MA 3403Algebraic Topology

Lecturer: Gereon Quick

Lecture 03

3. Singular chains and homology

We would like to make the idea to study a topological space X by considering

all continuous maps from test spaces into X precise. We start with defining an
important class of test spaces:

Definiton: The standard n-simplex

For n ≥ 0, the standard n-simplex ∆n is the set ∆n ⊂ Rn+1 defined by
n
X
n n+1
∆ = {(t0 , . . . ,tn ) ∈ R : ti = 1, ti ≥ 0 for all i}.
i=0
n
Another way to describe ∆ is to say that it is the convex hull of the
standard basis {e0 , . . . ,en } in Rn+1 :
( )
X X
∆n = ti ei : ti = 1, ti ≥ 0 .
i i
The ti are called barycentric coordinates.
It will be convenient to keep both these descriptions in mind.

The standard simplices are related by face maps for 0 ≤ i ≤ n which can be
described as

φni (t0 , . . . ,tn−1 ) = (t0 , . . . ,ti−1 ,0,ti , . . . ,tn−1 )

with the 0 inserted at the ith coordinate (t0 is the 0th coordinate).
1
2

Using the standard basis, φni can be described as the affine linear map (a
translation plus a linear map)
(
ej j<i
φni : ∆n−1 ,→ ∆n determined by φni (ej ) =
ej+1 j ≥ i.

A short way of expressing the above formula for φni is that it embedds ∆n−1
into ∆n by omitting the ith vertex (that is what the hat in the following formula
means):
φni = [e0 , . . . ,ei−1 ,b
ei ,ei+1 , . . . ,en ] : ∆n−1 → ∆n .

Definiton: Faces
Note that φni maps ∆n−1 onto the subsimplex opposite to the ith corner, or
in the standard basis, opposite to ei . We call the image of φni the ith face
of ∆n (which is opposite to ei ).
Note that the union of the images of all the face inclusions is the boundary
of ∆n .

The face maps satisfy a useful identity, sometimes called simplicial identity:

Lemma: A useful identity

For all 0 ≤ j < i ≤ n + 1 we have
(1) φni ◦ φn−1
j = φnj ◦ φn−1
i−1 .
 3

The first composition, φni ◦ φn−1

j , results in a 0 at the jth and i + 1st place.
n n−1
The second composition, φj ◦φi−1 , has the effect to insert a 0 at the (i−1)st
place and then one at the jth place. But since j < i, this means that, in
both cases, we have an extra 0 at the jth and at the (i + 1)st spot.
Thus both compositions yield
φni ◦ φn−1
j (t0 , . . . ,tn−2 )
=(t0 , . . . ,tj−1 ,0,tj , . . . ,ti−2 ,0,ti−1 , . . . ,tn−2 )
=φnj ◦ φn−1
i−1 (t0 , . . . ,tn−2 ).

We are going to study a topological space X by looking at all the continuous

maps from simplices into X. We give those sets of maps a name:

Definition: Singular n-simplices

Let X be any topological space. A singular n-simplex in X is a continu-
ous map σ : ∆n → X. We denote by Singn (X) the set of all n-simplices in
X. For example, Sing0 (X) is just the set of points of X. But, in general,
Singn (X) carries more interesting information for n ≥ 1.

For 0 ≤ i ≤ n, we can use the face maps φni to define maps

dni : Singn (X) → Singn−1 (X), σ 7→ σ ◦ φni

by sending an n-simplex σ to the n − 1-simplex defined by precomposition with

the ith face inclusion. The image dni (σ) = σ ◦ φni is called the ith face of σ. We
will sometimes use the notation σ (i) := σ ◦ φni for the ith face.
4

Since the collection of all face inclusions φni forms the boundary of ∆n , we can
use the maps dni to talk about the boundary of an n-simplex. The boundaries
of simplices will actually play a crucial role in the story.
We need to make this precise. First let us look at a simple example. Let X
be some space and σ : ∆1 → X be a 1-simplex in X. Assume σ(e0 ) = x0 6= x1 =
σ(e1 ). Then we would like to say that the boundary of σ is given by x0 and x1 .
Now let us assume that σ : ∆1 → X is another 1-simplex in X which forms a
closed loop, i.e., σ(e0 ) = σ(e1 ) = x ∈ X. Now we would like to say that σ has
no boundary (since it is a loop). Our face maps express σ(e0 ) = σ(e1 ) as
d10 (σ) = d11 (σ).
It would be nice if we had a short way to formulate that the boundary of σ
vanishes. For example, it would be nice if we were allowed to rewrite this equation
as
∂(σ) = d10 (σ) − d11 (σ) = 0.

But, so far, Sing0 (X) is just a set and we are not allowed to add or subtract
elements. We are now going to remedy this defect, since algebraic operations
make life much easier. Therefore, we formally extend Singn (X) into an abelian
group.
The general way to turn a set B into an abelian group, is to form the associated
free abelian group. The idea is to add the minimal amount of structure
and relations to turn B into an abelian group. Since this is an important
construction, we recall how this works:

Good to know about free abelian groups

• Any abelian group A can be seen as a Z-module with n·a := a+· · ·+a
(n summands), for n ∈ N and a ∈ A, and (−n) · a := −n · a.
Thus, abelian groups are in bijection with Z-modules. An abelian
group A is called free over a subset B ⊂ A if B is a Z-basis, i.e.,
if any element a ∈ Z can be written uniquely as a Z-linear
combination of elements in B. The cardinality of a basis is the
same for any choice of basis and is called the rank of A.
• The group Zr is free abelian with basis {e1 , . . . ,er } with ei =
(0, . . . ,i,0, . . . ,0) (the 1 in the ith position).
• Note that, for example, the group Z/2Z is not free, since it does
not admit a basis: the vector 1 ∈ Z/2Z cannot be in a basis since
2 · 1 = 0.
 5

• Given a set B, there is an associated free abelian group ZB with

basis B which is characterized by the following universal property:
any map f : B → A of sets into an arbitrary abelian group A can be
extended uniquely to a group homomorphism φ : ZB → A with
φ(b) = f (b) for all b ∈ B.
In terms of category theory, this means that the functor
AbGroups → Sets which forgets the group structure, is right ad-
joint to the functor
Sets → AbGroups, B 7→ ZB.
In other words,
HomSets (B,A) = HomAbGroups (ZB,A).
• Any subgroup of a free abelian group F is a free abelian group.

We apply this construction to the set B = Singn (X):

Definition: Singular n-chains

The group Sn (X) of singular n-chains in X is the free abelian group
generated by n-simplices
Sn (X) := ZSingn (X).
Thus an n-chain is a finite Z-linear combination of simplices
k
X
ai σi , ai ∈ Z, σi ∈ Singn (X).
i=1

Note: If n < 0, Singn (X) is defined to be empty and Sn (X) is the trivial
abelian group {0}. So whenever we talk about n-chains, n will be assumed to be
nonnegative.

Definition: Boundary operators

We define the boundary operator by
n
X n
X
∂n : Singn (X) → Sn−1 (X), ∂(σ) = (−1)i dni σ = (−1)i σ (i) .
i=0 i=0
6

We can then extend this to a homomorphism, which we also call boundary

operator, by additivity, i.e.,
m
! m
X X
∂n : Sn (X) → Sn−1 (X), ∂ aj σj := aj ∂(σj ).
j=1 j=1

Note that we will often just write ∂ instead of ∂n .

In particular, for the loop σ we considered above we are allowed to write in

S0 (X)
∂1 (σ) = d10 (σ) + (−1)d11 (σ) = d10 (σ) − d11 (σ) = 0.

A loop is an example of a particularly important class of chains. For, the

equation ∂(σ) = 0 expresses algebraically that σ has no boundary. We give
such chains a special name:

Definition: Cycles
An n-cycle in X is an n-chain c ∈ Sn (X) with ∂n c = 0. We denote the
group of n-cycles by
Zn (X) := Ker (∂n : Sn (X) → Sn−1 (X))
= {c ∈ Sn (X) : ∂n (c) = 0} ⊆ Sn (X).

Note that the group of 0-cycles is all of S0 (X), since every 0-chain is mapped
to 0:
Z0 (X) = S0 (X).

To find another example of a 1-cycle we could consider a 1-chain c = α+β +γ

where α, β, γ : ∆1 → X are singular 1-simplices such that
α(e1 ) = β(e0 ), β(e1 ) = γ(e0 ), γ(e1 ) = α(e0 ).
For then we get
∂(c) = d0 (α) − d1 (α) + d0 (β) − d1 (β) + d0 (γ) − d1 (γ)
= α(e1 ) − α(e0 ) + β(e1 ) − β(e0 ) + γ(e1 ) − γ(e0 )
= 0.

As the notation suggests, we are going to think of a chain of the form ∂(c) as
the boundary of c:
 7

Definition: Boundaries
An n-dimensional boundary in X is an n-chain c ∈ Sn (X) such that
there exists an (n + 1)-chain b with ∂n+1 b = c. We denote the group of
n-boudnaries by
Bn (X) := Im (∂n+1 : Sn (X) → Sn−1 (X))
= {c ∈ Sn (X) : there is a b ∈ Sn+1 (X) with ∂n+1 (b) = c}.

As an aside, here is another way of thinking of the algebraic process.

Signs are like orientations... just not exactly

We want to express the fact that a loop has no boundary by saying that
the signs of the boundary points cancel out. The following picture
illustrates that the something similar happens when several vertices are
involved:

In general, we can think of the signs as giving the faces of the simplices
an orientation. And if an n-simplex is a face of an (n + 1)-simplex, then it
inherits an induced orientation which is determined by how it fits into the
bigger simplex. Going down two steps of inherited signs means things
cancel out.
However, thinking of signs as orientations is formally not correct as we will
notice in an example below. But, as we will see soon, we can algebraically
remedy this defect.

As the above picture suggests, every boundary is a cycle:

Theorem: Boundaries of boundaries vanish

For every topological space X, the boundary operator satisfies ∂ ◦ ∂ = 0, or
more precisely
∂n ◦ ∂n+1 = 0 : Sn+1 (X) → Sn−1 (X).
8

Proof: It suffices to check this for an (n + 1)-simplex σ. The general case

follows, since each ∂ is a homomorphism. For σ, we just calculate:
n+1
! n+1
X X
i n+1
∂n ◦ ∂n+1 (σ) = ∂n (−1) di σ = ∂n (σ ◦ φn+1
i )
i=0 i=0
n+1
X n
X
= (−1)i (−1)j σ ◦ φn+1
i ◦ φnj
i=0 j=0
X X
= (−1)i+j σ ◦ φn+1
i ◦ φnj + (−1)i+j σ ◦ φn+1
i ◦ φnj
0≤j<i≤n+1 0≤i≤j≤n
(∗) X X 0 0
= (−1)i+j σ ◦ φn+1
j ◦ φni−1 + (−1)j +i −1 σ ◦ φn+1
j0 ◦ φni0 −1
0≤j<i≤n+1 0≤j 0 <i0 ≤n+1
= 0.
Note that at (∗) we applied identity (1) to the left hand sum and just changed
the labels of the indices as i → j 0 and j → i0 − 1. Since both sums run over the
same indices (it does not matter how we label them) and the right hand sum is
the left hand sum multiplied by (−1), both sums cancel out. QED
As an immediate consequence we get:

Corollary: Every boundary is a cycle

For every n ≥ 0, we have
Bn (X) ⊆ Zn (X).

This basic resut shows that the sequence {Sn (X), ∂n }n has an important prop-
erty:

Definition: Chain complexes

A graded abelian group is a sequence of abelian groups, indexed by the
integers. A chain complex is a graded abelian group {An }n together with
homomorphisms ∂n : An → An−1 with the property that ∂n−1 ◦ ∂n = 0.

Hence we have shown that we obtain for any topological space X a complex of
(free) abelian groups
∂ n ∂ ∂n−1
2 1∂ 0 ∂ ∂
··· →
− Sn (X) −→ Sn−1 (X) −−−→ · · · −
→ S1 (X) −
→ S0 (X) −
→ 0.
 9

It is called the singular chain complex of X. We will see next lecture what
such chain complexes are good for.