Академический Документы
Профессиональный Документы
Культура Документы
Modules
R×M −→ M
(r, m) �−→ r ∗ m =: rm ∈ M
E XAMPLES 3.1.2.
(1) Let F be a field, then an F–module is precisely a vector space over F.
(2) Let (M, +) be an abelian group. Define the map
Z×M −→ M
(r)
if r > 0,
m + m + · · · + m
(r, m) �−→ rm := (−r)
(−m) + (−m) + · · · + (−m) if r < 0,
0 if r = 0.
From the group axioms it can be asily deduced that the previous map defines
a Z–action on M . In other words, every abelian group is a Z–module. The
converse is also true, if (M, +) is a Z–module the for any r > 0 in Z one has
(r)
r = 1 + 1 + · · · + 1, and thus
(r) (r)
rm = (1 + · · · + 1)m = 1m + · · · + 1m = m + · · · + m,
and for r < 0 one has r = (−1)+ thus Z–modules are precisely abelian groups.
(3) Let R be a ring. Define an action of R on (R, +) by r ∗ s := rs. By the ring laws
this action makes R into an R–module. This action is called the (left) regular
action of R on itself. When we want to refer to R as a module over itself rather
than as a ring we will write it as R R.
21
22 3. MODULES
(5) For any R–module R M there are two distinguished submodules: the trivial sub-
module 0 := {0} ≤ M and the total submodule M ≤ M .
P ROPOSITION 3.1.6. If R M is an R–module, and A, B ≤ M are submodules of M ,
then the intersection A ∩ B is also a submodule of M . Moreover, A ∩ B is the largest (with
respect to inclusion) sumbodule of M contained in�both A and B. More generally, for any
family {Pα } of sumbodules of M the intersection Pα is also a submodule of M .
P ROOF. The proof is left as an exercise. �
is a submodule of M .
P ROOF. The proof is left as an exercise. �
The inclusions of submodules given by the last two propositions can be encoded in the
following diagram:
A �+ � B
A� B
�
A∩B
3.1.2. Cyclic modules and finitely generated modules.
D EFINITION 3.1.8. Let R M be an R–module, x ∈ M . Define Rx := {rx | r ∈ R}.
It is easy to check that Rx ≤ M is a submodule. The submodule Rx is called the cyclic
submodule of M generated by x.
R EMARK 3.1.9. Rx is the smallest submodule of M containing x.
D EFINITION 3.1.10. If R M is an R–module, and there exist some x ∈ M such that
M = Rx, we say that M is a cyclic module.
D EFINITION 3.1.11. Let R M be an R–module, and let x1 , . . . , xn ∈ M ; the set
Rx1 + · · · + Rxn ≤ M is the smallest sumbodules of M containing all the xi . This
submodule is called the submodule generated by x1 , . . . , xn . If M = Rx1 + · · · + Rxn
we say that M is finitely generated.
D EFINITION 3.1.12. Let P ≤ M be a submodule of the R–module R M . Consider
the abelian group
M
= {m = m + P | m ∈ M } ,
P
where addition is defined in the usual way as m + n := m + n. For each r ∈ R and each
m ∈ M/P , define rm := rm. This operation yields a well defined action of R on M/P ,
thus making M/P into an R–module. This module is called the quotient of M b yP .
24 3. MODULES
R EMARK 3.1.13. For the previous definition to make sense, one needs to check that
the action of R on M/P is well defined. This is the case as m = n if and only if m−n ∈ P ,
and thus for all r ∈ R one has r(m − n) ∈ P , yielding rm = rn.
E XAMPLES 3.1.14.
(1) For each R–module R M , the submodule 0 = R0 is a cyclic submodule.
(2) If V is a vector space over F, 0 �= x ∈ V , then the cyclic submodule Fx is the
1–dimensional subspace of V generated by x.
(3) Let M be a Z–module (i.e. an abelian group), x ∈ M , then Zx is the cyclic
subgroup of M generated by x.
(4) Let x ∈ R R, then the cyclic submodule Rx is precisely the principal ideal (x)
generated by x. In particular R R = R1 is a cyclic R–module.
(5) The quotient of any R–module R M by the trivial submodule 0 is M/0 = M .
(6) The quotient of any R–module R M by the total submodule M is M/M = 0.
(7) The quotient of the Z–module Z Z by the cyclic submodule Zn = (n) is the
cyclic group of order n, Zn = Z/(n).
P ROPOSITION 3.1.15. Let R M be an R–module. If P ≤ M is a submodule and
{x1 , . . . , xn } ⊆ M generate M , then {x1 , . . . , xn } ⊆ M/P generate the quotient module
M/P .
P ROOF. Since
� x1 , . . . , xn generate M , for each m ∈ M there exist r1 , . . . , rn ∈ R
such that m = ri xi , and hence
� �
m= ri x i = ri x i ,
therefore {x1 , . . . , xn } generate M/P . �
C OROLLARY 3.1.16. If M is a cyclic R–module, then for any submodule P ≤ M the
quotient M/P is also cyclic.
P ROOF. By the previous proposition, if M = Rx then M/P = Rx. �
R EMARK 3.1.17. Unlike it happens for groups, a submodule of a cyclic module
doesn’t need to be cyclic. For instance, one can take the R–module R R = R1. Its sub-
modules are the ideals of R, and in general ideals do not have to be principal, as long as
we take R to be a ring which is not a principal ideal domain.
E XAMPLE 3.1.18. Let us consider the R–module R R = R1, which is a cyclic module,
and let I � R be an ideal of R, i.e. a submodule of R R, then the quotient R/I is also a
cyclic R–module. Indeed, it is generated by 1 = 1 + I.
3.1.3. Module homomorphisms.
D EFINITION 3.1.19. Let R be a ring, R M and R N two R–modules; a map α : M →
N is said to be an R–module homomorphism (also called R–homomorphism or R–linear
map) if it satisfies the following conditions:
(1) α is an additive group homomorphism, i.e. α(0) = 0 and α(m1 + m2 ) =
α(m1 ) + α(m2 ) for all m1 , m2 ∈ M .
(2) For all r ∈ R and all m ∈ M one has α(rm) = rα(m)
E XAMPLES 3.1.20.
(1) For any two R–modules M and N , there is always a trivial module homomor-
phism from M to N , the zero homomorphism, defined by 0(m) = 0N for all
m ∈ M . This homomorphism is simply denoted by 0.
(2) For any module M , the identity map Id : M → M is a module homomorphism.
Similarly, if P ≤ M is a submodule, the inclusion map ı : P → M is also a
module homomorphism.
3.1. SOME GENERALITIES ABOUT MODULES 25
P ROOF. One has α(m) = α(n) if and only if 0 = α(m) − α(n) = α(m −
n), i.e. if and only if m − n ∈ Ker α. Thus if Ker α = 0 we have α(m) = α(n)
if and only if m = n, i.e. α is injective. Conversely, if α is injective, then if
0 = α(m) we have α(m) = α(0) and by injectivity we get m = 0, and hence
Ker α = 0. �
E XAMPLES 3.1.28.
(1) 0 : M → M the zero homomorphism, Ker 0 = M , Im 0 = 0, so the first
isomorphism theorem tells us that M/M ∼= 0.
(2) Id : M → M , the identity map, Ker Id = 0, Im Id = M , thus M/0 ∼= M.
(3) Let R M be an R–module, P ≤ M a submodule, πP : M → M/P the canonical
projection, so Ker πP = P , Im πP = M/P , and the first isomorphism theorem
tells us that M/P ∼
= M/P .
T HEOREM 3.1.29 (Classification of cyclic modules). Let R M be an R–module, then
M is cyclic if and only if M ∼
= R/I for some ideal I � R. Moreover, the ideal I is unique.
R EMARK . This result only holds for commutative rings.
P ROOF.
⇐ The module R = R1 is cyclic, so for any ideal I � R the quotient R/I is also cyclic
(cf. corollary 3.1.16).
⇐ Let M be a cyclic R–module, then there is some x ∈ M such that M = Rx. Define
α : R → Rx = M by α(r) := rx. For any a, b ∈ R R, r, s ∈ R we have
α(ra + sb) = (ra + sb)x = rax + sbx = r(ax) + s(bx) = rα(a) + sα(b),
thus α ∈ HomR (R R, Rx) is a module homomorphism. As M is cyclic, we get Im α = M ,
i.e. α is surjective. By the first isomorphism theorem for modules, we get M ∼
= R/ Ker α,
where Ker α ≤ R R is a submodule of R. But we know that submodules of R are precisely
ideals, thus Ker α = I � R, and thus M ∼ = R/I.
It remains to prove the uniqueness of the ideal I. Suppose M ∼ = R/I ∼= R/J for two
∼
ideals I, J � R. Then there is some R–module isomorphism β : R/I −→ R/J (note that
β does not need to be a ring homomorphism). As β is surjective, there must exist some
r = r + I such that β(r + I) = 1 + J ∈ R/J. For any i ∈ I, on ehas ir = ir = 0, as
ir ∈ I, thus
0 + J = 0 = β(0 + I) = β(ir) = iβ(r) = i(1 + J) = i + J,
and hence i ∈ J, so we obtain I ⊆ J. Doing the same reasoning with the inverse isomor-
phism β −1 : R/J → R/I one gets J ⊆ I, and therefore I = J. �
3.1. SOME GENERALITIES ABOUT MODULES 27
Endowed with these operations, M becomes an R–module, called the external direct sum
of the modules Mi . We will represent this as
n
�
M = M1 ⊕ · · · ⊕ M n = Mi .
i=1
�
Given an external direct sum M = Mi , we will denote by
Mi� := {(0, . . . , mi , 0, . . . , 0) | mi ∈ Mi } ,
where the only possibly nonzero coordinate is in position i. Obviously Mi� is a submodule
of M , and the maps mi ←→ (0, . . . , mi , 0, . . . , 0) provide a module isomorphism Mi ∼
=
Mi� . Henceforth, we can identify the modules Mi with the Mi� and thus think of Mi as
submodules of M .
Question: Given an R–module R M and submodules Mi ≤ M , when can we ensure that
M∼ = M1 ⊕ · · · ⊕ Mn ? What properties must the submodules Mi satisfy?
If it is the case that we can rewrite a module M as a direct sum of certain submodules
Mi ≤ M , we say that M is� an internal direct sum of the Mi ’s.
Firstly, let m ∈ M = Mi , then
m = (m1 , . . . , mn ) = (m1 , 0, . . . , 0) + · · · + (0, . . . , 0, mn ),
�
and hence we get M = M1 + · · · + Mn = Mi . Secondly, if mi ∈ Mi then mi =
(0, . . . , 0, mi , 0, . . . , 0), and m1 +· · ·+mn = 0 if and only if (m1 , . . . , mn ) = (0, . . . , 0),
i.e if and only if mi = 0 for all i. This condition leads us to the following notion:
D EFINITION 3.1.38. Let R M be an R–module, M1 , . . . , Mn ≤ M submodules of M .
We say that {Mi }ni=1 is an independent set of submodules if whenever m1 +· · ·+mn = 0
for some mi ∈ Mi , then one must have mi = 0 for all i = 1, . . . , n.
�
R EMARK 3.1.39. From what we have stated above, if M = Mi , then {Mi } is
always an independent set of submodules. This also means that “being independent” is not
an intrinsic property of the submodules, but rather of the way they are included inside M .
P ROPOSITION 3.1.40. Let R M be an R–module, Mi ≤ M for i = 1, . . . , n submod-
ules. The following are equivalent:
(1) {M1 , . . . , Mn } is an independent set of modules.
(2) Any m ∈ M1 + · · · + Mn can be written as m = m1 + · · · + mn for unique
mi ∈ Mi .
(3) For each i = 1, . . . , n one has Mi ∩ M �i = � i �= jMi .
�i = 0, where M
P ROOF.
1 ⇒ 2 If m = m1 +· · ·+mn = m�1 +· · ·+m�n m then 0 = (m1 −m�1 )+· · ·+(mn −m�n ),
where m − i − m�i ∈ Mi , thus mi − m�i = 0 and thus mi = m�i for all i = 1, . . . , mn , i.e.
the mi are unique.
2 ⇒ 3 Let m ∈ Mi ∩ M �i , then m = mi = m1 + · · · + mi−1 + mi+1 + · · · + mn , thus
m1 + · · · + mi−1 − mi + mi+1 + · · · + mn = 0, and hence m1 = · · · = mn = 0, implying
m = 0.
3⇒1 �
P ROOF.
1 ⇒ 2 Already shown above. �
1 ⇒ 1 Define α : M → Mi by m �→ (m1 , . . . , mn ), where m1 , . . . , mn are the
unique elements with mi ∈ Mi such that m = m1 + · · · + mn . Clearly, α is a surjective
module homomorphism. Let m ∈ Ker α, then α(m) = 0 = (0, . . . , 0) and thus m =
0 + 0 + · · · + 0 = 0, hence α is injective and therefore an isomorphism.
�
E XAMPLE 3.1.43. Let R M be an R–module, A, B submodules, then M = A ⊕ B if
and only if M = A + B and A ∩ B = 0. In this case, A and B are called direct summands
of M and B is called a complement of B in M .
R EMARK 3.1.44. If M = A ⊕ B, then using the second isomorphism theorem we
have
M A⊕B A+B ∼ B B ∼
= = = = = B.
A A A A∩B 0
Similarly, we obtain (A ⊕ B)/B ∼
= A.
D EFINITION 3.1.45. If M1 = M2 = · · · = Mn = M , we will represent the external
direct sum M1 ⊕ · · · ⊕ Mn = M ⊕ · · · ⊕ M simply by M n .
3.1.5. Free modules.
(n)
D EFINITION 3.1.46. Let R be a ring. A module of the form R Rn = R R ⊕ · · · ⊕ R R
is called a free module over R.
D EFINITION 3.1.47. Let {e1 , . . . , en } ⊆ M be a subset of an R–module M . We say
that the set {e1 , . . . , en } is a basis for M if for all m ∈ M there exist unique ri ∈ R such
that m = r1 e1 + · · · + rn en .
P ROPOSITION 3.1.48. Let R M be an R–module, the following are equivalent:
(1) M is a free module.
(2) M has a basis.
P ROOF.
1 ⇒ 2: Suppose M = Rn is a free module, let ei = (0, . . . , 0, 1, 0, . . . , 0), where the 1
is in the i–th position. If m ∈ M = Rn , then m = (r1 , . . . , rn ) with ri ∈ R, and hence
m = r1 e1 + · · · + rn en for unique ri ∈ R.
2 ⇒ 1: Let {e1 , . . . , en } be a basis of M , then for each m ∈ M there exist unique ri ∈ R
such that m = r1 e1 + · · · + rn en . Define the map α : M → Rn by α(m) := (r1 , . . . , rn ).
It is easy to check that this is an R–module isomorphism, showing that M ∼ = Rn . �
R EMARK 3.1.49. If {ei } is a basis of M and r ∈ R is such that rei = 0, then
0 = 0e1 + · · · + 0ei−1 + rei + 0en+1 + · · · + 0en , and hence r = 0, thus ann(ei ) = 0 and
Rei ∼
= R/(0) = R. Henceforth, for any module with a basis we have M ∼ = Re1 ⊕ · · · ⊕
(n)
∼ R ⊕ ··· ⊕ R = R .
Ren = n