ELEMENTARY HOMOLOGY THEORY WITH COMPUTATIONS

ALEXANDER BURKA

Abstract. This paper aims to expose the reader unfamiliar with algebraic
topology to elementary topics in homology theory. In particular, we intend to
briefly motivate homology and introduce the basic definitions and properties,
namely invariance. We conclude with some low-dimensional computations to
reinforce the ideas presented.

Contents
1. Introduction and Motivation 1
2. Simplicial Homology 2
3. Homology as an Invariant 4
4. Some Computations 6
Acknowledgements 11
References 11

1. Introduction and Motivation

In algebraic topology, the basic aim is to construct algebraic invariants clas-
sifying topological spaces. Such invariants include the homotopy, homology, and
cohomology groups; this paper focuses on introducing homology, which we reinforce
with explicit computations for S 1 , S 2 , and RP 2 . It assumes some linear algebra,
group theory, and point-set topology.
Though we present homology in isolation, it would be remiss to ignore the his-
torical roots underpinning its more formal treatment. Very informally, homology
begins with the intuitive notion of distinguishing between spaces by examining the
presence of n−dimensional holes; since such holes persist in a space under continu-
ous deformation, they might reasonably amount to an invariant. Euler’s polyhedron
formula and Riemann’s definition for a genus are some early instances of this con-
cept at play. It would not be until the end of the nineteenth century, however, that
the theory would assume the semblances of its modern form, when Henri Poincaré
introduced homology classes in his seminal Analysis Situs.
As we venture into the more formal developments presented in the subsequent
section, bear in mind the intuition. Indeed, by rendering a more precise formulation
from this notion, we will have established an invariant whose value lies in part in
its computability (in comparison to the homotopy groups).

Date: 18 September, 2018.

1
2 ALEXANDER BURKA

2. Simplicial Homology
Our discussion begins with the standard n−simplex, a generalization of the tri-
angle to n dimensions. In order to lend a more arithmetical formulation to the
boundary of a simplex, the ordered n−simplex is introduced. From there we spec-
ify a manner by which to combine simplexes by defining a simplicial complex.
Definition 2.1. The standard n−simplex ∆n with affinely independent vertices
v0 , ..., vn (i.e. v1 − v0 , ..., vn − v0 are linearly independent) is the convex hull of the
vertices represented as vectors in Rn :
n
X
∆n = {x0 v0 + ... + xn vn ∈ Rn+1 | xi = 1 and for all i, xi ≥ 0}
i=0

Note that each ∆n is completely determined by its vertices, allowing us to represent

them as their set of vertices.
Removing all but some arbitrary choice of k ≤ n vertices leaves a k−face.
z

1 1
1 1 y y
x x
Figure 1: Graphical representation of the standard 1- and 2−simplexes
Definition 2.2. An ordered n−simplex is an n−simplex equipped with an ordering
on its vertices, denoted [v0 , ..., vn ]. This induces an orientation on the k−faces
[vi1 , ..., vik ] of the simplex.
Definition 2.3. Consider a space X and a family Σ of continuous maps from a
(ordered) simplex ∆k to X indexed by A, for which each λ determines some k:
Σ = {σλ |σλ : ∆k → X is cts. and λ ∈ A}
We call Σ a simplicial complex structure on the space X if its members satisfy
(1) The restriction of any map to the simplex’s interior is injective.
(2) The restriction of any map to any k-face is also an element of Σ, particularly
one whose index determines a map having a domain of dimension k.
(3) A set U is open in X iff each preimage σλ−1 (U ) is open.
We abuse notation by not representing simplexes as maps.
While the first two conditions are straightforward, the final condition is not as
clear. It is included because it requires the finest topology possible on the space X,
in order to rule out trivialities.
Intuitively, a simplicial complex on X is a nice decomposition of the space into
simplexes with some deformation, hence the continuous maps. It moreover lends
itself to a clear geometric interpretation: to each map we associate one simplex
and hence obtain a collection of simplexes, from which we can construct our com-
plex by gluing together their faces in a manner adherent to the aforementioned
requirements, the result of which is homeomorphic to the space (this justifies our
 ELEMENTARY HOMOLOGY THEORY WITH COMPUTATIONS 3

aforementioned abuse of notation). The disjoint union of the simplexes associated

to each map, quotiented by the face identifications, offers another representation.
Now that we have specified a manner by which to combine simplexes, we are
prepared to introduce the more technical mechanisms of chain groups, boundary
maps, and the chain complex. Chain groups capture the traversals of a series of
k−simplexes as formal linear combinations.
Definition 2.4. For a simplicial complex, the k−th chain group Ck is the free
abelian group generated by its k-simplexes. Its members are called k−chains.
Definition 2.5. The boundary operator of an ordered n−simplex [v0 , ..., vn ] is the
alternating sum of its (n − 1)−faces:
n
X
∂n [v0 , ..., vn ] = (−1)i [v0 , ..., vˆi , ..., vn ]
i=0
Each boundary homomorphism ∂k : Ck → Ck−1 is defined in the expected way:
j
X
∂k (a1 ∆k1 + ... + aj ∆kj ) = ai ∂k ∆ki
i=0

Definition 2.6. For a simplicial complex, the chain complex is a diagram consisting
of the chain groups of the complex, where successive chain groups connect via the
appropriate boundary maps; it terminates at the trivial group:
∂n+1
n ∂ 2 1 ∂n−1
0 ∂ ∂ ∂
...Cn+1 −−−→ Cn −→ Cn−1 −−−→ ... −→ C1 −→ C0 −→ 0
Defining the boundary map as an alternating sum of the (n − 1)−faces of a
simplex admits a crucial property.
Proposition 2.7. For any k−chain K, the boundary of its boundary vanishes:
(∂k−1 ◦ ∂k )(K) = 0
Hence we have that each image Im ∂k+1 is a subgroup of the corresponding kernel
Ker ∂k for all k.
Proof. Each map ∂k is a homomorphism, so it suffices to check its behavior on an
ordered k−simplex [v0 , ..., vk ]. From the definition we have
k
X
∂k [v0 , ..., vk ] = (−1)i [v0 , ..., vˆi , ..., vk ]
i=1
Let us examine the behavior of each term in the summation under the (k − 1)−th
boundary map:
∂k−1 [v0 , ..., vˆi , ..., vk ] = [vˆ0 , ..., vˆi , ..., vk ] + ... + (−1)i−1 [v0 , ..., v̂i−1 , vˆi , ..., vk ]
+(−1)i [v0 , ..., vˆi , v̂i+1 , ..., vn ] + ... + (−1)k−1 [v0 , ..., vˆi , ..., vˆn ]
But observe that for the terms in the last expansion whose index j exceeds i, the
coefficient becomes (−1)j−1 since the i−th term is removed. Hence we can write
i−1
X k
X
∂k−1 [v0 , ...vˆi , ..., vk ] = (−1)j [v0 , ...vˆj , ..., vˆi , ..., vk ]+ (−1)j−1 [v0 , ...vˆi , ...vˆj , .., vk ]
j=0 j=i+1

Then we have two instances of every [v0 , ..., vˆi , ...vˆj , ...vk ] ∈ ∂k−1 ◦ ∂k [v0 , ..., vk ]: one
from ∂k−1 [v0 , ..., vˆi , ..., vk ] with coefficient (−1)i+j , and another from ∂k−1 [v0 , ...vˆj , ..., vk ]
4 ALEXANDER BURKA

with coefficient (−1)i+j−1 . These pairs annihilate one another, and we obtain
(∂k−1 ◦ ∂k )(K) = 0 for any k−chain K.
If K is the boundary of a (k + 1)−chain L (K ∈ Im ∂k+1 and ∂k+1 (L) = K), then
∂k (K) = (∂k ◦ ∂k+1 )(L) = 0
It follows that K ∈ Ker ∂k , hence Im ∂k+1 ⊂ Ker ∂k for all k ≥ 0. Now because
each map ∂k is a homomorphism, the k−th kernel and (k + 1)−th image are both
subgroups of each k−th chain group:
Ker ∂k , Im ∂k+1 ≤ Ck for all k ≥ 0
But provided the inclusion of each (k + 1)−image in each k−kernel, we have
Im ∂k+1 ≤ Ker ∂k for all k ≥ 0

For homology, we want to count holes, not boundaries; defining it as the quotient
of k−cycles (Ker ∂k ) by k−boundaries (Im ∂k+1 )—whose cosets are equivalence
classes of k−cycles identified by their corresponding k−hole—accomplishes this.
Definition 2.8. Consider a space X along with an associated simplicial complex
and chain complex. We define the k−th (simplicial) homology group Hk∆ (X) of
X by the quotient of the k−th kernel modulo the (k + 1)−th image under the
boundary maps: .
Hk∆ (X) = Ker ∂k Im ∂ k+1

Elements of Hk∆ (X)are referred to as (simplicial) homology classes. Two cycles in

the same homology class are called homologous.
Definition 2.9. We regard a chain complex as exact if the k−th kernel and the
(k + 1)−th image coincide for all k ≥ 0. Observe that any space whose simplicial
complex yields an exact chain complex has trivial homology groups for all k ≥ 0.
In light of this definition, we can succinctly refer to homology as a measure of
the inexactness of the chain complex associated with a space.

3. Homology as an Invariant
Since several different simplicial complexes may represent the same space, how
can we guarantee that they all admit the same homology? We do not answer this
question in this paper; Prerna Nadathur’s 2007 REU paper, titled “An Introduc-
tion to Homology,” completes with a proof of this. It involves the more abstract
formulation of singular homology, which begins with the definition of a singular sim-
plex (to follow), demonstrating its equivalence to simplicial homology for a space
admitting a suitable simplicial complex.
Definition 3.1. A singular k−simplex in X is a continuous map σk : ∆k → X. In
singular homology, we consider all continuous maps from a k−simplex into X:
{σk : ∆k → X|σk is cts.}
Chain groups, k−chains, boundary maps/homomorphisms, and the singular ho-
mology groups are defined analogously to the simplicial definitions.
It remains to demonstrate the invariance of homology. We will characterize
it as a functor from the category of topological spaces Top to that of abelian
 ELEMENTARY HOMOLOGY THEORY WITH COMPUTATIONS 5

groups Ab. That functors preserve isomorphisms between categories means that a
homeomorphism amongst spaces implies an isomorphism in homology.
Definition 3.2. A category C consists of a class of objects Obj(C), a set of mor-
phisms Hom(A, B) for any A, B ∈ Obj(C), and a composition law
◦ : Hom(B, C) × Hom(A, B) → Hom(A, C) denoted (g, f ) 7→ f ◦ g. It satisfies
(1) Each morphism f ∈ Hom(A, B) of C has a unique domain A and unique
target B.
(2) For every object A, there exists an identity morphism 1A ∈ Hom(A, A)
(3) The composition law is associative.
The objects and morphisms of Top are topological spaces and continuous func-
tions; those of Ab are abelian groups and group homomorphisms.
Definition 3.3. In a category C, a morphism f ∈ Hom(A, B) is an isomorphism if
there exists a morphism f −1 ∈ Hom(B, A) such that f ◦f −1 = 1B and f −1 ◦f = 1A .
Isomorphisms in Top and Ab are homeomorphisms and group isomorphisms.
Definition 3.4. For categories C and D, a map F : C → D is a functor if
(1) If A is an object of C, then F (A) is an object of D.
(2) If f ∈ Hom(A, B), then F (f ) ∈ Hom(F (A), F (B)).
(3) If f ∈ Hom(A, B) and g ∈ Hom(B, C), then F (g ◦ f ) = F (g) ◦ F (f )
(4) For every object A ∈ C, we have F (1A ) = 1F (A) .
Proposition 3.5. Let F : C → D be a functor. If f is an isomorphism in C, then
F (f ) is an isomorphism in D.
Proof. If f ∈ Hom(A, B) is an isomorphism in C, it has an inverse f −1 ∈ Hom(B, A).
Using (3) and (4) from Definition 3.4, one can show that F (f ) and F (f −1 ) satisfy
Definition 3.3.

To characterize homology as a functor, a mechanism by which to relate the chain
groups of two spaces given a continuous map is required.
Definition 3.6. For any singular n−simplex σ ∈ X and continuous map f : X →
Y , f ◦σ is a singular n−simplex in Y by closure. To extend this to singular n−chains
we define the chain map f# : Cn (X) → Cn (Y ) as
X X
f# ( nσ σ) = nσ f ◦ σ
σ σ

We omit references to n, X, and Y in this definition to avoid excessive notation;

the next lemma justifies this.
Lemma 3.7. If f : X → Y is continuous, then ∂n ◦ f# = f# ◦ ∂n . Here, it is
implicit that the first boundary map refers to Y while the second refers to X.
Proof. Checking this for a basis element σ suffices for homomorphisms f# and ∂n :
X X
(f# ◦ ∂n )(σ) = f# ( (−1)i σ) = (−1)i f ◦ σ = ∂n (f ◦ σ) = (∂n ◦ f# )(σ)

Theorem 3.8. Homology is an invariant of a space.
6 ALEXANDER BURKA

Proof. To show Hn : Top → Ab is a functor, we define a well-defined action on

the objects and morphisms of Top while checking the functorial properties.
Obtaining the homology group Hn (X) from a space X is well-defined, since
consider all continuous simplicial maps in singular homology.
For this proof, let Zn (X) = Ker ∂n and Bn (X) = Im ∂n+1 . Consider a contin-
uous map f : X → Y . Define Hn (f ) : Hn (X) → Hn (Y ) by sending a coset in
Hn (X) to the corresponding coset in Hn (Y ) under the induced map f# . Here, zn
is a cycle in Zn (X):
Hn (f )(zn + Bn (X)) = f# (zn ) + Bn (Y )
To see that it is well-defined, first note that f# (zn ) ∈ Zn (Y ), given zn has zero
boundary. We also check that cosets are sent to corresponding cosets. If ∂n+1 (u) ∈
Bn (X), then by Lemma 3.7 we have
(f# ◦ ∂n+1 )(u) = (∂n+1 ◦ f# )(u)
Because f# (u) is a (n + 1)−chain for Y , (f# ◦ ∂n+1 )(u) ∈ Bn (Y ) follows. Recall
yn + Bn (X) = zn + Bn (X) iff yn − zn ∈ Bn (X). Then
f# (yn − zn ) = f# (yn ) − f# (zn ) ∈ Bn (Y )
Thus f# (yn ) + Bn (Y ) = f# (zn ) + Bn (Y ).
Now we check that Hn satisfies the functorial properties. First Hn (f ) should be
a group homomorphism. Take any two yn + Bn (X), zn + Bn (X) ∈ Hn (X). Then
Hn (f )[(yn + Bn (X)) + (zn + Bn (X))] = Hn (f )[(yn + zn ) + Bn (X)]
= f# (yn + zn ) + Bn (Y ) = [f# (yn ) + Bn (Y )] + [f# (zn ) + Bn (Y )]
= Hn (f )(yn + Bn (X)) + Hn (f )(zn + Bn (X))
P
We also want Hn (1X ) is the identity morphism. Let zn = nσ σ. Observe:
Hn (1X )(zn + Bn (X)) = 1X# (zn ) + Bn (X)
X
= nσ 1X ◦ σ + Bn (X) = zn + Bn (X) = 1Hn (X) (zn + Bn (X))
Hence Hn (1X ) = 1Hn (X) . Finally we check that Hn obeys composition:
X
Hn (g ◦ f )(zn + Bn (X)) = (g ◦ f )# (zn ) + Bn (Z) = nσ (g ◦ f ) ◦ σ + Bn (Z)
X X
= nσ g ◦ (f ◦ σ) + Bn (Z) = g# ( nσ f ◦ σ) + Bn (Z)
= Hn (g)(f# (zn ) + Bn (Y )) = (Hn (g) ◦ Hn (f ))(zn + Bn (X))
And so we have Hn (g ◦ f ) = Hn (g) ◦ Hn (f ). Thus Hn : Top → Ab is a functor. 

4. Some Computations
Now that we have characterized homology as an invariant, we will compute the
homology groups of some elementary spaces—namely, the sphere and real projective
space in 1 and 2 dimensions, some properties of which follow.
Recall that S n is a compact n−manifold; refer to [3] for a proof.
Real projective n−space RP n consists of all lines in Rn+1 passing through the
origin. Its topology is generated by the open cones with the cone point on the
origin. It can be represented as a quotient space with the inherited topology:
.
n+1
RP n = R \ {0} (x ∼ λx for λ ∈ R )
 ELEMENTARY HOMOLOGY THEORY WITH COMPUTATIONS 7

Similarly to S n , RP n is a compact n−manifold. ([2] shows compactness, [4]

demonstrates that it is a manifold.) Moreover, RP 1 is homeomorphic to the circle
S 1 and RP 2 is homeomorphic to the 2−sphere S 2 with its antipodal points identi-
fied, which is in turn homeomorphic to the closed 2−disk with antipodal points on
the boundary identified:
. .
2 2
RP 2 ∼
= S (x ∼ −x) ∼ = D (x ∼ −x when x ∈ ∂D2 )
Three homology computations are to follow.
Computation 4.1. To compute the homology of the circle S 1 , we begin with a
simplicial complex with proper identifications, yielding an object homeomorphic
to the circle. This simplifies the problem, since the simplicial maps for the for-
mer figure amount to nothing more than identity maps, so we do not need to deal
with them explicitly. Start with an oriented 1−simplex [x1 , x2 ], which we denote
a. Identify the vertices x1 and x2 , referring to them as x to reduce notation. The
figure below summarizes this procedure:

x x
a
Figure 2: simplicial complex for S 1
It admits the following chain complex. Here, hp1 , ..., pn i denotes the free abelian
group generated by the set {pi } and 0 denotes the trivial group:
∂2 ∂
1 0 ∂
...0 −→ hai −→ hxi −→ 0
Since the zeroth boundary homomorphism maps all 0-simplexes to 0, we immedi-
ately get Ker ∂0 = hxi. To compute Im ∂1 , we must observe its behavior on some
arbitrary 1−chain na, with n ∈ Z:
∂1 (na) = n∂1 (a) = n(x − x) = 0
Thus Im ∂1 is trivial. We now compute the zeroth homology group:
H0 = hxi /0 = hxi
Hence H0 ∼ = Z.
Since the function ∂1 maps every 1−chain to zero, it follows that the simplex a
generates Ker ∂1 , i.e. Ker ∂1 = hai. The domain of the map ∂2 is trivial, implying
the triviality of Im ∂2 in turn. Similarly to before, we can characterize the first
homology group:
H1 = hai /0 = hai
∼
Thus H1 = Z. As with Im ∂2 , each higher image and kernel amount to nothing more
than the trivial group and hence admit trivial homology groups. This completes the
computation for the homology of the circle. Moreover, because the real projective
line is homeomorphic to the circle, we have also computed its homology.
Computation 4.2. Similarly to the previous computation, we begin by construct-
ing a simplicial complex with identifications rendering it homeomorphic to the
2−sphere S 2 . Begin with two oriented 2-simplexes [x1 , y1 , z1 ] and [x2 , y2 , z2 ]. First,
attach the simplexes by identifying one of the 1−faces, denoted c, which connects
vertices x1 to z1 and x2 to z2 . To indicate the subsequent vertex identifications, we
refer to them as x and z. We will identify the edges connecting vertices x and y1,2
and denote the resulting edge a. Similarly, we will identify the edges connecting
8 ALEXANDER BURKA

vertices y1,2 and z and denote the resulting edge b. Under these identifications,
it makes sense to refer to the former vertices y1,2 as y. Finally, we orient the ac-
tual simplexes A counter-clockwise (bottom) and B clockwise (top), respectively,
to respect the orientation of the vertices. See the figure below:
y z
b

a b
c

x y
a
Figure 3: simplicial complex for S 2
Provided this, we can characterize the chain complex:
∂ 3 ∂
2 1 ∂ 0 ∂
...0 −→ hA, Bi −→ ha, b, ci −→ hx, y, zi −→ 0
Again we already have Ker ∂0 = hx, y, zi. In order to determine Im ∂1 , we must
examine the behavior of the map ∂1 on any arbitrary 1−chain la+mb+nc. Observe:
∂1 (la + mb + nc) = l∂1 (a) + m∂1 (b) + n∂1 (c)
= l(y − x) + m(z − y) + n(z − x) = (−l − n)x + (l − m)y + (m + n)z
Since any finitely-generated free abelian group with rank n is isomorphic to Zn , we
can represent this chain as a 3−tuple [l, m, n]T ∈ Z3 . Then we observe the following
behavior of the vector under ∂1 :
[l, m, n]T 7→ [−l − n, l − m, m + n]T
We can represent the linear transformation ∂1 as a matrix:
 
−1 0 −1
[∂1 ] =  1 −1 0 
0 1 1
Performing (integral) row reduction yields
 
1 0 1
0 1 1
0 0 0
Elementary techniques from linear algebra give us the basis vectors for Im [∂1 ]:
[−1, 1, 0]T , [0, −1, 1]T
They correspond to y − x and z − y, so we will compute the following quotient to
obtain the zeroth homology group:
.
H0 = hx, y, zi hy − x, z − y i
By identifying y − x and z − y, the quotient homomorphism q : Ker ∂0 → H0
behaves as follows:
q(y − x) = q(y) − q(x) = 0 =⇒ q(y) = q(x)
 ELEMENTARY HOMOLOGY THEORY WITH COMPUTATIONS 9

q(z − y) = q(z) − q(y) = 0 =⇒ q(z) = q(y)

We obtain the quotient by first swapping instances of y by x:
. .
H0 = hx, x, zi h0, z − xi = hx, zi hz − xi
Subsequently we swap instances of z with y, i.e. x:
H0 = hx, xi /0 = hxi
Thus H0 = ∼ Z.
With our matrix representation for ∂1 , we can compute the basis for Ker ∂1 to be
[−1, −1, 1]T i.e. −a−b+c by recalling that the row-reduced matrix and the original
matrix have the same kernels. We implement a similar procedure as described before
to determine Im ∂2 by considering some arbitrary 2−chain mA + nB:
∂2 (mA + nB) = m∂2 (A) + n∂2 (B)
= m(a − c + b) + n(a − c + b) = (m + n)a + (m + n)b + (−m − n)c
This admits a matrix representation:
 
1 1
[∂2 ] =  1 1
−1 −1
Then its row-reduced form is the following:
 
1 1
0 0
0 0
The basis set for of this matrix is [1, 1, −1]T , which corresponds to the chain a+b−c.
But then −a − b + c is another suitable basis for. From this information we can
compute the first homology group:
.
H1 = h−a − b + ci h−a − b + ci = 0
So H1 is trivial.
Using the row-reduced form of the matrix representation of the map ∂2 , we
compute the basis of Ker ∂2 to be [−1, 1]T , i.e. A − B. Im ∂3 is trivial, since the
map ∂3 has domain 0. We now have the second homology group:
H2 = hA − Bi /0
Thus we see H2 ∼
= Z. All of the higher homology groups are trivial, so we are done.
Computation 4.3. We begin by considering a more tractable representation of the
real projective plane than the original definition provided. Namely, we consider the
closed 2−disk with antipodal points on the boundary identified. Our construction
of a suitable simplicial complex begins by noting that the closed square is homeo-
morphic to the closed 2−disk. To begin accounting for the boundary identifications,
we identify opposite vertices on the square and denote them x and y. Beginning
with the left 1−face and moving clockwise, we denote the first edge connecting x
and y by a and the second edge connecting y and the subsequent x by b. We will
continue by identifying the subsequent to edges with a and b, respectively. Since
antipodal points are identified on the boundary, it makes sense that the orienta-
tion of the remaining faces be oriented oppositely to their identifications. But, we
must separate the square into two 2−simplexes, which we achieve by connecting
10 ALEXANDER BURKA

the opposite x vertices by an edge c and orienting it diagonally upwards. To orient

the actual 2−simplexes, which we denote by A (top) and B (bottom), we take into
account the directionality of the faces a and b. Hence both A and B are oriented
clockwise. The included figure summarizes this construction:
y x
b

a a
c

b y
x
Figure 4: simplicial complex for RP 2
This construction admits the following chain complex:
∂ 3 ∂
2 1 ∂ 0 ∂
...0 −→ hA, Bi −→ ha, b, ci −→ hx, yi −→ 0
We immediately get that Ker ∂0 = hx, yi. We now characterize Im ∂1 :
∂1 (la + mb + nc) = l∂1 (a) + m∂1 (b) + n∂1 (c)

= l(y − x) + m(x − y) + n(x − x) = (m − l)x + (l − m)y

Provided this we can represent the map ∂1 as a matrix:
 
−1 1 0
[∂1 ] =
1 −1 0
Its row-reduced form is  
1 −1 0
0 0 0
Then we compute the basis of Im [∂1 ] to be [−1, 1]T , which corresponds to the chain
y − x. Hence Im ∂1 = hy − xi. Now we can compute the zeroth homology group:
.
H0 ∼= hx, yi hy − xi = hx, xi /0 = hxi

Thus H0 ∼ = Z.
Using the row-reduced of the map [∂1 ], we can compute the basis of Ker ∂1 to be
[1, 1, 0]T and [0, 0, 1], so Ker ∂1 = ha + b, ci. Now consider some 2−chain mA + nB
and its behavior under the map ∂2 :
∂2 (mA + nB) = m∂2 (A) + n∂2 (B)

= m(b − c + a) + n(b + c + a) = (m + n)a + (m + n)b + (n − m)c

Then its matrix representation is
 
1 1
[∂2 ] =  1 1
−1 1
 ELEMENTARY HOMOLOGY THEORY WITH COMPUTATIONS 11

Under row-reduction we obtain  

−1 1
 0 2
0 0
The basis vectors for Im [∂2 ] are then [1, 1, −1] and [1, 1, 1], which means Im ∂2
is generated by a + b − c and a + b + c. We can now compute the quotient H1 ,
following a similar procedure to that in the previous computation; in particular, we
replace instances of a + b with c, since they correspond to the same thing under the
quotient map:
. . .
H1 = ha + b, ci ha + b − c, a + b + c = hc, ci h0, 2ci = hci h2ci
Hence H1 ∼ = Z2 .
With the row-reduced form, we can quickly compute that Ker ∂2 is trivial. Since
we know Im ∂3 is already trivial, we deduce that H2 is also trivial. Of course, so
are the higher homology groups; hence we have computed the homology of RP 2 .

Acknowledgements
It is my pleasure to thank my mentor, Ronno Das, in helping me to learn the
material, in directing me to a suitable topic for my paper, and in facilitating my
writing process. And, although he only served as my mentor for the first couple
of weeks, I would also like to thank Diego Bejarano-Reyes for the valuable insights
and advice he provided me early on in the REU. Finally, thank you to Peter May
for his algebraic topology lectures, his valuable comments on my paper, and, most
importantly, for organizing this excellent program.

References
[1] Allen Hatcher. Algebraic Topology. Cambridge University Press. 2002.
[2] James R. Munkres. Topology. Pearson Education, Inc. 2013.
[3] Jimmie Lawson. Differential Geometry. LSU Press. 2006.
[4] John C. Loftin. The Real Definition of a Smooth Manifold. Rutgers University Press. 2009.
[5] Joseph J. Rotman. An Introduction to Homological Algebra. Springer Media, LLC. 2009.
[6] J. P. May. A Concise Course in Algebraic Topology. University of Chicago Press. 1999.