You are on page 1of 6

Algebra Exam – Fall 2005

Alexander J. Wertheim
Last Updated: December 6, 2017

Contents
1 Groups 2
1.1 Problem 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Problem 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.3 Problem 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

2 Rings 3
2.1 Problem 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.2 Problem 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
2.3 Problem 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

3 Fields 4
3.1 Problem 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.2 Problem 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
3.3 Problem 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1
1 Groups
1.1 Problem 1
Theorem 1.1.1. Let R be a principal ideal domain, and let M be a free R-module on n < ∞
generators. Then any submodule N ⊂ M is a free R module on at most n generators.
Proof. Without loss of generality, we may take M = Rn . We proceed by induction on n. If
n = 1, then a submodule of M = R corresponds with an ideal of R, which must be generated
by one element of R by hypothesis. Assume the claim is true for the case n − 1, and let
N ⊂ M = Rn be a submodule. Consider the projection homomorphism π : Rn → Rn−1
given by π(r1 , r2 , . . . , rn ) = (r2 , r3 , . . . , rn ). Then N 0 = π(N ) is a submodule of Rn−1 , and
hence is free on 6 n − 1 generators by the induction hypothesis. Letting N 00 = ker(π|N ), we
have an exact sequence
π|N
0 → N 00 ,→ N −−→ N 0 → 0
which must be split, because N 0 is free and therefore projective. Thus, N = N 0 ⊕ N 00 , so
it suffices to show that N 00 is a free module on 1 generator, if we take for granted that
the direct sum of two free R-modules of finite rank k, l respectively has rank k + l. It is
evident that every element of ker(π) has the form (r, 0, . . . , 0) for some r ∈ R, so we have
an injection ρ : ker(π) → R given by (r, 0, . . . , 0) 7→ r. Then we have a series of injections
ρ
N 00 ,→ ker(π) →
− R, whence N 00 is (isomorphic to) a submodule of R, and hence is free on 1
generator as remarked above.
Corollary 1.1.2. If G is a free abelian group of finite rank r, then any subgroup of G is free
of rank at most r.
Proof. This is the conclusion of the theorem above applied with R = Z.
Let G be an abelian group finitely generated by elements g1 , . . . , gn , and let H ⊂ G be a
subgroup. Then there is a surjection Φ : Zn → G which maps the basis element ei ∈ Zn to
gi for each 1 6 i 6 n. Note Γ := Φ−1 (H) is a subgroup of Zn , and hence is free of finite rank
k 6 n by the corollary above. The restriction Φ|Γ is surjective onto H, and letting f1 , . . . , fk
be the basis elements for Γ, the images Φ(f1 ), . . . , Φ(fk ) must generate H.

1.2 Problem 2
Let N be a normal subgroup of a finite group G, and let P be a Sylow p-subgroup of G.
Let |G| = pk m for some m coprime to p, so |P | = pk . Since N is a normal subgroup of G,
P N is a subgroup of G, whence |P N | divides |G| by Lagrange. Write |P N | = pj s for some
0 6 j 6 k and s coprime to p, and similarly |N | = pl r for some 0 6 l 6 k and r coprime to
p. We have the formula |P N | = (|P | · |N |)/|P ∩ N |, so |P ∩ N | = (|P | · |N |)/(|P N |), and

pk · pl r r
|P ∩ N | = j
= pk+l−j ·
ps s
Since P ∩ N is a subgroup of P , |P ∩ N | divides |P | = pk by Lagrange, so r/s = 1 and
|P ∩ N | = pk+l−j . From above, j 6 k, so k + l − j > l; but since P ∩ N is a subgroup of N

2
too, |P ∩ N | divides |N | by Lagrange, so k + l − j 6 l, whence k = j and |P ∩ N | = pl , so
P ∩ N is a Sylow p-subgroup of N .

1.3 Problem 3
First, note that A4 does not have a subgroup of order 6. Indeed, suppose G ⊂ A4 were such
a subgroup; then G is necessarily normal, since |A4 | = 12 and any subgroup of index 2 is
necessarily normal. Since A4 has eight 3-cycles, there must be some 3-cycle x ∈ G which
does not belong to G. The quotient group A4 /G has order two, and so two of the cosets
G, xG, x2 G must coincide. If G = xG or xG = x2 G, then x ∈ G, a contradiction.If G = x2 G,
then likewise xG = x3 G = G, so x ∈ G, a contradiction.
Let X be a set with two elements, and suppose there were a nontrivial action of A4 on X,
i.e. ϕ : A4 → S(X) ∼ = S2 . Then since |S2 | = 2, ϕ must be surjective, so A4 / ker(ϕ) ∼
= S2 ,
whence Lagrange implies
|A4 | 12
|A4 / ker(ϕ)| = = =2
| ker(ϕ)| | ker(ϕ)|
i.e. | ker(ϕ)| = 6, contradicting the above.

2 Rings
2.1 Problem 1
Let I and J be ideals of a commutative ring R with unity such that I + J = R. The
containment IJ ⊆ I ∩ J holds, since IJ ⊂ IR = I and IJ ⊂ RJ = J. Let k ∈ I ∩ J. Since
I + J = R, there exist i ∈ I, j ∈ J such that i + j = 1, so k = k · 1 = k(i + j) = ki + kj.
Since k ∈ J, ki = ik ∈ IJ, and since k ∈ I, kj ∈ IJ, so k ∈ IJ.

2.2 Problem 2
Lemma 2.2.1. Let R be a UFD and K its quotient field. Then R is integrally closed in K.
Proof. Let r/s ∈ K such that r and s have no common irreducible factors, and suppose
p(X) = X n + a1 X n−1 + · · · + an ∈ R[X] has r/s as a root. Then
 r   r n  r n−1
p = + a1 + · · · + an = 0.
s s s
Multiplying both sides by sn , we obtain
rn + a1 srn−1 + · · · + sn an = 0
and so
rn = a1 srn−1 + · · · + sn an
Any irreducible divisor d of s divides the RHS, and hence divides rn . Since the notions of
prime and irreducible coincide in a UFD, any such d must divide r. Since r and s have no
common irreducible factors, the list of irreducible divisors of s must be empty, i.e. s ∈ R× ,
whence r/s ∈ R.

3
The factor ring J := R[X, Y ]/(Y 2 − X 3 )R[X, Y ] is not a UFD. Intuitively, if X, Y rep-
2 3
resent the equivalence classes of X and Y in J, the equality Y = X gives two distinct
factorizations of the same element into irreducibles. Proving that X and Y are irreducible
elements of J is cumbersome, however. An easier approach is to use the lemma above, and
show that J is not integrally closed in its quotient field F .
Note that X is a nonzero class of J by degree considerations, so put α := Y /X ∈ F . Then
α is a root of the monic polynomial T 2 − X ∈ J[T ], and hence is integral over J. However,
α∈ / J, since otherwise there would exist f ∈ R[X, Y ] such that Y − f X ∈ hY 2 − X 3 i, which
is impossible by degree considerations. Explicitly, every nonzero monomial term X n Y m of
any element of hY 2 − X 3 i satisfies min{n, m} > 2, whereas Y − f X has the monomial term
Y of degree 1 for any f ∈ R[X, Y ].

2.3 Problem 3
Consider the (left) R-module homomorphism Φ : R ⊕ R → R given by (r, s) 7→ rx + sy. This
map is surjective, since (z, t) 7→ zx + ty = 1. If (r, s) 7→ 0, then

(rx + sy)z = rxz + syz = r(1) + s(0) = 0

so r = 0; similarly,
(rx + sy)t = rxt + syt = r(0) + s(1) = 0
so s = 0, which establishes injectivity. Note that the inverse map is given by r 7→ (rz, rt).

3 Fields
3.1 Problem 1
This is a special case of W05.F3.

3.2 Problem 2
Let ϕ(X) be an irreducible polynomial of degree 4 in Q[X] and let K be the field generated
by the complex roots α1 , α2 , α3 , α4 of ϕ. Let F be the subfield of K generated by

β1 = (α1 + α2 )(α3 + α4 )

β2 = (α1 + α3 )(α2 + α4 )
β3 = (α1 + α4 )(α2 + α3 )
Note that α1 , α2 , α3 , α4 are pairwise distinct, since ϕ is irreducible over a field of characteristic
0, hence separable. Furthermore, K is the splitting field of ϕ over Q, so K/Q is indeed a
Galois extension. First, note that β1 , β2 , β3 are pairwise distinct as follows:

β1 − β2 = α1 α3 − α1 α2 + α2 α4 − α3 α4 = (α1 − α4 )(α2 − α3 ) 6= 0

β1 − β3 = α2 α3 − α1 α2 + α1 α4 − α3 α4 = (α2 − α4 )(α3 − α1 ) 6= 0

4
β2 − β3 = α2 α3 − α2 α4 + α1 α4 − α1 α3 = (α2 − α1 )(α3 − α4 ) 6= 0
Since Gal(K/Q) must permute the roots of ϕ, Gal(K/Q) can be realized as a subgroup of
S4 , with S4 acting on the roots of ϕ by permuting the subscripts. In particular, S4 acts
on the set {β1 , β2 , β3 }, and one can compute the stabilizer subgroups Γ1 , Γ2 , Γ3 of β1 , β2 , β3
respectively as follows:

Γ1 = {e, (12), (34), (12)(34), (13)(24), (14)(23)}

Γ2 = {e, (13), (24), (12)(34), (13)(24), (14)(23)}


Γ3 = {e, (14), (23), (12)(34), (13)(24), (14)(23)}
Since β1 , β2 , β3 are pairwise distinct, the subgroup V of S4 which fixes all three generators
of F is given by the intersection of the stabilizer subgroups Γ1 , Γ2 , Γ3 , so

V := Γ1 ∩ Γ2 ∩ Γ3 = {e, (12)(34), (13)(24), (14)(23)}

which isomorphic to the Klein 4-group Z/2Z × Z/2Z, hence abelian. The automorphisms
of K fixing F are elements of Gal(K/Q) which fix each of the three generators β1 , β2 , β3 , so
Gal(K/F ) = Gal(K/Q) ∩ V . Hence, Gal(K/F ) is a subgroup of V , so Gal(K/F ) is abelian.

3.3 Problem 3
Let f (X) = X 11 − 7 ∈ Q[X], and let ξ be a primitive 11th root of unity. Note that f
is irreducible over Q by Eisenstein at 7, so letting α be a root of f , [Q(α) : Q] = 11.
Further, recall that the minimal polynomial mξ (X) ∈ Q[X] of ξ over Q is given by mξ (X) =
X 10 + X 9 + · · · + X + 1, so [Q(ξ) : Q] = 10. Over C, f (X) splits as
10
Y
f (X) = (X − ξ i α)
i=0

so the splitting field K of f over Q is given by Q(α, ξα, . . . , ξ 10 α) = Q(α, ξ). Note that K
is Galois over Q, since it is the splitting field of a separable polynomial over Q. Further,
since deg(α), deg(ξ) are relatively prime, [K : Q] = [Q(α, ξ) : Q] = [Q(α) : Q] · [Q(ξ) :
Q] = 11 · 10 = 110. Let G = Gal(K/Q); by the fundamental theorem of Galois theory,
|G| = [K : Q] = 110. Any element σ ∈ G is determined by the images σ(ξ) and σ(α), and
any automorphism of K must map roots to roots. Hence, we must have σ(ξ) = ξ i for some
1 6 i 6 10 and σ(α) = ξ k α for some 0 6 k 6 10. Since there are 11 · 10 possible assignments
and G has 110 elements, every such σ must be realized, so in particular, there exist σ, τ ∈ G
defined on generators by
σ(α) = ξα σ(ξ) = ξ
τ (α) = α τ (ξ) = ξ 2
It is straightforward to see that σ has order 11 and τ has order 10, so H := hσi and K := hτ i
are subgroups of G of order 11, 10 respectively. By the third Sylow theorem, the number
N11 of Sylow 11-subgroups of G is congruent to 1 mod 11 and divides 10, so N11 = 1. Hence,

5
H is the unique Sylow 11-subgroup of G and thus must be normal. Then HK is a subgroup
of G, and by order considerations, H ∩ K = {e}, so

|H| · |K| 110


|HK| = = = 110
|H ∩ K| 1

whence HK = G. Thus, G is the semidirect product of H and K, induced by a group


homomorphism ϕ : K → Aut(H). In particular, G is generated by σ, τ , and τ στ −1 must be
a power of σ. Explicitly,

τ (σ(τ −1 (α))) = τ (σ(α)) = τ (ξα) = τ (ξ)τ (α) = ξ 2 α = σ 2 (α)

τ (σ(τ −1 (ξ))) = τ (σ(ξ 6 )) = τ (ξ 6 ) = ξ 12 = ξ = σ 2 (ξ)


so τ στ −1 = σ 2 (α), i.e. τ σ = σ 2 τ . Hence, G ∼ = hσ, τ | σ 11 = τ 10 = e, τ σ = σ 2 τ i ∼
= H o K.
Finally, by the fundamental theorem of Galois theory, the subfields E of K over Q such that
[E : Q] = 2 are in bijection with the subgroups Q of G with [G : Q] = 2, i.e. subgroups of G
of order 55. Since every index 2 subgroup is normal, Q is the kernel of the natural projection
homomorphism π : G → G/Q ∼ = Z/2Z, so it suffices to count the number of surjective
homomorphisms G → Z/2Z. Any group homomorphism ϕ : G → Z/2Z is determined by
ϕ(σ), ϕ(τ ), and ϕ(σ) and ϕ(τ ) must obey the relations that σ, τ obey in G. There are
therefore two group homomorphisms G → Z/2Z, one which sends σ, τ to [0]2 (the trivial
map) and one which sends σ to [0]2 and τ to [1]2 , which is manifestly surjective. G therefore
has one subgroup of order 55, and hence K has one subfield E such that [E : Q] = 2.

You might also like