ALGEBRA - LECTURE VI

1. Localization
Let R be a commutative ring with 1. Let S ⊆ R be a subset of R closed under multi-
plication. It will be convenient to assume that 1 is in S, but not necessary. Let M be an
R-module. The S-torsion of M is
T orS (M ) = {m ∈ M | a · m = 0 for some a ∈ S}.
Note that T orS (M ) is a submodule. (This follows since S is closed under multiplication.)
Note that M/T orS (M ) has no S-torsion. A typical example of S is R \ 0 if R is an integral
domain. In this case we write T or(M ) for T orS (M ).
Let M an R-module. In particular, any element of M can be multiplied by an element of
R. We would like to define a module where we can also divide by any element in S. This
is called a localisation of M by S. Roughly speaking, elements of S −1 M are fractions m/a
where m is in M and a in S. Then
m n
=
a b
if bm = an (note the analogy with rational numbers). Formally, S −1 M is constructed as
follows. Define a relation ∼ in M × S by (m, a) ∼ (n, b) if
x(bm − an) = 0
for some x in S. This is an equivalnce relation. If T orS (M ) = 0 then the factor x can be
omitted in the definition of the relation. The addition in S −1 M is defined by
[(m, a)] + [n, b] = [(bm + an, ab)]
which corresponds of the usual addition of fractions. The zero element is [(0, 1)], the class of
(0, 1). Note that S −1 R is a ring and S −1 M is an S −1 R-module where multiplication is given
by
[(c, b)] · [(m, a)] = [(cm, ab)].
Example: Let R be an integral domain and let S = R \ 0. Then T or(R) = 0, clealry, and
S −1 R is the field of fractions of R.
Proposition 1.1. If M 6= T orS (M ), then S −1 M 6= 0.
Proof. Let m 6= 0 be in M \ T orS (M ). Then (m, 1) is not equivalent to (0, 1), the proposition
follows. 
Corollary 1.2. If m is an element in M \ T orS (M ) then 1 ⊗ m is a non-zero element in
S −1 R ⊗R M . In particular, if M 6= T orS (M ), then S −1 R ⊗R M 6= 0.
Proof. There is a natural bilinear form on S −1 R × M with values in S −1 M :
([(r, s)], m) 7→ [(rm, s)].

1
2 ALGEBRA - LECTURE VI

2. Projective Modules
An R-module P is projective if for every diagram of R-modules

?
M - N - 0

there exists a homomorphism h : P → M such that the following diagram is commutative.

?
M - N - 0

Example: Any free module is projective!

Note that for every R-module P , every exact sequence of R-modules 0 → L → M → N → 0
gives rise to an exact sequence
0 → HomR (P, L) → HomR (P, M ) → HomR (P, N ).
Proposition 2.1. The following are equivalent:
(i) P is projective.
(ii) Every exact sequence 0 → L → M → P → 0 splits. This means that M ∼
= P ⊕ L.
(iii) For every exact sequence of R-modules 0 → L → M → N → 0, the sequence
0 → HomR (P, L) → HomR (P, M ) → HomR (P, N ) → 0
is exact. One says that the functor HomR (P, ·) is exact!
Proof. The first and the last statement are trivially equivalent as exactness of the sequence
is equivalent to surjectivity of the map from Hom(M, P ) to Hom(N, P ).
(i) implies (ii): Let f denote the map from M to P in the given exact sequence. Since P
is projective, there exists h : P → M such that f ◦ h : P → P is the identity. This shows
that h is injective and Im(h) ∩ ker(f ) = 0. Also, every m in M can be written as
m = h(f (m)) + (m − h(f (m))) ∈ Im(h) + ker(f ).
This shows that M = Im(h) ⊕ ker(f ) ∼ = I ⊕ L.
(ii) implies (i): We need the following lemma:
Lemma 2.2. Let P be an R-module. Then there exists a projective module Q such that
Q → P → 1.
 ALGEBRA - LECTURE VI 3

Proof. Let S be a set of generators of P . (You could take S to be the whole P , for example.)
Let FS be the free module generated by elements es for s in S. Then FS is projective and
f : FS → P given by
X X
f( rs es ) = rs s
surjective. 

Let L be the kernel of the projection Q → P → 0. Then (ii) imples that Q ∼

= P ⊕ L. Note
that a diagram

?
M - N - 0

can be extended to a diagram

P ⊕L

?
M - N - 0

so that L maps trivially to N . Since P ⊕L ∼

= Q is projective, there exists a map h0 : P ⊕L → M
making the diagram commutative. Now put h to be the restriction of h0 to P . This shows
that P is projective.


In the course of the proof we have also proved:

Corollary 2.3. Every projective module is a direct summand of a free module.

Corollary 2.4. The following are easy consequeces:

(i) A finitely generated Z-module is projective if and only if it is free.
(ii) Q is not projective as a Z-module.

Proof. Clearly, a direct summand of a free Z-module has no torsion. Therefore a finitely
generated projective Z-module cannot have any torsion. Thus it must be free. This shows
the first statement. The second is as easy, since Q cannot be a summand of a free Z-module.
(A free Z-module is not divisible.) 

Corollary 2.5. Let P and Q be two projective R-modules. Then P ⊗R Q is also a projective
R-module.
4 ALGEBRA - LECTURE VI

Proof. We need to show that the functor HomR (P ⊗R Q, ·) is exact. Recall from the definition
of the tensor product that, for any R-module K, we have
HomR (P ⊗ Q, K) ∼= BilR (P × Q, K)
where BilR (P × Q, K) is the set of R-bilinear maps from P × Q to K. Since, trivially,
BilR (P × Q, K) ∼ = HomR (P, HomR (Q, K))
it follows that the functor HomR (P ⊗R Q, ·) is a composition of functors HomR (Q, ·) and
HomR (P, ·). Since P and Q are projective, these two functors are exact. So is the functor
HomR (P ⊗R Q, ·). The corollary is proved. 

3. Injective modules
The theory of injective R-modules is obtained by reversing arrows and replacing the letter
P by I. An R-module I is injective if for every diagram of R-modules

I
6

M N  0

there exists a homomorphism h : M → I such that the following diagram is commutative.

I
 6

M N 0

In order to give some examples of injective modules, let us assume that R = Z. In this
case we have the following
Proposition 3.1. A Z-module I is injective if an only if it is divisible.
Proof. Let I be an injective module. We want to show that I is divisible, that is, for every
x in I and n in Z there is y in I such that x = ny. Consider the diagram

I
6

Z  Z  0
 ALGEBRA - LECTURE VI 5

where the vertical map is given by 1 7→ x and horizontal by 1 7→ n. Then there exists
h : Z → I making the diagram commutative. Now y = h(1) satisifies ny = x.
In the other direction, assume that I is divisible, and consider a diagram

I
6

M N 0

For simplicity, assume that M is generated by N and another element m. If M ∼ = N ⊕ Zm,

then m can be mapped to any element in I. Otherwise, there exists a smallest positive integer
n such that nm is in N . Let x be the image of nm in I. Let y be in I such that x = ny.
Now map m to y. In general, use Zorn’s lemma. 
The previous characterization of injective Z-modules shows that Q and Q/Z are injective
Z-modules.
Proposition 3.2. Any Z-module I is contained in an injective Z-module.
Proof. Let S be a set of generators of I and FS a free Z-module with a basis es for all s in S,
and f : FS → I as in the proof of Lemma 2.2. In particular, I ∼ = FS /K where K is the kernel
of f . Let FS0 be the free Q-module with the same basis es for all s in S. Then K ⊆ FS ⊆ FS0
and, therefore,
I∼= FS /K ⊆ FS0 /K.
Note that FS is divisible Z-module. Since a quotient of a divisible module is again divisible
module, it follows that FS0 /K is divisible and, therefore, injective. Proposition is proved. 
Note that for every R-module I, every exact sequence of R-modules 0 → N → M → L → 0
gives rise to an exact sequence
0 → HomR (L, I) → HomR (M, I) → HomR (N, I).
Proposition 3.3. The following are equivalent:
(i) I is injective.
(ii) Every exact sequence 0 → I → M → L → 0 splits. This means that M ∼
= I ⊕ L.
(iii) For every exact sequence of R-modules 0 → N → M → L → 0, the sequence
0 → HomR (L, I) → HomR (M, I) → HomR (N, I) → 0
is exact. One says that the functor HomR (·, I) is exact!
Proof. The first and the last statement are trivially equivalent as exactness of the sequence
is equivalent to surjectivity of the map from HomR (M, I) to HomR (N, I).
(i) implies (ii): Analogous to the case of projective modules.
(ii) implies (i): We need the following lemma:
Lemma 3.4. Let I be an R-module. Then there exists an injective module Q such that
0 → I → Q.
6 ALGEBRA - LECTURE VI

Proof. Conisider I as a Z-module. By Proposition 3.2, there exists an injective Z-module QZ

containing I. Note that
I∼= HomR (R, I) ⊆ HomZ (R, QZ ).
Now put Q = HomZ (R, QZ ). This is an R module, via the action (rT )(r0 ) = T (rr0 ) for every
T in Q. I claim that Q is an injective R module. To this end, note that, for every R-module
K
HomR (K, HomZ (R, QZ )) = HomZ (K, QZ ).
This shows that the functor HomR (·, Q) is exact since HomZ (·, QZ ) is so. The lemma is
proved. 
The rest of the proof is analogous to the case of projective modules. 
Remark: The lemma is the only difference between what we have done for projective modules
and what we are doing for injective modules, except for reversing all arrows and replacing
P by I. While it is very easy to see that any R-module is a quotient of a projective (even
free) module, showing that any R-module is contained in an injective module is much harder.
This is, I guess, an inherent feature of the category of R-modules.

Exercises

1) Prove that P1 and P2 are projective if and only if P1 ⊕ P2 is.

2) If 0 → K → P → M → 0 and 0 → K 0 → P 0 → M → 0 are exact sequences of R-modules
and P and P 0 are projective, show that K ⊕ P 0 is isomorphic to K 0 ⊕ P .
√ √ √ √
3) Show that Z[ 2] is not a projective Z[2 2]-module. Hint: consider Z[ 2] ⊗Z[2√2] Z[ 2].
4) Let R be a commutative ring with 1. Prove that the following two are equivalent
(i) Every R-module is injective.
(ii) Every R-module is projective.
This proof consists of two if and only if statements and the conclusion. I will NOT read
anything longer.