Solutions Ark2

From the book: Number 10, 11 and 12 on page 32.

Number 10 : Let A be a ring, a an ideal contained in the Jacobson radical of A;
let M be an A-module and N a finitely generated A-module, and let u : M → N be
homomorphism. If the induced homomorphism M/aM → N/aN is surjective, then u
is surjective.
Solution: Let ū denote the induced homomorphism. Then Im ū = Im (u + aN )/aN
(since ū(m̄) = n̄ means that u(m) − n ∈ a.). So if ū is surjective, Im u + a = N , and
we can apply version 2.7 of Nakayama’s lemma and conclude that Im u = N , i.e., u
is surjective. �

Number 11 : Let A be a ring. Show that if Am ≈ An , then m = n. If Am → An is

surjective, then m ≥ n. If Am → An is injective, is it always true that m ≤ n?
Solution: Let m⊆ A be a maximal ideal, and look at the induced map (A/m)m →
(A/m)n which also is an isomorphism. Now A/m is a field, and the isomorphism is
an isomorphism of finite dimensional vector spaces which consequently must have the
same dimension. Hence m = n.
If the map Am → An is surjective, it follows immediatly that (A/m)m → (A/m)n
is a surjective map between finite dimensional vector spaces, hence m ≥ n.
The last one is in fact always true, but I do not know a proof that can be given at
this stage of the course. Warning: There are several proofs on the web, but the ones
I have checked are either flawed or incomplet. �

Number 12 : Let M be a finitely generated A module and φ : M → An a surjective

homomorphism. Show that Ker φ is finitely generated.
Solution: Let e1 , . . . , en be a basis for An and let ui be elements in M with φ(ui ) = ei .
We shall see that M is the direct sum of Ker φ and the submodule N generated by
u1 , . . . , un . This will do, because then there is projection map M → Ker φ which is
surjective, and images of the generators of M under this map will generate Ker φ.
� �
Now, Ker φ ∩ N = 0 for if φ( ai ui ) = ai ei = 0, it follows that each ai = �
0 since
n
the ei ’s form � a basis for A . On the other hand if u ∈ M we may write φ(u) = ai ei .
Then u − ai ui ∈ Ker φ, so M = N + Ker φ. And the two conditions M = N + Ker φ
and Ker φ ∩ N = 0 are what we need to ensure M being the direct sum of N and Ker φ.
�

Oppgave 1. Let A be a ring and let M , M � , N and N � be four A-modules.

Solutions Ark 2 MAT4200 — autumn 2011

Show that HomA (M ⊕ M � , N ) � HomA (M, N ) ⊕ HomA (M � , N ) and that

HomA (M, N ⊕ N � ) � HomA (M, N ) ⊕ HomA (M, N � ).
Solution: Let i : M → M ⊕ M � be the homorphism i(m) = (m, 0) and i� : M � →
M ⊕ M � the one given by i� (m� ) = (0, m� ). Then we get a homorphism φ �→ (φi, φi� )
from HomA (M ⊕ M � , N ) to HomA (M, N ) ⊕ HomA (M � , N ).
It is injective since any element of M ⊕ M � is of the form (m, m� ) = i(m) + i� (m� ),
so if both φ(i) = 0 and φ(i� ) = 0, it follows that φ = 0.
On the other hand, if φ : M → N and φ� : M � → N are given, the map Φ(m, m� ) =
φ(m) + φ� (m� ) maps to the pair (φ, φ� ).
The second part of the exercise is done in a similar manner, but using the projection
maps π : M ⊕ M � → M and π � : M ⊕ M � → M � given by π(m, m� ) = m and
π � (m, m� ) = m� . They induce a map φ �→ (πφ, π � φ) which one checks is an isomorphism.
�

Oppgave 2. If a and b are ideals in the ring A and a + b = A, then (b : a) = b.

Solution: Since a + b = A there are elements a ∈ a and b ∈ b such that a + b = 1.
So if f a⊆ b, then f = f a + f b ∈ b. Hence (b : a)⊆ b. The other inclusion is obvious (it
follows from b being an ideal). �

Oppgave 3. Show that we have HomA (A/a, A/b) = (b : a)/b. Show further that
HomA (A/(x)A, A) = Ann(x) for an element x ∈ A.
Solution: An A-homorphism φ from A to any A-module is given by the element
m = φ(1) ∈ M ; indeed φ(a) = aφ(1) = am since φ is A-linear. There is no restriction
on m, any element in M gives a homorphism.
An A-homorphism φ from A/a to M is also given by m = φ(1̄), because φ(ā) =
aφ(1̄). But now there are conditions on m. Since ā = 0 whenever a ∈ a, we must have
aφ(1̄) = 0 for all a ∈ a. From the fundamental theorem on quotients, it follows that
this is the only condition. Hence HomA (A/a, M ) ≈ {m ∈ M | am = 0}.
Putting M = A/b, we we have HomA (A/a, A/b) ≈ { y ∈ A/b | ay = 0 } = { x̄ | x ∈
A and ax ∈ b } = (b : a)/b. �

Oppgave 4. If a and b are two comaximal ideals, then HomA (A/a, A/b) = 0.
Solution: This follows directly from the two previous exercises: We have (b, a) = b
and therefore HomA (A/a, A/b) = (b : a)/b = 0. �

Oppgave 5. Determine HomZ (Z/8Z, Z/2Z), HomZ (Z/2Z, Z/8Z) and HomZ (Z/4Z, Z/8Z).

—2—
Solutions Ark 2 MAT4200 — autumn 2011

Solution: We use the previous exercise:

HomZ (Z/8Z, Z/2Z) ≈ Z/2Z because any element in Z/2Z is killed by 8. Furthermo-
re, HomZ (Z/2Z, Z/8Z) ≈ 4Z/8Z, i.e., the ideal generated by 4 in Z/8Z, this because
any integer x such that 2x is divisible by 8 must be divisible by 4.
Finally, HomZ (Z/4Z, Z/8Z) � 2Z/8Z, i.e., the ideal generated by 2 in Z/8Z, again
because if 4x is divisible by 8 for an integer x, then x is divisible by 2. �

Oppgave 6. What is HomZ (Z/55Z, Z/121Z)? What about HomZ (Z/55Z, Z/565Z)?
Solution: We have 55 = 5×11 and 121 = 112 . Hence Z/55Z = Z/5Z⊕Z/11Z, We get
HomZ (Z/55Z, Z/121Z) = HomZ (Z/5Z ⊕ Z/11Z, Z/121Z) = HomZ (Z/5Z, Z/121Z) ⊕
HomZ (Z/11Z, Z/121Z) = HomZ (Z/11Z, Z/121Z) since 5 and 11 are relatively prime.
Now HomZ (Z/11Z, Z/121Z) ≈ 11Z/121Z ≈ Z/11Z since an integer x with 11x
divisible by 121 must be divisible by 11. �

Oppgave 7. If k is a field, then Homk (k n , k m ) � Mn,m (k) where Mn,m (k) stands for
the vectorspace of m × n-matrices with entries in k.
Solution: This is just linear algebra! �

Oppgave 8. Determine HomZ (Z[i], Z[i]) and HomZ[i] (Z[i], Z[i]).

Solution: In general it is true that for a ring A and an A-module M we have
HomA (A, M ) ≈ M . The isomorphism is given by sending φ to φ(1) (this is certainly a
very natural isomorphism and merits to be called “canonical”). That the homomorp-
hism φ is determined by φ(1) follows since φ being A-linear gives φ(a) = aφ(1). On
the other hand, posing φ(a) = am gives a homomorhism, for any choice of m from M .
Hence HomZ[i] (Z[i], Z[i]) ≈ Z[i].
For the other homomorphism group — HomZ (Z[i], Z[i]) — we are looking at Z-
module homomorphisms, i.e., group homomorphism (so we are ignoring the multipli-
cative structure on Z[i]). As an abelian group Z[i] is free of rank two, i.e., Z[i] ≈ Z2 .
Hence HomZ (Z[i], Z[i]) ≈ Z4 . (By oppgave 1 above and the remark at the begining of
this exercise.)
One may identify HomZ[i] (Z[i], Z[i]) with the subgroup of HomZ (Z[i], Z[i]) consisting
of those additive (i.e., Z-linear) maps which also respects the multiplication. It is
easy to see that an additive map φ is among the multiplicative ones if and only if
φ(i) = iφ(1).
One may identify HomZ (Z[i], Z[i]) with the additive group of 2 × 2 matrices with
entries in Z (which is also a ring, with multiplication corresponding to composition of

—3—
Solutions Ark 2 MAT4200 — autumn 2011

maps), i.e., with the set of matrices

� �
a b
with a, b, c and d in Z.
c d
The subgroup (or even subring) of matrices corresponding to maps in HomZ[i] (Z[i], Z[i])
are the ones of the form � �
a b
with a and b ∈ Z.
−b a
�

Oppgave 9. Let K be a field and let A ⊂ K be a local ring which is not a field. Show
that K can not be a finitely generated A-modul.
Let B ⊂ K be a ring which is not a field. Show that K is not finitely generated as
an B-modul (Hint: If m ⊂ B is a maximal ideal, let A = {a/b | b �∈ m}. Show that A
is a local ring with maximal ideal mA. This is a special case of a general construction
we shall do later on).
Solution: This is an application ofNakayama’s lemma. If A is a local ring which
is not a field, it has a non-zero maximal ideal m. And clearly mK = K since non-zero
elements in K are invertible. If K were finitely generated over A, Nakayama would
tell us that K = 0 which is not the case.
In the second part, let m be a maximal ideal in A and replace A by the localisation
Am. �

Oppgave 10. Let M be a finitely generated A-module and φ : M → M a A-

homomorphism. If φ is surjectiv, then φ is an isomorphism.(Hint: Regard M as a
module over the polynomial ring A[X] by letting X act on M as φ, i.e., Xm = φ(m)
for m ∈ M . Then use Corolary 2.5 with a = (X)A[X].)
Solution: As hinted, we regard M as a module
� over the ring
�of polynomials
� A[X], by
i i i
letting Xm := φ(m). (Hence a polynomial ai X acts as ( ai X ).m = ai φ (m).)
Certainly M is finitely generated over A[X] — generators over A are also generators
over the bigger ring.
Furthermore let a = (X)A[X]. Since φ is surjective, XM = M and thence aM = M .
By version 2.5 of Nakayama, we can find an element x killing M with x ≡ 1 mod a,
that is x = 1 + P (X)X for some polynomial P (X). But as xM = 0, it follows that
idM = −P (φ)φ, and consequently −P (φ) is an inverse map to φ. �

Oppgave 11. Use Zorn’s lemma to show that any finitely generated module has a
maximal, proper submodule. Use this to give another proof of Nakayama’s lemma.

—4—
Solutions Ark 2 MAT4200 — autumn 2011

Give an example of a module — necessarily not finitely generated — without maximal,

proper submodules.
Solution: Let {Mi }i∈I be an accending chain of proper submodules � (which are
not necessaryly finitely generated). We shall see that the union i∈I Mi is a proper
submodule. �
Indeed, assume M = i∈I Mi and let m1 , . . . , mr be generators for M . Then each
mi lies in Mρ(i) for some ∈ I, and hence in Mj for j ≥ ρ(i) as the chain is acceending.
One of the Mi ’s therefore contains all of the generators mi . This is not the case since all
the Mi ’s were supposed to be proper submodules, and the union is a proper submodule.
By Zorn’s lemma we conclude that there exists a maxiamal, proper submodule.
To derive Nakayama from this, let N be such a maximal, proper submodule. Pick
an e ∈ M but e �∈ N . Now the module < e >= Ae generated by e, contains e and hence
is not contained in N . Since N is maximal, proper, it follows that M = Ae + N . If now
aM = M , we get e = ae+n for some a ∈ a and some n ∈ N . Thus (1−a)e ∈ N , and as
a is contained in the Jacobson-radical, (1−a) is invertible, and e ∈ N . Contradiction.
�

Oppgave 12. The aim of this exercise is to investigate the behavior of a prime ideal
(p)Z when extend to the ring of Gaussian integers Z[i]. Throughout the exercise p will
be a prime.
Recall that Z[X]/(X 2 + 1) � Z[i] with X corresponding to i.
a) Show that Z[i]/(p)Z[i] � Fp [X]/(X 2 + 1) with X corresponding to i. (Hint: Both
are isomorphic to Z[X]/(p, x2 + 1).)
Solution: The map φ : Z[X] → Z[i] given by X �→ i, identifies Z[i] with the quotient
Z[X]/(x2 + 1). Now φ−1 (pZ[i]) = (p, X 2 + 1). Hence Z[i]/pZ[i] ≈ Z[X]/(p, X 2 + 1). On
the other hand Z[X]/(p, X 2 + 1) ≈ Fp [X]/(x2 + 1) as Z/pZ = Fp . �

There are three cases:

i) The case p = 2.
In this case show that X 2 + 1 = (X + 1)2 in F2 [X] and hence the equality (2) = (i + 1)2
of ideals in Z[i]. (Which one of course also can see directly).
Solution: Over F2 we have (X + 1)2 = X 2 + 2X + 1 = X 2 = 1 since 2 = 0 in F2 . The
map Z[i]/2Z[i] → F2 [X]/((X +1)2 )F2 [X] where i �→ X shows that (2)Z[i] = (i+1)2 Z[i].
(Or directly, using that (1+i)2 = 2i: (i+1)2 Z[i] = (2i)Z[i] = (2)Z[i] since i is invertible
in Z[i]. �

ii) The case when −1 is square mod p.

—5—
Solutions Ark 2 MAT4200 — autumn 2011

Then the polynomial X 2 + 1 has a root in Fp , say the residue class n̄ of an integer n.
Hence X 2 + 1 = (X − n̄)(X + n̄). Show that (p)Z[i] = (i − n, p) ∩ (i + n, p) and that
those two ideals both are prime.
Solution: In this case we have the equality (X 2 + 1)Fp [X] = (X − n̄)Fp [X] ∩ (X +
n̄)Fp [X] of ideals in Fp [X], so the (p, X 2 + 1)Z[X] = (p, X − n)Z[X] ∩ (p, X + n)Z[X]
and thus (p)Z[i] = (p, X − n)Z[i] ∩ (p, X + n)Z[i].
The ideal (p, X − n) is prime, because
Z[i]/(p, X−n) ≈ Z[X]/(p, X 2 +1, X−n) ≈ Fp [X]/(X 2 +1, X−n̄) = Fp [X]/(X−n̄) ≈ Fp
since X 2 + 1 = (X − n̄)(X + n̄) in Fp [X]. Now Fp is a field and hence (p, X − n) is a
maximal ideal. �

iii) The case −1 is not a square mod p.

In this case X 2 + 1 is irreducible in Fp [X] and (X 2 + 1) is a prime ideal. Use this to
show that (p)Z[X] is prime.
Solution: Since Z[X]/(p)Z[i] ≈ Fp [X]/(X 2 + 1), and the latter is an integral domain
since X 2 + 1 is irreducible in Fp [X], it follows that (p)Z[i] is prime. (In fact both ideals
are maximal). �

iv) Here we go further and analyse when cases ii) and iii) occure, i.e., we shall give a
criterion for a prime, which we assume different from 2, to have the property that −1
is a square mod p. Recall that F∗p denotes the mutiplicative group of non zero elements
in the finite field Fp , and that this group is cyclic of order p − 1. (Every finite subgroup
of the group of units in field is cyclic. This is a theorem).
p−1
Let σ : F∗p → F∗p be the map sending x to x2 , and let τ be the one sending x to x 2 .
Show that there is an exact sequence
σ τ
1 −−−→ {±1} −−−→ F∗p −−−→ F∗p −−−→ F∗p (1)
meaning that the kernel Ker σ = {±1} and that Ker τ = Im σ. Conclude that −1 is a
square mod p if and only if (p − 1)/2 is even, i.e., p ≡ 1 mod 4.
Solution: One has τ σ = 1 (we are working with multiplicative, abelian groups,
and the constant homomorphism x �→ 1 plays the role as the “zero” map), since
p−1
(x2 ) 2 = xp−1 = 1 — taken into account that F∗p is cyclic of order p − 1.
In a field the equation X 2 − 1 = has only ±1 a solutions, hence Ker σ = { ±1}.
Now τ (t) �= 1 since the order of F∗p is p − 1 and not p−1
2
. Clearly it takes values
p−1
in {±1}, so the kernel has order 2 , But this is exactly the order of Im σ, since
|Im σ| = |F∗p |/|Ker σ| = (p − 1)/2. Consequently, Im σ = Ker τ . �

—6—