a
unique : AR(BC) (BC) (ARB)(ARC) such that = where : A(BC) AR(BC)
is canonical. To complete the proof, we check that and given the appropriate identity maps:
(a b, 0) = (a (b, 0)) = (a b, a 0) = (a b, 0), and
(0, a c) = (a (0, c)) = (a 0, a c) = (0, a c).
Thus, is the identity on generators of (AR B) (AR C), so = 1
(A
R
B)(A
R
C)
. Finally,
(a (b, c)) = (a b, a c) = ((a b, 0) + (0, a c)) = a (b, 0) + a (0, c) = a (b, c).
Thus is the identity on generators of AR(BC), so = 1
A
R
(BC)
. This proves that : (ARB) (ARC)

=
AR (B C).
Lemma 2.6 If h Hom
R
(A, B), g Hom
R
(B, C), and f Hom
R
(C, D) are injective, then fgh Hom
R
(A, D)
is injective.
Proof: fgh(a) = 0 gh(a) = 0 (since f is injective) h(a) = 0 (since g is injective) a = 0 (since h is injective).
Lemma 2.7 If g Hom
R
(A, B), f Hom
R
(B, C), if g is surjective, and if fg is injective, then f is injective.
Proof: Suppose f(b) = 0. Since g is surjective, there is an a A such that g(a) = b. Then fg(a) = f(b) = 0, which implies
a = 0, since fg is injective. Therefore, b = g(a) = g(0) = 0 since g is a homomorphism.
Lemma 2.8 If g Hom
R
(A, B), f Hom
R
(B, C), and if fg is injective, then g is injective.
Proof: ker g = {a A : g(a) = 0} {a A : fg(a) = 0} = ker fg = {0}. The set containment holds since f is a
homomorphism, so f(0) = 0.
8
3 GROUPS
PH.D. COMPREHENSIVE EXAM
November 21, 2008
(Corrections and Additions: December 1, 2008)
3 Groups
4. Let G be a nite group, all of whose Sylow subgroups are normal. Prove that G is the product of its Sylow
subgroups.
Solution:
3
Let P
1
, . . . , P
r
be the Sylow subgroups of G with orders [P
i
[ = p
ei
i
, where p
1
, . . . , p
r
are distinct
primes and e
1
, . . . , e
r
are positive integers. Since P
i
G for each 1 i r, P
1
P
r
is a subgroup of G. If
a P
i
has order o(a), then o(a)[p
ei
i
, by Lagranges theorem. Since p
1
, . . . , p
r
are distinct primes, o(a) p
ej
j
, for
j = i. Therefore, again by Lagranges theorem, a / P
1
P
i1
P
i+1
P
r
. This proves that, for each 1 i r,
P
i
(P
1
P
i1
P
i+1
P
r
) = (e). Two facts almost
4
immediately:
(i) P
1
P
r

= P
1
P
r
, and
(ii) [P
1
P
r
[ = [P
1
[ [P
r
[ = p
e1
1
p
er
r
.
Finally, P
1
P
r
< G and [P
1
P
r
[ = p
e1
1
p
er
r
= [G[ together imply that G = P
1
P
r
. Therefore, by (i),
G

= P
1
P
r
. .
3
I believe my original solution to this problem was essentially correct, as the problem asks for proof that Gis the product of its Sylow subgroups.
Nonetheless, here it is again, with a minor addition: we can conclude that G is (isomorphic to) the direct product of its Sylow subgroups.
4
See Theorem 8.6 and Corollary 8.7 of Hungerford, p. 61, for the proof of (i).
9
4 FIELDS
PH.D. COMPREHENSIVE EXAM
November 21, 2008
(Corrections and Additions: December 1, 2008)
4 Fields
2. Let F be a eld with p
n
elements (p a rational prime). Prove that F is a Galois extension of its prime subeld with
cyclic Galois group. (You may assume any result not specic to nite elds.)
Solution 1: Consider f(x) = x
p
n
x Z
p
[x]. Since F is a eld with p
n
elements, the multiplicative group
F

= F `0 is cyclic of order p
n
1. By Lagranges theorem, then, a
p
n
1
= 1 holds for all a F

. Therefore,
for all a F (including 0), a
p
n
= a. That is, the p
n
elements of F give p
n
distinct roots of f(x). Since f(x) has
at most p
n
roots, these are all the roots, so F is a splitting eld for f(x) over Z
p
. Since the roots are all distinct,
f(x) is separable. This proves that F/Z
p
is a nite Galois extension.
To see that Gal(F/Z
p
) is cyclic, consider the map : F F given by (a) = a
p
. Since F has characteristic
p, (a b) = (a b)
p
= a
p
b
p
. Also, (ab) = (ab)
p
= a
p
b
p
. Thus, End(F). Fix a F. Then
b = a
p
n1
F, so (b) = (a
p
n1
)
p
= a
p
n
= a, and is surjective. Also, a ker = a F : a
p
= 0 implies
a = a
p
n
= (a
p
)
p
n1
= 0, so ker = 0, and is injective. Therefore, Aut(F). Finally, (r) = r
p
= r, for
r Z
p
. To see this, note that Z

p
is a cyclic group of order p 1. Therefore, r
p1
= 1, for all r Z

p
, so r
p
= r,
for all r Z
p
. This proves that Gal(F/Z
p
).
If a F, then
n
(a) =
n1
(a
p
) =
n2
(a
p
2
) = = a
p
n
= a. Therefore,
n
= 1
F
. Also, if
k
= 1
F
for some 1 k < n, then a
p
k
= a for all a F, which says that the polynomial x
p
k
x has p
n
roots,
where p
k
< p
n
(nonsense). Therefore,
n
= 1
F
, and n is the smallest positive integer with this property, so
() = ,
2
, . . . ,
n
= 1
F
is a set of n distinct elements in Gal(F/Z
p
). Finally, n = [F : Z
p
] = [Gal(F/Z
p
)[,
since F/Z
p
is Galois. This proves that Gal(F/Z
p
) is cyclic with generator : a a
p
. .
Solution 2: Dene (a) = a
n
. As above, Gal(F/Z
p
), and () is a subgroup of order n. By the fundamental
theorem of Galois theory, [F : Inv()] = n and Gal(F/Inv()) = (). On the other hand, Gal(F/Z
p
)
implies Z
p
Inv(). Since n = [F : Z
p
] = [F : Inv()][Inv() : Z
p
] = n[Inv() : Z
p
], we have [Inv() :
Z
p
] = 1, so Inv() = Z
p
. Equivalently, by the theorem on Galois extensions,
5
F/Z
p
is Galois, with Galois group
Gal(F/Z
p
) = (). .
5
See page 4 of Math 612 Notes: Galois Theory, William Lampe, July 2008.
10
4 FIELDS
3. Let F be a eld of characteristic zero such that every nite extension of F has even degree. Show that
(a) every nite extension of F has degree a power of 2,
(b) all proper minimal nite extensions of F are quadratic.
(Hint: Sylow theory can be applied to Galois groups of normal extensions of F.)
Solution: (a) Suppose K is a nite extension of F. Let

K be the normal closure of K over F. Since F has
characteristic zero,

K is separable and therefore Galois with [Gal(

K/F)[ = [

K : F], which is even, by
assumption. Thus, without loss of generality, we can assume [Gal(

K/F)[ = 2
k
m, for some positive integers
k and m, with m odd. Let G = Gal(

K/F). By the rst Sylow theorem, G has a Sylow 2-subgroup, H, of
order [H[ = 2
k
. Therefore, by the fundamental theorem of Galois theory, [

K : InvH] = [H[ = 2
k
, where
InvH = a

K : (a) = a for all H. Thus 2
k
m = [

K : F] = [

K : InvH][InvH : F], which
implies [InvH : F] = m, an odd integer. If InvH is a (proper) nite extension of F, then, by assumption,
it must have even degree. The only way out is m = [InvH : F] = 1, which gives [

K : F] = 2
k
. Finally,
[

K : F] = [

K : K][K : F] implies [K : F] [ 2
k
. Thus , [K : F] is a power of 2.
(b) Let E be a proper nite extension of F, so that [E : F] > 1. By part (a), [E : F] = 2
k
for some k 1. If
[E : F] = 2, I claim that there is a proper intermediate eld F < K < E such that 1 < [K : F] < [E : F].
(This will establish that, if [E : F] = 2, then E is not a proper minimal nite extension of F, which will
complete the proof.)
Suppose [E : F] = 2. Then [E : F] = 2
k
with k > 1. Let

E be the normal closure of E over F. Then, by part
(a), [G[ = [Gal(

E/F)[ = 2
K
= [

E : F], where 1 < k K < . Consider the subgroup
FixE = G : (a) = a for all a E.
As a subgroup of G, [FixE[ = 2
j
for some 0 j K. Also, [G[/[FixE[ = [G : FixE] = [E : F] > 2, so
FixE is a proper subgroup of G.
Next, consider N
G
(FixE), and recall the following standard lemma about normalizers of p-subgroups: if H is
a p-subgroup of a nite group G, then [N
G
(H) : H] [G : H] (mod p). By this lemma, we have
1 [N
G
(FixE) : FixE] [G : FixE] 0 (mod 2).
Therefore, 2 divides [N
G
(FixE)/FixE[, so N
G
(FixE)/FixE contains a subgroup of order 2, by Cauchys
theorem. The third isomorphism theorem implies that this subgroup has the formJ
1
/FixE, where FixEJ
1
<
N
G
(FixE), and [J
1
/FixE[ = 2, which implies [J
1
[ = 2[FixE[ = 2
j+1
. If J
1
G, stop. Otherwise, repeat the
foregoing argument and obtain a subgroup J
2
with the properties FixE J
1
J
2
< N
G
(J
1
) and [J
2
[ = 2
j+2
.
After a nite number of steps, we obtain a normal subgroup J
r
G of order [J
r
[ = 2
j+r
. (Note: this will be a
proper normal subgroup, i.e., [G : J
r
] > 1, since any subgroup of order 2
K1
is of index 2, hence normal, in
G.)
Finally, consider the subeld InvJ
r
= a

E : (a) = a for all J
r
. Since Inv is order reversing on the
poset 'Sub[G], `, it is clear that FixE < J
r
implies
6
InvJ
r
< InvFixE = E. Putting it all together, we have
1 < [InvJ
r
: F] = [G : J
r
] < [G : FixE] = [InvFixE : F] = [E : F].
That is, F < InvJ
r
< E and 1 < [InvJ
r
: F] < [E : F], as claimed. .
6
The equality InvFixE = E follows from the subgroup lemma of Professor Lampes 612 notes on Galois theory: If

E/F is a nite Galois
extension and F < E <

E, then

E is Galois over E. Consequently, in the algebra of intermediate elds, E is closed (with respect to the closure
operator InvFix).
11

Você também pode gostar