Vector spaces over a field are a special case of the more general notion of modules over a ring. Normally, textbooks define a vector space as a set equipped with two operations which obey a long list of axioms:
An (abstract) vector space (V,F,+,⋅) consists of
A field F, whose elements are called scalars
A set V, whose elements are called (abstract) vectors
A rule (+):V×V→V for vector addition,
satisfying
a. (associativity) u+(v+w)=(u+v)+w
b. (commutativity) u+v=v+u
c. (additive identity) exists 0∈V with v+0=v for all v∈V
d. (additive inverse) for all v∈V, exists (−v)∈V with v+(−v)=0
A rule (⋅):F×V→V for scalar
multiplication, satisfying
a. (scalar identity) 1F⋅v=v for all v∈V
b. (compatibility) (αβ)v=α(β(v))
c. (distributes over vector addition) α(v+w)=αv+αw
d. (distributes over field addition) (α+β)v=αv+βv
We can state these properties more concisely by noticing that Property III is equivalent to the requirement that (V,+) forms a commutative group.
An (abstract) vector space over the field F is a commutative group (V,+) together with a rule (⋅):F×V→V satisfying
(scalar identity) 1F⋅v=v for all v∈V
(compatibility) (αβ)v=α(β(v))
(distributes over addition) α(v+w)=αv+αw
(distributes over field addition) (α+β)v=αv+βv
Definitions 1a and 1b seem to present the set V as the primary object of interest, relegating the scalars F to the sidelines. The key to understanding modules is to turn this presumption on its head by treating F as the distinguished object instead.
By partial application of the scaling operator (⋅):F×V→V, each scalar α∈F corresponds to a linear map φa:v↦αv from V to itself. Linear self-maps on V constitute the endomorphism ring (End(V),+,∘), whose operations are pointwise addition and function composition. The vector space axioms ensure that the map φ⊡:F→(V→V) from field elements to linear self-maps is a ring homomorphism. We arrive at our third and final definition,
An (abstract) vector space over the field F is a commutative group (V,+) together with a ring homomorphism φ:F→End(V).
The ring homomorphism defines the additive and multiplicative group actions on V by scalars from the field F.
For modules, we require only that the set acting on V be a ring, rather than a field.
A module over the ring R is a commutative group (M,+) together with a ring homomorphism φ:R→End(M) defining an action of R on M, where End(M) is the set of group homomorphisms M→M.
Modules over a ring R are called R-modules, for short. An R-module is called left if it arises from a left action, and right otherwise. As for vector spaces, we could unfold this definition into a list of axioms, but this would obfuscate the real purpose of modules: Many mathematical objects happen to be rings, and modules allow us to study rings by their action on a set (much like we can study groups via their representations).
Let M be an R-module. An R-submodule of M is a subgroup N⩽(M,+) closed under the ring action, rn∈N for r∈R, n∈N.
Several important examples of modules are listed below.
If F is a field, then F-modules and F-vector spaces are identical.
Every ring R is an R-module over itself. In particular, every field F is an F-vector space. Submodules of R as a field over itself are ideals.
If S is a subring of R with 1S=1R, every R-module is an S-module.
If G is a commutative group of finite order m, then m⋅g=0 for all g∈G, and G is a (Z/mZ)-module. In particular, if G has prime order p, then G is a vector space over the field (Z/pZ).
The smooth real-valued functions C∞(M) on a smooth manifold form a ring. The smooth vector fields on M form a C∞(M)-module.
For a ring R, every R-algebra has natural (left/right) R-module structure given by the (left/right) ring action of R on A.
(Z-modules) By definition, every Z-module is a commutative group. Likewise, every commutative group (G,+) becomes a Z-module under the ring action defined for n∈Z, g∈G by
n⋅g=⎩⎨⎧−a+a+⋯+a−(n times)−0−a−a−⋯−a(−n times)if n>0if n=0if n<0
We conclude that Z-modules and commutative groups are one in the same.
The polynomial ring F[x] is the space of formal linear combinations of powers of an indeterminate x, with coefficients drawn from an underlying field F.
p(x)=p0+p1x+p2x2+⋯+pdxm(m∈N)
Polynomials form a ring\footnote{the polynomial ring F[x] actually has the additional property of being an algebra, since F embeds into the center of F[x] via the ring homomorphism (α∈F)↦(α⋅1∈F[x]).} under entrywise addition and discrete convolution of coefficient sequences. The sum and product of p,q∈F[x] have coefficients
[p+q]k=pk+qk[p⋅q]k=j=0∑max(n,m)pjqk−j
Consider what it would mean for an F-vector space V to be an F[x]-module. We need a ring homomorphism φ:F[x]→End(V) describing the action of polynomials on vectors. Since φ preserves sums and products between F[x] and (End(V),+,∘) as rings\footnote{We take some notational shortcuts. For instance, ϕ(x)k is ϕ(x) composed with itself k times, and pk refers to both the element of F and to the map (v↦pkv)∈End(V).}, we find that the choice of a single linear map φ(x)∈End(V) determines the value of φ on arbitrary polynomials~p∈F[x],
φ(p)v=φ(k=1∑mpkxk)v=k=1∑mpkφ(x)kv
Similarly, any choice of ϕ(x)∈End(V) yields a valid ring homomorphism, exposing a bijection between F[x]-modules and pairs (V,T∈End(V)).
{F[x]-modules V}⟷{F-vector spaces V with alinear map T:V→V}
In general, there are many different F[x]-module structures a given F-vector space V, each corresponding to a choice of linear T:V→V.
The F[x]-submodules of an F[x]-module~V are precisely the T-invariant subspaces of V, where T∈End(V) denotes the action of x.
Each F[x]-submodule of V is closed under actions by ring elements, including T. Likewise, every T-invariant subspace is closed under ring actions, which are all polynomials in T.
An R-module homomorphism is a map ϕ:M→N between modules which respects the R-module structure, by preserving addition and commuting with the ring action on M,
ϕ(x+y)ϕ(r⋅x)=ϕ(x)+ϕ(y)=r⋅ϕ(x)∀x,y∈M∀x∈M,r∈R
The kernel of a module homomorphism is its kernel kerϕ=ϕ−1{0S} as an additive group homomorphism. A bijective R-module homomorphism is an isomorphism. For any ring R, the set HomR(M,N) of homomorphisms between two R-modules forms a commutative group under pointwise addition, (ϕ+ψ)(m)=ϕ(m)+ψ(m) for ϕ,ψ∈HomR(M,N). Moreover,
For a commutative ring R, the group HomR(M,N) forms an R-module under the ring action R→End(HomR(M,N)) given by
(r⋅ϕ)(m)≡r⋅ϕ(m)∀r∈R,m∈M,ϕ∈HomR(M,N)
Commutativity of R guarantees that (r⋅ϕ)∈HomR(M,N), since
(r⋅ϕ)(s⋅m)=r⋅ϕ(s⋅m)=rs⋅ϕ(m)=sr⋅ϕ(m)=s⋅(r⋅ϕ(m))(by definition)(ϕ is a homomorphism)(commutativity)(by definition)
Endomorphisms HomR(M,M) form a unital ring, where
(ϕ+ψ)(m)(ϕψ)(m)1HomR(M,M)=ϕ(m)+ψ(m)=(ϕ∘ψ)(m)=IdM(pointwise addition)(composition)(multiplicative identity)
We write EndR(M)=HomR(M,M) for the endomorphism ring of M.
Let M be a module over a commutative ring R. The endomorphism ring EndR(M) forms an R-algebra, under the same ring action r↦φ(φr:m↦rm) which defines M as an R-module.
This property is normally stated without reference to ring homomorphisms, but in these notes we wish to emphasize that the study of modules is really the study of \emph{ring actions}. There is at least one subtlety, though: When defining M as an R-module, we required that φ⊡:R→End(M,+) be a ring homomorphism from R to the additive group endomorphisms on (M,+). Now, we are asking whether each φr is also an R-module homomorphism.
First, the additive group homomorphism φr∈End(M,+) is also a module homomorphism, since for r,s∈R and m∈M,
φr(s⋅m)=r⋅(s⋅m)=(rs)⋅m1=s⋅(r⋅m)=s⋅φr(m)(by definition)(associativity of scalars)(associativity of scalars)(by definition)
Futher, φ⊡:R↦EndR(M) sending r↦φr is a ring homomorphism.
φr1+r2(m)φr1r2(m)=(r1+r2)⋅m=r1⋅m+r2⋅m=φr1(m)+φr2(m)=(r1r2)⋅m=r2⋅(r1⋅m)=(φr2∘φr1)(m)(by definition)(distributivity of scalars)(by definition)(by definition)(R commutative)(by definition)
Finally, each φr commutes with every element ϕ∈EndR(M),
(φr∘ϕ)(m)=φr(ϕ(m))=r⋅ϕ(m)=ϕ(r⋅m)=ϕ(φr(m))(composition)(by definition)(module homomorphism)(by definition)
By definition, every field F is a commutative ring. Therefore, the endomorphisms EndF(V) of any F-vector space form an F-algebra.
For groups and rings, recall that quotients are well-defined only for \emph{normal} subgroups and \emph{multiplication-absorbing} subrings (ideals), respectively. For modules M, it turns out that \emph{any} submodule N⩽M has a quotient M/N, and the natural projection map π:M→M/N is a ring homomorphism with kernel kerπ=N. Similarly, each F-vector subspace has a quotient F-vector space arising as the kernel of some linear map.
Let R be a ring. Let N⩽M be a submodule of the R-module M. The (additive, commutative) quotient group M/N can be made into an R-module under the ring action R→End(M/N) given by
r⋅(x+N)=(r⋅x)+N∀r∈R,x+N∈M/N
The natural projection π:M→M/N mapping x↦x+N is an R-module homomorphism with kernel kerπ=N.
(First Isomorphism Theorem) Let M,N be R-modules. The kernel of any module homomorphism ϕ:M→N is a submodule of M, and
M/kerϕ≅ϕ(M)
The vector space concepts of linear combinations, bases, and span all have analogues in R-module theory. We normally assume R is a ring with identity.
Let M be an R-module. The submodule of M generated by a subset A⊂M is the set of finite R-linear combinations
RA={r1a1+⋯+rmam∣rk∈R,ak∈A,m∈N}⩽M
A submodule N=RA⩽M is finitely generated if A⊂M is finite. A cyclic submodule N=Ra is generated by a single element a∈M.
An R-module F is free on the subset A⊂F if each nonzero x∈F expands uniquely as an R-linear combination of elements from A, in which case A is called a basis for F.
x=r1a1+⋯+rmam∃!rk∈R,ak∈A,∀x∈F
In general, more than one basis may exist. If R is commutative, every basis has the same cardinality, called the module rank of F. Unlike for vector spaces, not every module has a basis (not every module is free).
Recall that every linear map T∈HomF(V,W) between F-vector spaces is uniquely determined by its value on n=dimV points. R-linear maps between free modules enjoy the same property, which is normally stated in the following way:
(Universal Property) For any set A, there is a unique (up to isomorphism) free R-module Free(A) satisfying the following universal property: for any R-module M and any function φ:A→M, there is a unique R-module homomorphism Φ:Free(A)→M such that Φ(a)=φ(a),
\begin{figure}[h!tb]
\centering
\begin{tikzcd}
A \arrow[r, “\iota”, hook] \arrow[rd, “\varphi”’]
& \mathrm{Free}(A) \arrow[d, ”{\exists!,\Phi}”, dashed] \ & M
\end{tikzcd}
\end{figure}