A Beginner’s Guide to Homological

arXiv:2208.11199v2 [math.HO] 4 Sep 2022

Algebra: A Comprehensive Introduction

for Students
Andy Eskenazi1,2 Kevin You1
andyeske@sas.upenn.edu kevinyou@sas.upenn.edu
Will Vauclain1,3 Robin Murugadoss1
willvauc@seas.upenn.edu gmdoss@sas.upenn.edu
1
Department of Mathematics, University of Pennsylvania
2
Department of Mechanical Engineering and Applied Mechanics, University of
Pennsylvania
3
Department of Computer and Information Science, University of Pennsylvania

Abstract. Homological algebra is often understood as the translator be-

tween the world of topology and algebra. However, this branch of mathematics
is worth studying by itself, given that it provides fascinating perspectives about
other disciplines, most notably, category theory. In this paper, we seek to pro-
vide an introductory guide for advanced students of mathematics and related
specialties seeking to get started on homological algebra, covering the neces-
sary central topics to later delve deeper into more complex aspects of this field
and beyond. This work starts by presenting the notion of chain complexes
of algebraic structures, and then moves into exploring homology modules and
chain homotopies. Next, we provide an overview of projective resolutions and
conclude by entering the world of category theory by looking at Tor functors.

1 Introduction
Homological algebra is a very fascinating branch of mathematics, one that finds
its origins, generally speaking, in algebraic topology, in an attempt to construct
bridges between topological shapes and algebraic structures. Indeed, since its
inception, homological algebra has become a very valuable tool for topologists,
providing an algebraic approach for the classification of various types of shapes.
However, beyond the geometrical applications, this field is worth studying inde-
pendently, due to its wonderful insights and the mechanism it supplies to view
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

and understand complex concepts in the world of mathematics, such as category

theory1 , in a different and simpler way.
Through this work, we hope to provide the advanced undergraduate or first
year math graduate student with a complete, self-contained guide to this branch
of mathematics, assuming the knowledge of basic concepts from an introductory
abstract algebra course. And although various references are included through-
out this work to established, advanced-level resources such as Weibel’s “An
Introduction to Homological Algebra” [9], our work is written specifically with
our above-mentioned audiences in mind, serving as a navigation tool, with clear
and simple language, as well as a multitude of examples.
Indeed, using examples sourced from related fields such as group theory,
topology and category theory, this work begins by presenting the notion of chain
complexes (sequence of maps between Abelian groups or modules) in Section 2,
to then move into introducing the concept of homology modules (an algebraic
tool to compute the number of holes of a topological shape) in Section 3 and
chain homotopies (the ability to continuously deform one shape to get another
one) in Section 4. To illustrate the beauty of homological algebra in and of itself,
this work also explores projective resolutions (a type of sequence of algebraic
objects used to describe invariant properties) in Section 5 and ventures into the
world of category theory by looking at Tor functors (a key tool that simplifies
the analysis and classification of topological shapes) in Section 6.
However, before setting-off in the exploration of these various concepts, in
the remaining portion of the introduction, we will review one of the most funda-
mental theorems from group theory, and use it alongside an example that will
pave the way for the latter sections of this work.
Effectively, algebra, on its most fundamental level, is mostly concerned with
studying the structures of objects, and one of the most useful theorems to
achieve this task is the first isomorphism theorem, which we recall.
Theorem 1.1. Let ϕ : X → Y be a homomorphism (of groups G, rings R, or
R-modules). Then, X/ ker(ϕ) ∼
= ϕ(X).
Proof.
ϕ
X Y
γ
f
X/ ker ϕ
1 Category Theory is a wonderful branch of mathematics that can be understood, vaguely

speaking, as an abstraction from mathematical structures. In essence, this field deals with
mathematical objects, such as sets, groups, topologies, and the relationships between them.
In that sense, categories are composed of objects such as those listed previously, as well as
the relationships (or morphisms) between them. For instance, the category of sets contains
all sets as its objects and functions as its morphisms, while the category of groups contains
groups as its objects and group homomorphisms as its morphisms. This work will often times
point out that an algebraic structure is part of a category to make special emphasis on the
properties it might posses. The definition of a category will be provided more rigorously in
Section 2 of this work
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

We follow the commutative diagram above to give a version of the proof, where γ
is the map sending x 7→ x + ker ϕ. Suppose that ϕ : X → Y is a homomorphism
of the stated objects. Define the map f : X/ ker ϕ → ϕ(X) by x + ker ϕ 7→ ϕ(x).
First, we show that f is well defined. If x1 + ker ϕ = x2 + ker ϕ, then x1 − x2 ∈
ker ϕ, implying that ϕ(x1 ) = ϕ(x2 ), so thus f agrees on both arguments. One
can show that f is a homomorphism in any of the three categories posited
by using the fact that ϕ is a homomorphism. We see that f is surjective; if
ϕ(x) ∈ ϕ(X), then f (x+ker ϕ) = ϕ(x). Finally, f is injective. If f (x1 +ker ϕ) =
f (x2 + ker ϕ), then ϕ(x1 ) = ϕ(x2 ). This implies that ϕ(x1 − x2 ) = 0, so hence
x1 − x2 ∈ ker ϕ. Evidently, x1 + ker ϕ = x2 + ker ϕ. 
Now that we are equipped with this powerful theorem, let’s examine the
following motivating example.
Example 1.1. View Z as an Abelian group; let Z2 be the cyclic group of two
elements, {e, x}. We define ϕ : Z → Z2 by
(
e, n ≡ 0 (mod 2)
n 7→
x, n ≡ 1 (mod 2)
where we leave the verification that ϕ is a surjective group homomorphism to
the reader. Note that ker(ϕ) = 2Z; the first isomorphism theorem precisely
asserts that Z/2Z ∼
= Z2 .
An immediate way of interpreting this result is that the cosets 2Z and 2Z+1,
when endowed with the proper addition, form an Abelian group isomorphic to
Z2 . Indeed, this gives us insight into the structure of Z, in particular, that Z can
be divided into cosets, and that these cosets themselves form another Abelian
group. Observe that Z ∼ = 2Z as Abelian groups via the map n 7→ 2n. Moreover,
it is true in general that any two cosets in a quotient group are in bijection with
each other; this is not hard to prove.
This leads us to yet another interpretation. Effectively, we have used a
subgroup 2Z of Z to create a distinct Abelian group out of elements of Z, where
the elements of this new group are in bijection with Z. However, what if we want
to reverse this process? That is to say, suppose we are given Z2 ; can we find an
Abelian group B containing a subgroup that looks like Z, such that the cosets
when quotienting by this subgroup in B form an Abelian group isomorphic to
Z2 ? We say we are extending Z2 by Z. In other words, we are trying to replace
the elements of Z2 with sets that look like Z.
Example 1.2. We rewrite our original example. Here, Z bijects onto 2Z ⊆ Z
via the map ψ(n) = 2n, and Z surjects onto Z2 via ϕ as defined before:
ψ ϕ
Z → Z → Z2 .
Notice that im(ψ) = ker(ϕ) = 2Z. This is precisely what we want, for this
means that Z/ im(ψ) ∼ = ϕ(Z). Furthermore, since ψ : Z → Z (ϕ : Z → Z2 ) is
injective (surjective), Z ∼
= im(ψ) and ϕ(Z) = Z2 . We say that Z is an extension
of Z2 by Z.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

The reason why these examples are interesting to analyze is because they
illustrate the need to simultaneously use the injectivity of ψ, the surjectivity
of ϕ, and the fact that im(ψ) = ker(ϕ) to find the extension of our Abelian
group. In fact, generally speaking, the problem of finding an extension of Z2 by
Z, denoted B, is precisely the task of determining when the sequence
ψ ϕ
Z → B → Z2

satisfies the aforementioned conditions. When it does, it is called a short exact

sequence, and this is one of the key foundational concepts behind homological
algebra. We will generalize this idea in the following section.

2 Chain Complexes
The story of chain complexes, much like the story of most branches of mathe-
matics, can be understood as successive enhancements on a concept to tackle
ever greater challenges. In this case, chain complexes allow us to study a wide
variety of structures that behave similarly to these short exact sequences.
For the first part of our journey, it will be helpful to introduce the following
mental model: given a map φ : X → Y between two algebraic structures X
and Y (such as groups or modules), we can view it in two separate ways. In
one respect, the most straightforwards interpretation is that the map φ creates
an object (or element) of Y using an object (or element) of X. In this sense,
we have that im(φ) ⊆ Y represents the objects that are constructible by φ. In
another, less obvious, respect, the map φ associates each element x ∈ X with
an element y ∈ Y according to some rule (dictated by the map), so one can also
interpret φ as testing whether or not x ∈ X satisfies a certain predicate that
would cause it to map to the zero element 0 ∈ Y . Under this secondary view,
one can consider ker(φ) as representing the objects of X that pass the test of φ,
the test being whether these elements map to 0 ∈ Y .
To visualize this duality in the essence of a function concretely, consider the
two maps ψ and ϕ from the introduction section. Here, the map ψ : n 7→ 2n
constructs the even integers from the integers, since every even integer can be
expressed as 2n for some n ∈ Z. On the other hand, the map ϕ : n 7→ [n]2 tests
(or can be understood as a test to see) if an integer is divisible by 2, because an
integer is divisible by 2 if and only if [n]2 = [0]2 (where [x]2 is the equivalence
class of x modulo 2).
Philosophically, however, constructions are not always perfect. Consider, for
example, the following unsuccessful attempt to construct the multiples of three:
ψ ϕ
Z → Z → Z3

where ψ : n 7→ 6n and ϕ : n 7→ [n]3 . Given a number n ∈ im(ψ), n is in the

kernel of ϕ, because n can be written as 6z, z ∈ Z, and [6z]3 = [6]3 · [z]3 =
[0]3 · [z]3 = [0]3 . This construction is not, however, perfect, because we are
missing the odd multiples of 3, like 3, 9, etc., so im(ψ) ( ker(ϕ). We will see
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

ψ
a formal definition soon, but when this happens we say that the sequence Z →
ϕ
Z → Z3 is not exact. In this work, we will only be interested in examples where
im(ψ) ⊆ ker(ϕ), which allows for the possibility of “imperfect” constructions ψ,
but all outputs of ψ still must pass the test of ϕ.
Then, it might make sense to ask what happens if we chain multiple of
these construction-test sequences together. For example, consider the sequence
f ψ ϕ g
0 → Z → Z → Z2 → 0, where f : 0 → Z is the inclusion map from the trivial
subgroup of Z, g : Z2 → 0 is the trivial homomorphism, and ψ and ϕ are defined
as in the first example above. The first construction is f and the first test is ψ,
then ψ is also the second construction being tested by ϕ, and then finally ϕ is a
construction being tested by g. One can check (and we will verify below) that
im(f ) ⊆ ker(ψ), im(ψ) ⊆ ker(ϕ), and im(ϕ) ⊆ ker(g), so this sequence is indeed
of the form that we are interested in.
Now that we have seen sufficient motivation, we can begin to talk about the
formal definition of what it means to have a chain complex:
Definition 2.1 (Chain Complexes of Abelian Groups). A chain complex
(A• , d) of Abelian groups is a collection of groups {Ai }i∈Z together with group
homomorphisms (often called boundary morphisms) di : Ai → Ai−1 for each
i ∈ Z such that di−1 ◦ di = 0, ∀i ∈ Z.
Remark. When unambiguous, mathematicians usually refer to a chain complex
(A• , d) as A• or A and the boundary morphisms are inferred.
Notice that this formal definition does not quite match the above examples,
although we can recover what we were talking about before with the following
general fact from Group Theory:
Proposition 2.1. di−1 ◦ di = 0 ⇐⇒ im(di ) ⊆ ker(di−1 )
Proof. Let y ∈ im(di ). By definition, this means that ∃x ∈ Ai such that y =
di (x). As a result, assuming the left-hand side is true, we have that di−1 (y) =
(di−1 ◦ di )(x) = 0(x) = 0, so y ∈ ker(di−1 ). The same logic is applicable for the
backwards direction. 
Now, before moving on to ever higher levels of abstraction, it is worth taking
a moment to pause and check whether or not the construction of a chain complex
is indeed a natural extension of the concept of a short exact sequence of Abelian
groups. To do this, we will need to introduce one more definition.
Definition 2.2. We say that a chain complex (A• , d) is exact if im(di ) =
ker(di−1 ), ∀i ∈ Z
f
Proposition 2.2. Given a sequence of Abelian group homomorphisms A →
g f g
B → C, this sequence is a short exact sequence 0 → A → B → C → 0 if and
only if the following chain complex is exact:
ι f g τ
... → 0 → 0 → A → B → C → 0 → 0 → ...
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

where 0 → 0 are the identities (denoted id0 below), ι : 0 → A is the inclusion

from the trivial subgroup, and τ : C → 0 is the trivial group homomorphism.
f g
Proof. First, suppose that A → B → C is short exact. By definition, this
means that im(f ) = ker(g), f is injective, and g is surjective. Now, to check
that the above chain complex is exact, we need to verify that the other group
homomorphisms also satisfy the exactness criterion. Because im(id0 ) = 0 =
ker(id0 ), so we just need to check the following four:
?
• im(id0 ) = ker(ι)
ι maps 0 (the only element) to 0, so ker(ι) = {0} = im(id0 ).
?
• im(ι) = ker(f )
ι maps 0 (the only element) to 0, so im(ι) = {0}. f is injective, so
ker(f ) = {0} = im(ι).
?
• im(g) = ker(τ )
τ is the trivial homomorphsm so ker(τ ) = C. Because g is surjective,
im(g) = C = ker(τ ).
?
• im(τ ) = ker(id0 )
τ maps every element to 0, so im(τ ) = {0} = ker(id0 ).
Now, suppose that the given chain complex is exact. Because im(ι) = ker(f )
and im(ι) = {0}, ker(f ) = {0}, so f is injective. Because im(g) = ker(τ ) and
f g
ker(τ ) = C, im(g) = C, so g is surjective. Thus, A → B → C is short exact. 
It turns out, however, that we can make sense of chain complexes even
outside the context of Abelian groups. Recall that the modules over the ring
Z are precisely the Abelian groups, with the Z-action defined as the group
exponentiation (which in the context of Abelian groups is often written in terms
of Z-multiplication). Therefore, a reasonable conjecture would be that chain
complexes also make sense over R-modules for any ring R. This conjecture is
perhaps informed by the fact that other important results about Abelian groups,
like the classification of finitely-generated Abelian groups, also generalize to R-
modules. And, in fact, that this conjecture is correct, and we can upgrade our
definition from above as follows:
Definition 2.3 (Chain Complexes of R-modules). A chain complex (C• , d)
of modules over a ring R is a collection of R-modules {Ci }i∈Z together with R-
module homomorphisms (often called boundary morphisms) di : Ci → Ci−1
for each i ∈ Z such that di−1 ◦ di = 0, ∀i ∈ Z.
Remark. Chain complexes over R-modules are incredibly rich and interesting
to study. In fact, the other sections of this paper will focus almost entirely on
chain complexes over R modules, so outside of this section whenever you see the
term “chain complex”, you should assume that we are discussing R-modules.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

When studying chain complexes, we are often interested in how they behave
relative to other chain complexes, so we need a way to translate between chain
complexes. We can do this by way of a chain complex map:
Definition 2.4. A chain complex map u : (C• , d) → (D• , d′ ) is a collection of
R-module homomorphisms {ui : Ci → Di }i∈Z such that the following diagram
commutes for each n ∈ Z:
dn+2 dn+1 dn dn−1
... Cn+1 Cn Cn−1 ...
un+1 un un−1
d′n+2 d′n+1 d′n d′n−1
... Dn+1 Dn Dn−1 ...

Now that we have a notion of a map between chain complexes, one would
hope that these maps are well-behaved. Indeed, one can check that the iden-
tity morphism between chain complexes is the morphism which is the iden-
tity morphism at each level, there is an associative composition formed by the
element-wise compositions of each R-module homomorphism ui : Ci → Di , and
composition respects identities in the proper way.
At this point, we have defined most of the tools necessary for the study
which will be undertaken in the rest of the paper. For the remaining portion
of this section, we will explore more exotic chain complexes and conclude with
a beautiful result that ties all of these different examples together into one
all-encompassing definition.
Effectively, using our chain complex maps, it is possible to define a chain
complex over chain complexes, analogously as above to be an ordered pair
(C•,• , d), where Cn,• ∈ Ch, ∀n ∈ Z and dn−1 ◦ dn = 0, ∀n ∈ Z, and this chain
complex is exact if im(dn ) = ker(dn−1 ), ∀n ∈ Z. Here, Ch is used to denoted
the category of chain complexes, making Cn,• one of its objects.
In this context, 0 refers to the chain complex map {0 : Cn,i → 0}i∈Z , ker(dn )
refers to the chain complex formed by taking ker(dn,i ) at each position i and
restricting and co-restricting dn,i : ker(dn,i ) → ker(dn,i−1 ) for each i. This
restriction and co-restriction is well-defined because im(dn,i ) ⊆ ker(dn,i−1 ), so
each element in the domain maps to an element in the codomain. Similarly,
im(dn ) refers to the chain complex formed by taking im(dn,i+1 ) at each position
i and restricting and co-restricting dn,i : im(dn,i+1 ) → im(dn,i ), which is well-
defined because im(dn,i+1 ) ⊆ Cn,i and dn,i only maps to values in im(dn,i ).
This definition seems unmotivated right now, and in fact we do not formally
define what the kernel or the image of a chain complex is, but this is enough to
prove the following result. One can check that our notion of kernels and images
coincides with the more general notion defined later by satisfying the universal
property.
f g
Proposition 2.3. A sequence 0 → A• → B• → C• → 0 of chain complexes
fn gn
is exact in Ch(ModR ) if and only if each sequence 0 → An → Bn → Cn → 0
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

is exact in ModR . Here, Ch(ModR ) is used to denote the category of chain

complexes over R-modules.
Proof. Using the above definition of the kernel and image, the sequence of chain
complexes is exact if and only if ker(f ) = 0, im(f ) = ker(g), and ker(g) = 0,
which happens if and only if for all i ∈ Z (i.e. at all positions in the respective
chain complexes), ker(fi ) = 0, im(fi ) = ker(gi ), and ker(gi ) = 0, which happens
if and only if all of the individual sequences are exact. 
At first, it may seem surprising that we can form chain complexes of chain
complexes, and that these are well-behaved in a similar way as chain complexes
over Abelian groups or R-modules. It turns out that many other mathematical
objects support chain complexes as well. And furthermore, these different “chain
complexes” are not merely disparate structures with a common name, but they
really fundamentally behave in very similar ways.
Because of these similarities, mathematicians began searching for the un-
derlying structure that unites all of these ideas together. Like many results
unifying the study of similar structures in different contexts, the true universal
definition of a chain complex comes to us via Category Theory. Over the fol-
lowing definitions, we will build up to a characterization of the kinds of objects
that support the notion of a chain complex. To start with, we need a rigor-
ous way to talk about different “kinds” of objects and morphisms, like Abelian
groups and their homomorphisms, chain complexes over R-modules and their
chain complex maps. As was hinted before, Category Theory provides us with
such a definition. To start with:
Definition 2.5. A category C consists of the following:
• An unordered collection Ob(C) of objects.
• An unordered collection HomC (A, B) of morphisms for each pair of ob-
jects A, B. An element f ∈ HomC (A, B) is often denoted f : A → B.
• A composition operator ◦ such that for any f : A → B and g : B → C,
g ◦ f : A → C.
such that the following properties hold:
• ◦ is associative
• For each object X ∈ Ob(C) there exists a morphism idX such that for all
f : X → Y , f ◦ idX = f , and for all f : W → X, idX ◦f = f .
As a result, using the above, the proper definition of a category of chain
complexes of R-modules can be concretely written as follows:
Definition 2.6. The category of chain complexes of R-modules Ch =
Ch(ModR ) is the category whose objects are chain complexes (C• , d), whose
morphisms are chain complex maps, and whose compositions are defined as
above.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

One cam similarly define the categories Ab of Abelian groups and ModR for
modules over a ring R.
Definition 2.7 (Additive Categories). We say that a category A is additive
if the following conditions hold:
(i) Each hom-set HomA (A, B) is an Abelian group, where the binary oper-
ation + respects composition: i.e., for all morphisms e ∈ HomA (A, B),
f, g ∈ HomA (B, C), and h ∈ HomA (C, D), we have that h ◦ (f + g) ◦ e =
h ◦ f ◦ e + h ◦ g ◦ e.
(ii) A has an object 0 such that for all other objects X, there exists unique
morphisms 0 → X and X → 0, which we call the zero element.
(iii) For every pair of objects A and B in A, the product A × B is an object in
A.
An additive category gives us enough structure to define what kernels, cok-
ernels, and images are:
Definition 2.8. In an additive category A, given a morphism f : B → C, we
define the kernel of f (if it exists) to be a map i : A → B such that f ◦ i = 0,
and given any other map i′ : A′ → B such that f ◦ i′ = 0, there exists a unique
map φ : A′ → A such that i′ = i ◦ φ.

Dually, given a morphism f : B → C, we define the cokernel of f (if it exists)

to be a map e : C → D such that e◦f = 0, and given any other map e′ : C → D′
such that e′ ◦ f = 0, there exists a unique map φ : D → D′ such that e′ = φ ◦ e.

Finally, we define the image of f to be ker(coker(f )).

The above definition (which is already an elaboration of the definition given
by [9]) is still rather incomplete on its own, and deserves to be unpacked further.
First of all, when we talk about a zero morphism 0 : A → B, we really mean the
unique morphism A → 0 composed with the unique morphism 0 → B. In the
categories we are used to working with, this corresponds to the 0 homomorphism
we are familiar with.
A careful observer, however, might argue that this definition still does not
match the definition usually given for a kernel or cokernel in ModR . In ModR ,
there is a clear notion of a “subobject”, so we can understand ker(f ) as a
subobject of the domain. If we let K indicate this object, in the above definition
ker(f ) is really the R-module homomorphism K ֒→ A. For arbitrary additive
categories, it makes sense to define the kernel to be this morphism rather than
the object because we don’t always have a convenient notion of a “subobject”,
so A is not necessarily a “subobject” of B in the sense we would want it to be.
This leaves the definition of the image to be demystified. The following
proposition clarifies somewhat why this definition of the image coincides with
our usual idea of the image of a map in ModR .
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Proposition 2.4. In ModR , given a morphism f : A → B, let the image of f

be denoted by I. Then im(f ) = ker(coker(f )) : I → B.
Proof. Consider the canonical projection γ : B → B/I. Because f (x) ∈ I, ∀x ∈
A, (γ ◦ f )(x) = I, ∀x ∈ A, so it is the 0 morphism. I claim that γ is the cokernel
of f . Let g : B → C such that g ◦ f = 0. Because g ◦ f = 0, we must have
g(i) = 0, ∀i ∈ I, because by definition for each i ∈ I there exists an x ∈ A such
that f (x) = i. Thus, I ⊆ ker(g), so by the universal property of the quotient
∃!ḡ such that g = ḡ ◦ γ, so thus γ satisfies the universal property of the cokernel.
Now, consider ker(γ). I claim that ι : I ֒→ B is the kernel of γ. Let i ∈ I.
(γ ◦ ι)(i) = γ(i) = I, so γ ◦ ι = 0. Now let h : A → B such that γ ◦ h = 0.
Because γ ◦ h = 0, we must have (γ ◦ h)(x) = 0 for all x ∈ A, which means
that γ(h(x)) = γ(x) = I, so x ∈ I, ∀x ∈ A. This means that we can co-restrict
h to get a new R-module homomorphism ĥ : A → I such that h = ι ◦ ĥ, and
ĥ is unique because h and ĥ map all of the same values, just with a different
codomain. Thus, ι satisfies the universal property of the kernel.
Therefore, im(f ) = ker(coker(f )) : I → B. 
Thus, the image I that we are used to dealing with is really just the object
associated with the image as defined in a categorical sense, so the two definitions
go hand-in-hand.
An additive category gives us all of the structure that we need in order to
obtain chain complexes and check whether or not they are exact:
Definition 2.9 (Chain Complexes in an Additive Category). A chain com-
plex (A• , d) in some additive category A is a collection of objects {Ai }i∈Z where
Ai ∈ A, ∀i ∈ Z, together with morphisms (often called boundary morphisms)
di : Ai → Ai−1 for each i ∈ Z such that di−1 ◦ di = 0, ∀i ∈ Z.
Definition 2.10. A chain complex (A• , d) in some additive category A is exact
if im(di ) = ker(di−1 ), ∀i ∈ Z.
In practice, however, additive categories do not provide quite enough struc-
ture to be interesting to study, because we are not guaranteed that images and
kernels exist, so we have no way of measuring whether or not a chain complex is
exact. To remedy this, mathematicians restrict their focus to chain complexes
over Abelian categories, where Abelian categories are categories which satisfy
a few additional constraints2, one of which is that for each morphism f , ker(f )
and coker(f ) exist in A. This also guarantees that im(f ) exists, because im(f ) is
constructed out of kernels and cokernels. Some examples of Abelian categories
include Ab and ModR .
Finally, to return to the question of why Ch behaves so well, it all relies on
the following theorem:
Theorem 2.1. Ch(ModR ) is an Abelian category. Furthermore, for any Abelian
category A, the chain complexes over A form a category Ch(A), and Ch(A) is
Abelian.
2 A full definition of an Abelian category can be obtained in [9], which includes some extra

details that are not worth mentioning here.

 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Proof. We can define the morphisms between chain complexes in the same way
as we defined them above for Ch(ModR ), which one can check defines a category
with well-behaved identity morphisms and composition.
To see that Ch(A) is an additive category, observe that we can lift the
binary operation + for morphisms of A to Ch(A) by adding chain complex
maps levelwise. This means that composition distributes over addition because
it does so levelwise. Ch(A) also has a 0 object, which is given by the chain
complex . . . → 0 → 0 → . . . . The unique morphisms to and from any chain
complex A• are given by the chain maps {0 : Ai → 0}i∈Z and {0 : 0 → Ai }i∈Z .
The proof that products of pairs exist in Ch(A) and the proof that Ch(A)
is Abelian are less enlightening and require some more machinery from category
theory than has been defined thus far, so the curious reader can refer to Weibel
[9]. 
Equipped with a rigorous set of definitions, in the section that follows below,
we will explore some interesting applications of chain complexes, and how these
help encode information about topological shapes.

3 Homology
The origins of homology lie in the study of geometries, and realizing, in par-
ticular, that their topologies can be differentiated from one another and be
classified by examining their holes. This is, for instance, what makes a Boston
creme doughnut different from a regular glazed doughnut, at least in topolog-
ical terms. But to understand a shape’s holes, one must first study the holes’
boundaries, in all possible dimensions, including in 0D (points), 1D (edges),
and 2D (faces). For example, when looking at a solid 4-face pyramid in 3D, we
can define its boundary as the hollow pyramid of four 2D faces. This defini-
tion might not seem intuitive at first, although a useful analogy is to think of
wrapping the geometry with paper, and considering the wrap as the boundary.
Similarly, when looking at each of these four 2D faces that make the boundary
of the 3D pyramid, we find that each of these are in turn bounded by three 1D
edges (again, enclosing the faces); and each one of these 1D edges are bounded
by two 0D points each! This example3 illustrates two crucial points about the
study of the topologies of geometries. The first is that the dimension of the
boundary is lower than that of the studied shape itself. The second, and more
important point, is that the study of boundaries provides enough information
to create building instructions for any shape.
3 Without delving too deep into concepts of topology, this example helps demonstrate what

simplices are, generalizations of 2D triangles to multiple dimensions. In essence, a 0-simplex

is a point, a 1-simplex is a line, a 2-simplex is a triangle, and so on. The reason why studying
triangles is so important in the field of topology is due to triangulation; Hungarian mathe-
matician Tibor Rado demonstrated that essentially most 2D surfaces can be constructed, or
triangulated, using 2-simpleces. This result, for which an analogous version exists in 3D, is
extremely useful in a variety of disciplines, such as in meshing applications for finite element
analysis. More on this topic can be seen in [8], alongside additional fascinating facts such as
Euler’s formula.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Incredibly, it is precisely chain complexes that allow us to take this infor-

mation about a shape’s boundaries in all possible dimensions and construct its
building instructions. This would permit obtaining a sequence of maps such as:
ψ1 ψ2 ψ3
P oints → Edges → F aces → V olume
In the pyramid example from above, we had a sequence relating four points,
six edges, four faces and one volume. Notice that this sequence was limited to
dimensions we were familiar with (0D, 1D, 2D, 3D), but we can continue up
to an arbitrary number of dimensions, where visualizing boundaries is certainly
not trivial.
Now, the question is, how do we relate boundaries and chain complexes to
holes and homologies? Under this line of thought, following the example pro-
vided by Kelsey Houston [5], holes are defined by “not possessing boundaries”,
or in other words, arise whenever we have a fully closed hollow boundary/loop.
As an example, if we had 3 0D points but just 2 1D line segments connect-
ing these points, then adding a third 1D line segment would close the loop,
creating a hollow triangle. This gives rise to a hole, given that the boundary
of each segment (the 0D points) disappear. As a result, chain complexes, by
storing information about the boundaries of shapes in each dimension, provide
a framework from which holes can be determined, by examining closely each
boundary.
Homologies, or more precisely homology modules, provide precisely the means
through which we can use chain complexes to calculate the number of holes at
each dimension. And not surprisingly, these define the core of homological al-
gebra, the topic of this paper, which as we know, is the branch of mathematics
that acts as the translator between the world of shapes and topologies and the
more computable world of algebra.
To construct homology modules, consider the following definition:
Definition 3.1. Let (C• , d) be a chain complex, with maps dn : Cn → Cn−1 ,
as shown below:
dn+1n d dn−1 dn−2
...→Cn+1 → Cn → Cn−1 → Cn−2 → ... .
Then, define Zn (C) = ker(dn ) as the module of n-cycles of C and Bn (C) =
im(dn+1 ) as the module of n-boundaries of C. The nth homology module
of the chain complex C is then defined by the quotient of the nth -cycle and nth -
boundary of C, or more precisely Hn (C) = Zn (C)/Bn (C) = ker(dn )/ im(dn+1 )
Notice that ∀n ∈ Z, we have that 0 ⊆ Bn (C) ⊆ Zn (C) ⊆ Cn . As a result,
a morphism u : C → D between two chain complexes C and D would need
to preserve these structures, and map, ∀n ∈ Z, the nth -boundaries of C to
nth -boundaries of D, the nth -cycles of C to the nth -cycles of D, and most
importantly, would need to map the nth -homology groups of C to those of D.
One such morphism u : C → D is called a quasi-isomorphism if the maps
between homology modules, φn : Hn (C) → Hn (D), are all isomorphisms.
To illustrate this definition, consider the following proposition.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Proposition 3.1. For every chain complex (C• , d), we have that the following
are equivalent:
(i) C is exact, that is, at every Cn ;
(ii) C is acyclic, that is, Hn (C) = 0, ∀n ∈ Z;
(iii) The map u : 0 → C is a quasi-isomorphism, where 0 represents the chain
complex composed of zero modules and zero maps.
Proof. (i) =⇒ (ii) By assumption of (i), C being exact means, by definition,
that im(dn+1 ) = ker(dn ) at every n, so then we automatically get that
ker(dn )/ im(dn+1 ) = Hn (C) = 0 as well, making C acyclic.
(ii) =⇒ (iii) By assumption of (ii), we have that for the chain complex
C, Hn (C) = 0 for all the positions n. Furthermore, given that 0 is the chain
complex composed of zero modules and zero maps, then it must also follow that
Hn (0) = 0 as well (this is not surprising, since im(dn+1 ) = ker(dn ) = 0). As a
result, this means that the maps between the homology modules, φn : Hn (0) →
Hn (C) are all trivial maps, mapping 0 → 0. Thus, these are all isomorphisms,
making the map u : 0 → C a quasi-isomorphism.
(iii) =⇒ (i) By definition of quasi-isomophism, we have that all maps
φn : Hn (0) → Hn (C) must be isomorphisms. Hence, because of the structure of
the chain complex 0 and by assumption of (iii), then it is clear that the mapping
Hn (0) = 0 → Hn (C) must be trivial too in order for it to be an isomorphism
(we can’t map 0 to more than one element). Hence, Hn (C) has to necessarily
be equal to 0 for all n, proving that the chain complex C is exact. 
Now that the main definition for homology modules has been provided, below
we provide a computational example, which should clarify and illustrate how
exactly we can use them.
Example 3.1. Recall the sequence from Example 1.2, which had the maps
ψ : Z → Z, mapping n ∈ Z 7→ 2n ∈ Z, and ϕ : Z → Z/2Z, mapping z ∈ Z to its
equivalence class in Z/2Z. This resulted in:
ψ ϕ
Z → Z → Z/2Z

As a result, if we have that im(ψ) = ker(ϕ), then this means that an integer k
is divisible by 2 ⇐⇒ k = 2q, for some other integer q. In an more fundamental
and philosophical level, the integers k = 2q represent the things that can be
constructed (in this case, with the map ψ), while the integers that are divisible
by 2 are what we can test (in this case, through the map ϕ). Thus, the homology
module of this sequence, ker(ϕ)/ im(ψ), captures precisely how successful we
were in measuring what we wanted to test. If, for instance, ψ mapped n ∈ Z 7→
4n ∈ Z, then the homology module would certainly be non-zero, since we are
failing to capture certain multiples of 2 (like 2, 6, 10, ...).
Overall, this provides another way of understanding homologies that is not
strictly related to topological spaces.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

One interesting feature to mention about homology modules is that we can

extend their definition to products and direct sums of chain complexes. To see
this with more clarity, first consider the following definition:
Definition 3.2. Let (Ci • , di ) beQa family of chain complexes of R-modules,
L
where i ∈ I. Then, the product i∈I Ci and direct sum (coproduct) i∈I Ci
both exist in Ch (the category of chain complexes), and are defined degree-wise.
This means we have:
Y Y Y M M M
di,n : Ci,n → Ci,n−1 di,n : Ci,n → Ci,n−1
i∈I i∈I i∈I i∈I i∈I i∈I

where di,n represent the maps that go between Ci,n and Ci,n−1 , for a fixed i.
Hence, this allow us to introduce the following proposition, which asserts that
we can compute the homologies of products and direct sums of chain complexes.
Proposition 3.2. Products and directs sums of chain complexes commute with
homology modules. In other words, we have that:

Hn (Ci ) ∼ Hn (Ci ) ∼
Y Y M M
= Hn ( Ci ) = Hn ( Ci )
i∈I i∈I i∈I i∈I

As can be seen, the power of this proposition lies in the fact that taking the
homology module of a product or direct sum of chain complexes is the same, up
to isomorphism, than taking the homology module of each chain complex, and
then the product or direct sum.
Proof. The proof follows from simply applying the universal property of prod-
ucts and direct sums, and utilizing the definition of the homology module. 
To finalize the discussion and applications of homology module, we can intro-
duce one last concept, which is that of long exact sequences of chain complexes.
In essence, through the mechanism that will be described below, it is possible to
take a (short exact) sequence of chain complexes, compute the homology module
at each nth position of each chain, and obtain a sequence by concatenating all
of these homology modules together. More formally said, consider the following
theorem:
f g
Theorem 3.1. Let 0 → A → B → C → 0 be a short exact sequence of the
chain complexes (A• , dA ), (B• , dB ) and (C• , dC ). Then, we naturally obtain
maps ∂n : Hn (C) → Hn−1 (A) (between the homology at the nth position of C
and the (n−1)th position of A), which are known as connecting homomorphisms.
This results in the following long exact sequence of homology module:
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

f g ∂n+1
... Hn+1 (A) Hn+1 (B) Hn+1 (C)

f g ∂n
Hn (A) Hn (B) Hn (C)

f g
Hn−1 (A) Hn−1 (B) Hn−1 (C) ...

The key to proving this theorem lies in both demonstrating the existence and
defining the structure of these connecting homomorphism maps ∂n , since these
essentially permit the realization of the sequence of homology module shown
above. This task is accomplished by a famous lemma known as the “Snake
Lemma4 ”, whose statement is introduced below.
Lemma 3.1 (Snake Lemma). Consider the following commutative diagram of
the R-modules M1 , M2 , M3 and M1′ , M2′ , M3′ :
f g
M1 M2 M3 0
a b c
f′ g′
0 M1′ M2′ M3′

If we have that the two rows in the diagram are exact, then we obtain the
following exact sequence:
∂
ker(a) → ker(b) → ker(c) → coker(a) → coker(b) → coker(c)

Here, ∂ is precisely the connecting homomorphism mentioned in Theorem

3.1, and is given by ∂ = f ′−1 bg −1 . In addition, recall that coker(a) = M1′ / im(a),
coker(b) = M2′ / im(b), and coker(c) = M3′ / im(c). Now, to better understand
the origins of this connecting homomorphism map ∂ and why it is defined in
this way, it is best to look at the following diagram:
∂
ker(a) ker(b) ker(c)

f g
M1 M2 M3 0
a b c
f′ g′
0 M1′ M2′ M3′

coker(a) coker(b) coker(c)

4 The connecting homomorphism ∂ from ker(c) to coker(a), taking the shape of a snake, is

what gives the lemma its nickname, as is seen in the diagram below.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Proof. The proof for the Snake Lemma uses the concept of diagram chasing.
However, this elaboration is quite involved, so we refer to the complete proof
written by Bob Gardner [4]. 
Now, notice that when we replace M1 , M2 , M3 for the homology modules
Hn (A), Hn (B), Hn (C), and then replace M1′ , M2′ , M3′ for the homology modules
Hn−1 (A), Hn−1 (B), Hn−1 (C), this lemma automatically gives us the desired
connecting homomorphism map ∂n : Hn (C) → Hn−1 (A). Here, recall that A,
B and C were chain complexes part of the short exact sequence given by the
assumption of Theorem 3.1. Thus, one can check that the diagram in the lemma
commutes and that the rows are exact, due to this assumption.
One interesting result that follows from the application of Theorem 3.1 can
be found below:
f g
Proposition 3.3. Let 0 → A → B → C → 0 be a short exact sequence of
chain complexes (A• , dA ), (B• , dB ) and (C• , dC ). Then, if any two of these
chain complexes is exact, so is the third.
Proof. First, notice that by Theorem 3.1, we can construct a sequence of ho-
mology modules as given above that forms a long exact sequence. However,
recall that by Proposition 3.1 above, if A is a short exact sequence, then A is
acyclic, so Hn (A) = 0, ∀ n ∈ Z. Thus, if two of the chain complexes making the
short exact sequence were exact, then we would get that the long exact sequence
of homology modules would be composed of 0s in two-thirds of all positions.
Denoting this long exact sequence as L, then:
(
Hn (C), if n ≡ 0 mod 3
L=
0, otherwise

Graphically, this is illustrated as follows:

... → 0 → Hn+1 (C) → 0 → 0 → Hn (C) → 0 → 0 → Hn−1 (C) → 0 → ...

Here, we are assuming without loss of generality that the chain complex C is
not (necessarily) exact, while A and B are. Thus, in order to show the exactness
of C, then we would need Hn (C) = 0, ∀n ∈ Z. To do so, we can make use of
f
the fact that if we have a short exact sequence 0 → A → B → 0, then f is
an isomorphism. This is due to exactness at A requiring the image of the zero
map to be equal to the kernel of f , while exactness at B enforcing the image of
f to be equal to the kernel of the zero map. As a result, applying this fact to
our problem, by letting B = 0 and A = Hn (C), we get that due to f being an
isomorphism, then A must be equal to 0. Thus, Hn (C) = 0, ∀n ∈ Z, showing
that the chain complex C had to be exact to begin with.

Overall, this section allowed us to investigate and understand the importance
that homological algebra has in the study of topologies, through chain complexes
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

and homology modules. In essence, we can differentiate between topological

spaces/shapes by studying their boundaries and their respective number of holes
in all possible dimensions, whether that is in 0D, 1D, 2D or beyond. Now, in
the section that follows, we will continue building this algebraic toolbox so that
we can classify shapes and objects in topological spaces in an even more robust
way.

4 Homotopy
One other topic within the realm of topology whose study is of great interest is
that of homotopies. In order to motivate the study of homotopies from an alge-
braic perspective, consider the following example, inspired from the explanation
given by Nick Alger [1].
Suppose we have a malleable topological object A in a topological space5
B. To make this image more concrete, think of the object A as a dough of
Play-Doh, which can take many different configurations in space. Now, in order
to fully define the particular configuration of the dough of Play-Doh in space,
we would need to know the position of all the points of A in B. In other words,
we need a mapping f : A → B. Now, if we are to change the Play-Doh’s
configuration, such as by introducing a dent, we can be even more specific with
our data-keeping by defining a map h : A × I → B, where I stores all the times t
for which this deformation takes place. In particular, we can let I = [0, T ] ⊆ R,
where T is simply the time when we stop the deformation; then h(a, t), for
a ∈ A, t ∈ I, essentially describes the Play-Doh’s evolving configuration at a
particular snapshot of time t.
Now, the question that we can ask is, for two configurations of the Play-Doh
dough A in the topological space B, f : A → B and g : A → B, is it possible
to continuously transform the configuration defined by f into that defined by g,
through the time snapshots described by h(a, t)? This question is precisely the
same as asking whether there exists a homotopy h between the configurations
f and g.
In essence, we say that two continuous functions, f and g, between topolog-
ical spaces are homotopic if we can continuously transform/deform one into the
other one. Now, just as with holes in topologies, it is possible to understand
homotopies from an algebraic perspective, also using chain complexes. To do
so, consider the following constructions.
Let (C• , d) be a chain complex of vector spaces over a field F. Then, recall
that ∀n ∈ Z, we had that 0 ⊆ Bn (C) ⊆ Zn (C) ⊆ Cn , where for dn : Cn → Cn−1 ,
Zn (C) = ker(dn ) and Bn (C) = im(dn+1 ). Looking individually at each one of
the nth positions of the chain complex, we can construct the following two exact
sequences:
0 → Zn → Cn → Cn /Zn → 0
5 For the purposes of this example, a topological space B simply is the environment where

the object A lives, without going into further details nor technicalities.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

0 → dn+1 (Cn+1 ) = Bn → Zn → Zn /dn+1 (Cn+1 ) = Zn /Bn → 0

Here, we can understand Zn and Bn as vector sub-spaces of Cn and Zn
respectively. Thus, if we were to define Bn′ (Cn ) = Cn /Zn = dn (Cn ) = Bn−1 and
Hn′ (C) = Zn /Bn = Hn (C), then we can create the following two compositions
of vector spaces:
Cn ∼
M M
= Zn Cn /Zn = Zn Bn′
and
Zn ∼
M M
= Bn Zn /Bn = Bn Hn′
While these structures might seem redundant, the point of them is that they
provide a mechanism through which we can move along the chain complex. In
particular, they allow us to define an even more important map, which is the
splitting map, as can be seen below:
Definition 4.1. A splitting map sn : Cn → Cn+1 is a map such that we
obtain the composition dn = dn ◦ sn−1 ◦ dn . Now, a chain complex C is split
if we have these maps sn such that dn = dn ◦ sn−1 ◦ dn for all n. The chain
complex C is split exact if on top of the existence of the maps sn , we also have
that C is an exact sequence (or equivalently, if it is acyclic).
Notice that this definition makes sense, given that for dn : Cn → Cn−1 ,
the composition dn ◦ sn−1 ◦ dn is essentially performing the following series of
dn sn−1 dn
mappings Cn → Cn−1 → Cn → Cn−1 . In particular, dn+1 ◦ sn and sn−1 ◦ dn
are projection maps from Cn into Bn and Bn′ respectively. As a result, the direct
sum dn+1 ◦ sn + sn−1 ◦ dn is an endomorphism of Cn , and one can check that
the kernel of this map is precisely Hn′ or Hn (C). Thus, another way of checking
whether the chain complex C is exact is by looking at whether dn+1 ◦sn +sn−1 ◦dn
is the identity map on the Cn . Overall, the reason as to why the exact sequence
C of modules (or vector spaces) splits over a field F is related to the fact that
every module (or vector space) over F is free, and thus projective. This concept
will be explored in greater detail in the next section of this work.
Now that we have provided the definitions from above, it is possible to
understand homotopies from a homological algebra perspective. To do so, we
will consider two chain complexes, (C• , dC ) and (D• , dD ), which as we know
from the study of homologies, also describe topological spaces. Then, we can
introduce the maps sn : Cn → Dn+1 and maps fn : Cn → Dn between these
chain complexes, such that the following diagram results:
dC,n+1 dC,n
... Cn+1 Cn Cn−1 ...
fn+1 sn fn fn−1
sn−1
... Dn+1 dD,n+1
Dn dD,n
Dn−1 ...

Now, consider the following important definition:

 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Definition 4.2. A chain map f between the chain complexes (C• , dC ) and
(D• , dD ) is null homotopic if there exists these maps sn : Cn → Dn+1 such
that we get fn = dD,n+1 ◦ sn + sn−1 ◦ dC,n , ∀n ∈ Z, and the diagram commutes.
Then we call the collection of maps {sn } the chain contraction of f .
With this, we are ready to define what it means to have a homotopy between
topological spaces from a homological algebra perspective, or a homotopy be-
tween maps of chain complexes:
Definition 4.3. Any two chain maps f and g from the chain complex (C• , dC )
to (D• , dD ) are chain homotopic if the difference f − g is null homotopic.
That is, fn − gn = dD,n+1 ◦ sn + sn−1 ◦ dC,n . We denote the collection of maps
{sn } the chain homotopy from f to g. Finally, we say that f : C → D is a
chain homotopy equivalence (otherwise known as a homotopism of chains) if
there exists another chain complex map q : D → C such that q ◦ f and f ◦ q
are the identity maps of C and D respectively. We write C ≃ D to denote the
homotopism.
Notice that one can check that if g is any chain complex map between
(C• , dC ) and (D• , dD ), so will g + dD,n+1 ◦ sn + sn−1 ◦ dC,n , at every n. Finally,
to conclude the discussion of chain homotopies, notice that we can re-write the
diagram from above as follows by replacing the chain complex D for C:
dC,n+1 dC,n
... Cn+1 Cn Cn−1 ...
fn+1 sn fn fn−1
sn−1

... CC,n+1 dn+1

CC,n dn
Cn−1 ...

What is interesting about this diagram is that it tells us directly that a chain
complex C is split exact if the identity map on C is null homotopic. In this case,
the identity map would be given by the fn , which is dC,n+1 ◦ sn + sn−1 ◦ dC,n .
Overall, this section presented the necessary machinery to understand homo-
topies of two functions between topological spaces from an algebraic perspective,
through the use of chain complexes and splitting maps. In the following two sec-
tions, the focus will shift towards the mathematics behind the intrinsic beauty
of homological algebra.

5 Projective Resolutions
One very useful concept in homological algebra is that of resolutions. In essence,
these are one specific type of the exact sequences encountered earlier, only that
their elements are objects from an Abelian category. The importance of resolu-
tions lies in the idea that they can be used to describe the structure of the objects
from the Abelian category in question, in particular through the invariants na-
tive to these objects. In general terms, invariants denote those mathematical
properties of objects that don’t see a modification when applying an operation
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

or transformation. For instance, the magnitude of a complex number remains

unchanged regardless of whether we take the conjugate or not.
To formalize our background on resolutions and to illustrate their use in
describing structures, throughout this section, we focus our attention on the
category of unital R-modules (modules that admit the multiplicative identity).
And to start our discussion, we first recall the definition of free R-modules.
Definition
L 5.1. Let R be a ring; let I be a set. The free R-module on I is
i∈I R, where addition and R-action are defined component wise. We denote
this module as R[I]. If I is empty, we set R[I] = {0}.
Embed I in R[I] with the map ι : I → R[I] sending

i 7→ (xj )j∈I

where xj = 0 for j 6= i, and xj = 1 when j = i. This is verifiably an injective

map. One can then see that each element in R[I], (ri )i∈I , can be written
P
i∈I ri ι(i), a well defined sum since elements of direct sums are all but finitely
zero. It is immediate from this perspective that the ι(i) are generators for R[I].
Now, it is possible to define an important quality of free R-modules, as given
by the theorem below:
Theorem 5.1. (Universal property of free modules). For any R-module M and
map f : I → M , there exists a unique homomorphism f˜ : R[I] → M such that
f˜ ◦ ι = f .
Proof. Suppose f is a map from I to M . Define f˜ : R[I] → M by
X X
ri ι(i) 7→ ri f (i)
i∈I i∈I

One sees that f˜ ◦ ι = f ; we leave it to the reader to verify that this is a

homomorphism of R-modules, and that any other such homomorphism must be
equivalent. 
This theorem implies that one can define maps from a free module to any
arbitrary R-module M quite easily; simply assign its generators to any elements
of M , for suppose that M is an arbitrary R-module, and each ι(i) is assigned to
mi ∈ M . Then, we have a set map f : I → M defined by i 7→ mi , and the uni-
versal property of free modules (Theorem 5.1) gives us a unique homomorphism
f˜ : R[I] → M . This is a useful quality indeed.
Definition 5.2. An R-module P is said to be projective if for any surjective
homomorphism ϕ : M → N between R-modules M and N , and any homomor-
phism f : P → N , there exists a lifting map f˜ : P → M such that ϕ ◦ f˜ = f ,
i.e.,
M
f˜
ϕ
f
P N
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Having defined the notion of module projectiveness, we can introduce the

proposition below, which highlights why studying free R-modules is of impor-
tance, especially in the context of resolutions:
Proposition 5.1. Every free module is projective.
Proof. Let R[I] be the free module on a set I, let ϕ : M → N be an epi-
morphism6 , and let f : R[I] → N be a homomorphism. For any i ∈ I, we
have f (ι(i)) ∈ N . Since ϕ is surjective, there exists an mi ∈ M such that
ϕ(mi ) = f (ι(i)). Define g : I → M by
i 7→ mi
By the universal property of free modules, there exists a unique homomorphism
f˜ : R[I] → M suchPthat f˜◦ ι = g. It is clear from construction that ϕ ◦ g = f ◦ ι.
Moreover, for any i∈I ri ι(i) ∈ R[I],
X  X 
ϕ(f˜ ri ι(i)) = ϕ ri f˜(ι(i))
i∈I i∈I
X 
=ϕ ri g(i)
i∈I
X
= ri ϕ(g(i))
i∈I
X
= ri f (ι(i))
i∈I
X 
=f ri ι(i)
i∈I

Thus, R[I] is projective. 

Indeed, all free modules are projective, but notice the converse is not neces-
sarily true. The following is an example of a non-free projective module.
Example 5.1. Let R = Z/6Z. Then, Z/3Z is a Z/6Z module as 3 | 6; it is not
free, but it is projective. To prove this, we use the fact that Z/6Z ∼
= Z/2Z ⊕
Z/3Z. Let ι : Z/3Z → Z/2Z ⊕ Z/3Z be the embedding into the direct sum, and
let π : Z/2Z ⊕ Z/3Z → Z/3Z be the canonical projection. If ϕ : M → N is a
surjective homomorphism and f : Z/3Z → N a homomorphism, then we have
the following diagram.
M
ϕ
π f
Z/2Z ⊕ Z/3Z Z/3Z N
ι
6 Epimorphisms can be understood as the category theory abstraction of surjective maps.

Formally written, from a categorical perspective, an epimorphism ϕ is a morphism between

objects A and B of category X such that for ∀C ∈ X and ψ1 , ψ2 from B to C, we have that
ψ1 ◦ ϕ = ψ1 ◦ ϕ =⇒ ψ1 = ψ2 .
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Since Z/6Z is free as a Z/6Z-module, Proposition 5.1 implies that it is projective

as well. Therefore, there exists a lifting map g : Z/2Z ⊕ Z/3Z → M such that
ϕ ◦ g = f ◦ π. Define the homomorphism f˜ : Z/3Z → M by f˜ = g ◦ ι. Then,
ϕ ◦ f˜ = ϕ ◦ g ◦ ι
=f ◦π◦ι
=f
Consequently, Z/3Z is projective as a Z/6Z-module.
The proof of projectiveness in the above example hints at a more general
relationship. Indeed, all direct summands of a free module are in fact projective,
and the converse is true as well. However, some theory is needed in order to
establish this correspondence. We proceed by first recalling Definition 4.1 in the
context of short exact sequences, to then introduce two powerful propositions.
Definition 5.3. A short exact sequence
f g
0→A→B→C→0
is said to be split exact if there exists an h : C → B such that gh = idC .
Proposition 5.2. If the above sequence is split exact, then B ∼ = A ⊕ C.
Proof. Suppose that the sequence is split exact, so that there exists such an
h : C → B. Define ϕ : A ⊕ C → B by
(a, c) 7→ f (a) + h(c),
a verifiable homomorphism. We see that ϕ is injective, for suppose that ϕ(a, c) =
0. Then, f (a) + h(c) = 0, implying that g(f (a)) + g(h(c)) = 0. But, g(h(c)) = c,
and g(f (a)) = 0; thus, c = 0. This means that f (a) = 0, but since f is injective,
a = 0. To prove surjectivity, let b ∈ B. Then, b = (b − h(g(b))) + h(g(b)). It
follows that g(b − h(g(b))) = g(b) − g(b) = 0, that is, (b − h(g(b))) ∈ ker(g) =
im(f ). Hence, there exists an a0 ∈ A such that f (a0 ) = (b − h(g(b))). One can
then see that ϕ(a0 , g(b)) = b, as desired. 
Proposition 5.3. Every R-module is the homomorphic image of a free module.
Hence, every R-module is the homomorphic image of a projective module.
Proof. Let A be the set of generators for M , an arbitrary R-module. A exists,
as we could let A = M . Consider the free module on A, R[A]; the universal
property of free modules induces a map f˜ : R[A] → M , since we have an
embedding idA : A → M as well as ι : A → R[A]. It is necessarily true that
f˜ ◦ ι = idA , so A ⊆ im(f˜). Since A generates M , we can write for any arbitrary
m ∈ M,
m = r1 a1 + ...rn an
= r1 f˜(x1 ) + ... + rn f˜(xn )
= f˜(r1 x1 + ... + rn xn )

 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Having established these propositions, it is now possible to construct the fol-

lowing theorem, which greatly facilitates the task of finding projective modules.
Theorem 5.2. The following conditions are equivalent for an R-module P .
(i) P is projective.
(ii) Every exact sequence
f g
0→A→B→P →0
is split exact.
(iii) P is a direct summand of a free module.
Proof. (i) =⇒ (ii) The map g : B → P is an epimorphism (surjective ho-
momorphism), and we have the identity map P → P . Thus, there exists an
h : P → B such that g ◦ h = idP .
(ii) =⇒ (iii) Let A be a set of generators for P . We have by Proposition
5.3 the exact sequence
id g
0 → ker(g) → R[A] → P → 0

implying by Proposition 5.2 that P ∼ = ker(g) ⊕ R[A].

(iii) =⇒ (i) This proof is structurally identical to the one given in Example
5.1. 
To illustrate the power of this theorem in determining whether a module is
projective or not, consider the following two examples.
Example 5.2. Z/2Z is a non-free projective Z/6Z-module, as it is a direct
summand of a free Z/6Z-module, Z/2Z ⊕ Z/3Z.
Example 5.3. Let R = Z/2Z ⊕ Z/2Z, where multiplication is defined compo-
nent wise. Let Z/2Z be an R-module where the R-action is defined

(r1 , r2 ) · x = r1 x

Then Z/2Z is a non-free projective R-module.

Now that we are acquainted with projective modules, we are finally ready
to introduce the formal framework behind resolutions, which as we mentioned
at the beginning of this section, are useful in defining the structure of objects
from Abelian categories, such as the category of unital R-modules. In this
particular case, given an R-module M , we are interested in examining how
closely it resembles a free module. By Proposition 5.3, we know that there
exists a free module P0 and epimorphism ϕ0 : P0 → M , and Theorem 1.1 (First
Isomorphism Theorem) asserts that P0 / ker(ϕ0 ) ∼ = M . If ker(ϕ0 ) = 0, then we
are satisfied; M is itself a free module. If not, then M need not be free, but can
certainly be extended by ker(ϕ0 ) to a free module. We can measure how free
this extension is as well, for ker(ϕ0 ) likewise need not be free, so choose a free
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

module P1 and epimorphism ϕ1 : P1 → ker(ϕ0 ). Iterating this process infinitely

many times, we can construct a chain complex of free modules that maps onto
M . Proposition 5.1 asserts that we could have done this same procedure with
projective modules. Formally, we can make the following definition.
Definition 5.4. Let M be an R-module. A projective resolution of M,
denoted P , is a collection of projective modules {Pi }i∈N and homomorphisms
ϕi : Pi → Pi−1 where P−1 = M such that the following chain complex is exact.
ϕ3 ϕ2 ϕ1 ϕ0
... → P2 → P1 → P0 → M → 0

If each Pi is free, we call it a free resolution.

The discussion preceding Definition 5.4 shows that any R-module M has a
free resolution, defined inductively for each degree by choosing a free module
mapping onto the kernel of the prior degree homomorphism. Of course, we
can say the same thing about projective resolutions, as Proposition 5.1 implies
that a free resolution is a projective resolution. For subsequent results, we only
require that the resolution be projective. If for some i ∈ N we have that ϕi is
bijective, we say that the resolution is finite. Consider the following example of
a finite projective resolution.
Example 5.4. Let R = Z; let M = Z/2Z.
ϕ1 ϕ0
... → 0 → Z → Z → Z/2Z → 0

The above sequence where ϕ0 maps x 7→ x + 2Z, i.e., it is the quotient homo-
morphism, and ϕ1 maps x 7→ 2x is an exact chain complex, and since we have
a bijection with the mapping x 7→ 2x, it is a finite projective resolution. One
may assume that every Pi for i > 1 is 0.
Here’s an infinite projective resolution.
Example 5.5. Let R = Z/4Z; let M = Z/2Z.
ϕ2 ϕ1 ϕ0
... → Z/4Z → Z/4Z → Z/2Z → 0

The above sequence where ϕ0 maps x + 4Z 7→ x + 2Z, and ϕi for i ≥ 1 maps

x + 4Z 7→ 2x + 4Z is an exact chain complex, and since none of the ϕi are
bijective, it is an infinite projective resolution.
Now, going back to the original motivation for studying resolutions, notice
that for a projective resolution of M , we may remove M from the sequence
without loss of data since M ∼ = coker(ϕ1 ). Effectively, this shows that a pro-
jective resolution of M completely encodes the structure of M , and thus one
can present M by its projective resolution. Naturally, one might ask if this
presentation is unique. In a way, it is. This section’s final result shows that any
projective resolution of a given R-module is unique up to quasi-isomorphism.
We present it, prove it, and conclude this section.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Theorem 5.3. Suppose P and P ′ are projective resolutions of M , as shown in

the diagram below.
ϕ3 ϕ2 ϕ1 ϕ0
... P2 P1 P0 M 0

ψ3 ψ2 ψ1 ψ0
... P2′ P1′ P0′ M 0

Then, there exists a chain map f = {fi : Pi → Pi′ }i∈N that induces an isomor-
phism on homologies f˜i : Hi (P ) → Hi (P ′ ) for all i ∈ N.
Proof. We define f inductively. By hypothesis, ψ0 : P0′ → M is a surjection,
and since P0 is projective, we have a lifting f0 : P0 → P0′ such that ψ0 ◦ f0 = ϕ0 .

P0′
f0
ψ0
ϕ0
P0 M

Of course, idM ◦ ϕ0 = ψ0 ◦ f0 , so for the purposes of our induction, we say that

idM = f−1 .
ϕ3 ϕ2 ϕ1 ϕ0
... P2 P1 P0 M 0
f0 id
ψ3 ψ2 ψ1 ψ0
... P2′ P1′ P0′ M 0

Now, suppose that fn : Pn → Pn′ and fn−1 : Pn−1 → Pn−1

′
exist such that
ψn ◦ fn = fn−1 ◦ ϕn . For any x ∈ Pn+1 ,

ψn ◦ (fn ◦ ϕn+1 )(x) = (ψn ◦ fn ) ◦ ϕn+1 (x)

= (fn−1 ◦ ϕn ) ◦ ϕn+1 (x)
= fn−1 ◦ (ϕn ◦ ϕn+1 )(x)
=0

By Proposition 2.1, im(fn ◦ ϕn+1 ) ⊆ ker(ψn ). Since ψn+1 surjects onto ker(ψn ),
′
Pn+1 projective implies that there exists a fn+1 : Pn+1 → Pn+1 such that
ψn+1 ◦ fn+1 = fn ◦ ϕn+1 . This gives us fi for all i ∈ N.
Define f˜i : Hi (P ) → Hi (P ′ ) by

k + im(ϕi+1 ) 7→ fi (k) + im(ψi+1 )

We leave it to the reader to verify that this is a well defined homomorphism.

Observe that for any i ∈ N, Hi (P ) = ker(ϕi )/ im(ϕi+1 ) = {0 + im(ϕn+1 )} and
Hi (P ′ ) = ker(ψi )/ im(ψi+1 ) = {0 + im(ψi+1 )}. As a homomorphism, f˜i maps
0+im(ϕn+1 7→ 0+im(ψi+1 ); this is trivially onto and one-to-one. Consequently,
we can see that f induces an isomorphism between Hi (P ) and Hi (P ′ ) for each
i ∈ N. 
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

In the last section of this work, we continue exploring the intrinsic abstract
mathematics behind homological algebra and examine yet a further application
of projective resolutions, in the context of category theory.

6 Tor Functor
As was mentioned in the previous section, projective resolutions are useful in the
sense that they help describe the structure of objects in an Abelian category
through their invariants. However, this is related to yet one powerful appli-
cation. One of the central theorems in homological algebra is the “universal
coefficient theorem”, a key component of algebraic topology that seeks to relate
homology groups of different topological shapes to those of simplicial complexes
(multi-dimensional triangles). Conceptually, this is the algebraic framework
through which we can simplify our analysis and classification of topologies, by
looking at simplicial complexes instead of complicated shapes. Tor functors,
the objects of discussion in this section, are the mechanism through which this
association is done.
Before continuing, it is important to refamiliarize ourselves with an essential
concept of module theory. Recall that given R-modules M and N , we may
define the tensor product M ⊗R N as an R-module, equipped with a map ι :
M × N → M ⊗R N . Recall the tensor product satisfies the condition that any
R-bilinear map f : M × N → S induces an R-linear map f : M ⊗R N → S such
that f = f ◦ ι. The proofs of existence and uniqueness of the tensor product
are given in most abstract algebra books, including Dummit and Foote [2]: for
conciseness, we will not repeat them here.
We now introduce briefly another concept from category theory to develop
the language needed to progress with homological algebra. Functors are one
of the main building blocks of category theory, acting as the bridges between
categories. Formally defined, for two categories C and D, a functor F will map
all objects A ∈ C to respective objects F (A) ∈ D and will map all morphisms
(f : A → B) ∈ C to respective morphisms (F (f ) : F (A) → F (B)) ∈ D.
Functors must preserve identity morphisms, so that F (idX ) = idF (X) , for all
X ∈ C, and functors must preserve composition of maps, so F (f ◦ g) = F (f ) ◦
F (g), for all f : X → Y and g : Y → Z ∈ C. As can be seen, this is an
additional level of abstraction than the simpler set or group theory, and it can
continue growing, by considering the category of categories and the functors of
functors.
Note that by definition, functors preserve composition of maps, so they also
preserve commutative diagrams. Now, our goal in this section is to use projective
resolutions to construct something known as the left-derived functors of an R-
module. However, before explaining what left-derived functors are, we begin by
defining what is known as a right-exact functor.
Definition 6.1. A functor F : ModR → ModS between modules in R and
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

modules in S is right-exact if for every short exact sequence of R-modules

0→A→B→C→0

it follows that the sequence

F (A) → F (B) → F (C) → 0

is exact in S modules, but the map F (A) → F (B) may not be injective.
The notion of a left-exact functor can be defined dually; alternatively, we
can say right-exact functors preserve cokernels and left-exact functors preserve
kernels. These definitions can be extended with ease to any Abelian category,
but for this paper we are only concerned with these functors over R-modules. An
example of a right-exact functor is the tensor product M ⊗R − : M odR → M odR ,
for R commutative. Note that this functor takes each R-module N to the
R-module M ⊗R N , and each R-module homomorphism f : N → P to the
morphism M ⊗R f : M ⊗R N → M ⊗R P defined by (M ⊗R f )(m⊗n) = m⊗f (n).
Now, given a right exact functor F and a R-module M , consider any projective
resolution of M :
. . . → P2 → P1 → P0 → M
Then apply F to the projective resolution:

. . . → F (P2 ) → F (P1 ) → F (P0 )

Note that coker(F (P1 ) → F (P0 )) ∼

= F (M ). Computing the homology of this
sequence at the i-th coordinate gives an object, denoted:

Li F (M ) = ker(F (Pi ) → F (Pi−1 ))/ im(F (Pi+1 ) → F (Pi ))

f
We further define L0 F (M ) = F (M ). Furthermore, consider any function M →
N , and choose projective resolutions Pi and Qi for M and N respectively.

... P2 P1 P0 M 0
f

... Q2 Q1 Q0 N 0

f
Then consider the composite P0 → M → N . Then since Q0 is projective and
the map Q0 → N is surjective, it follows that there exists an induced map f0
such that the diagram commutes:

... P2 P1 P0 M 0
f0 f

... Q2 Q1 Q0 N 0
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

To construct f1 , note in particular since the diagram commutes we have that

the composite

(P1 → P0 → Q0 → N ) = (P1 → P0 → M ) → N = f ◦ 0 = 0.

So in particular:

im(P1 → P0 → Q0 ) ⊆ ker(Q0 → N ) = im(Q1 → Q0 )

Hence f0 restricts onto the image of Q1 → Q0 , and then since Q1 is projective

and the map Q1 → im(Q1 → Q0 ) is surjective, we do indeed induce a map f1
from P1 to Q1 . In the same way, we can define maps f2 , f3 . . . such that the
diagram commutes:

... P2 P1 P0 M 0
f2 f1 f0 f

... Q2 Q1 Q0 N 0

Then applying F maintains that the diagram commutes, so taking the homology
induces maps:
0 0 0
... L2 F (M ) L1 F (M ) F (M )
F (f2 ) F (f1 ) F (f0 )
0 0 0
... L2 F (N ) L1 F (N ) F (N )

We denote Li F (f ) = F (fi ). To summarize, we have constructed, for each i ∈ N,

a set of R-modules Li F (M ) and a set of maps Li F (f ) : Li F (M ) → Li F (N ).
We consolidate this information in the following definition:
Definition 6.2. Given an R-module M and a right-exact functor F , we com-
pute the left-derived functors Li F as above.
We leave it to the interested reader to show that Li F is indeed a functor from
ModR to ModR . Currently, we have defined the left-derived functors based on
a specific projective resolution: we will now see that any projective resolution
suffices.
Theorem 6.1. Li F (M ) always exists and is independent of the choice of pro-
jective resolution.
Proof. Existence follows immediately from the construction of a free resolution
over a module, as discussed preceding Definition 5.4, and the fact that any free
resolution is also a projective resolution. A proof of uniqueness can be found at
Weibel §2.4.1 [9]. 
Proposition 6.1. If M is projective, then Li F (M ) = 0 for all i > 0.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Proof. One particular projective resolution of M is

id id id
... → M → M → M → M

The functor preserves identity maps, so the result is clear. 

Now, to explain the motivation for left-derived functors, we can start by
considering the following theorem.
Theorem 6.2. Given an exact sequence
f g
0→A→B→C→0

and a right-exact functor F , it follows that the left-derived functors of F induce

the following long exact sequence:

L2 F (f ) L2 F (g) δ2
... L2 F (A) L2 F (B) L2 F (C)

L1 F (f ) L1 F (g) δ1
L1 F (A) L1 F (B) L1 F (C)

F (f ) F (g)
L0 F (A) L0 F (B) L0 F (C) 0

Proof. The proof is lengthy and technical, although it is based applying “Snake
Lemma” to get the connecting homomorphismss, as were shown above. More
details can be found at Rotman §6.27 [7] 
However, a perhaps bigger motivation for left-derived functors lies in using
them to define the Tor functor, the focus of this section.
Definition 6.3. The Tor functors are the left-derived functors of M ⊗R −. In
particular, we denote TorR
i (M, N ) as the i-th left-derived functor of M ⊗R −,
evaluated at N .
The above is isomorphic to the i-th left-derived functor of − ⊗R N evaluated
at M in a similar manner to the isomorphism M ⊗R N ∼ = N ⊗R M , though
this statement is not obvious. Hence we may also consider Tor as a functor
of two variables TorR i (−, −) : ModR × ModR → ModR . When the ring R is
understood, we frequently use the notation Tori (M, N ) instead of TorR i (M, N ).
To illustrate this definition of the Tor functor in action, we can consider the
following simple example.
Example 6.1. Let R = C[x, y], with the ideal I = (x, y). Then, Tor1 (R/I, I) =
Tor2 (R/I, R/I) = C.
Notice that the name Tor has to do with a link between the Tor functor and
torsion, exemplified in the following exercise:
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

Proposition 6.2. Given a ring R, a R-module M , and a nonzerodivisor x ∈ R,

we have that Tor1 (R/(x), M ) ∼
= {m ∈ M | xm = 0}.
Proof. Consider the exact sequence
·x
0 → R → R → R/(x) → 0

The map R → R/(x) is the canonical projection, and the map ·x is just ring
multiplication in R by x. Because R is free and hence projective, we can consider
this exact sequence as a projective resolution of R/(x):
·x
. . . → 0 → 0 → R → R → R/(x)

We then apply tensor by M :

·x⊗id
. . . → 0 → 0 → R ⊗R M −→M R ⊗R M → R/(x) ⊗R M

Hence

Tor1 (R/(x), M ) = ker(·x ⊗ idM )/ im(0 → R ⊗R M ) ∼

= ker(·x ⊗ idM )

However, we have R ⊗R M ∼ = M via the bijection r ⊗ m 7→ rm. So ker(·x ⊗ idM )

is exactly those elements r ⊗R m for which xrm = 0, which bijects to exactly
those elements rm ∈ M for which xrm = 0. Since x is a nonzerodivisor, in this
case the Tor functor does compute the torsion. 
Now, naturally from its definition, the Tor functor also provides some very
valuable insights into understanding the tensor product operator, as well as
some of its applications in Ring theory, as can be seen in the proposition below.
Proposition 6.3. Given a ring R with ideals I, J, we have that

Tor1 (R/I, R/J) ∼

= (I ∩ J)/IJ

Proof. Consider the exact sequence

0 → J ֒→ R → R/J → 0

The maps are of course the inclusion and natural projection. Consider the long
exact sequence induced by the Tor functor with R/I (Theorem 6.2):

... Tor1 (R/I, J) Tor1 (R/I, R) Tor1 (R/I, R/J)

R/I ⊗R J R/I ⊗R R R/I ⊗R R/J 0

Recall that R is a free and hence projective module. Furthermore, recall that Tor
is a left-derived functor, so we can see from Proposition 6.1 that Tor1 (R/I, R) =
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

0. We also note that R/I ⊗R R ∼

= R/I and R/I ⊗R J ∼
= J/IJ, so we can rewrite
the long exact sequence:

. . . → 0 → Tor1 (R/I, R/J) → J/IJ → R/I → R/I ⊗R R/J → 0

In particular we find that since the map 0 → Tor1 (R/I, R/J) has a trivial image,
the kernel of the map Tor1 (R/I, R/J) → J/IJ is trivial, so in fact we have by
the first isomorphism theorem and exactness

Tor1 (R/I, R/J) ∼

= im(Tor1 (R/I, R/J) → J/IJ) ∼
= ker(J/IJ → R/I)

We wish to understand the map J/IJ → R/I. Reversing the isomorphisms, we

find that this is the inclusion map

R/I ⊗R J → R/I ⊗R R

r + I ⊗ s 7→ r + I ⊗ s
So after the isomorphisms, this is simply the map sending r + IJ to r + J for
r ∈ I. So the kernel of this map is exactly those elements r + IJ for which r ∈ J
as well, giving indeed that ker(J/IJ → R/I) ∼ = (I ∩ J)/IJ as desired. 
We now take a moment to pause and receive the skeptical claims that the def-
initions of projective resolutions and Tor are not, by nature, very constructible
structures, especially when working in categories other than R-modules. Al-
though this is true, the above examples and exercises indeed reveal that even
in the cases where Tor is constructible, the functors do indeed carry a deep
meaning of some of the fundamental properties of objects. For example, the
previous proposition immediately implies that if I + J = R, then IJ = I ∩ J,
a central result about ideals of a ring. In addition, the Tor functor also has
deep connections to flat7 R-modules, which the interested reader can explore
further in Dummit and Foote §17.1 [2]. Yet, as was alluded at the beginning of
this section, the biggest application of the Tor functor lies in helping prove the
“universal coefficient theorem” for homology, one of the most important results
from homological algebra. Our hope is that, after reading this guide, abstract
algebra students will be equipped with the necessary framework to delve deeper
into and appreciate the beauty of these fascinating topics.

7 Conclusion
Homological algebra, like many parts of abstract algebra, is an exercise in gener-
alizing known ideas to create different ways of thinking about problems; as was
evident in this work, the treated topics naturally had deep connections to topol-
ogy and category theory, from which we motivated examples and applications
of this field.
7 By definition, a module M over a ring R is flat if ∀f : A → B injective linear maps of

R-modules, then the map f ⊗R M : A ⊗R M → B ⊗R M is injective too. An interesting result

about flatness is that ∀i > 0, TorR
i (M, N ) = 0 if either A or B themselves are flat.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

We started this work by exploring the main building blocks behind homologi-
cal algebra, most notably chain complexes, homology modules, and homotopies.
Throughout this process, we discovered ideas such as being able to construct
chain complexes of chain complexes through an application the “Snake Lemma”,
as well as new ways of thinking about the classification of topological shapes,
through the computation of holes at various dimensions, as enabled by homol-
ogy modules. We also introduced the notion of what it means for two chain
complexes to be chain homotopic, which is the algebraic analog of asking about
when can one shape by continuously deformed into another. Having established
this foundation, we then moved into more advanced topics, including projective
resolutions and Tor functors. In these sections, we highlighted the importance
of these two concepts in helping describe algebraic structures through their in-
variants, but also illustrated why homological algebra is a field worth studying
by itself, beyond the topological applications.
Overall, through this guide, we hope to not only have provided our reader
with the necessary background to further study homological algebra, but also
with enough motivation and excitement to do so.
 A Beginner’s Guide to Homological Algebra: A Comprehensive Introduction
for Students

References
[1] Nick Alger. Explain ”homotopy” to me. Math Stack Exchange, 2017.
[2] David Steven Dummit and Richard M. Foote. Abstract Algebra. John Wiley
& Sons, 2004.
[3] David Eisenbud. Commutative Algebra with a View toward Algebraic Ge-
ometry. World Publishing Corp., 2008.
[4] Bob Gardner. Supplement. A Proof of the Snake Lemma. East Tennessee
State University, 2018.
[5] Kelsey Houston-Edwards. How Mathematicians Use Homology to Make
Sense of Topology. Quanta Magazine, 2021.
[6] Thomas W. Hungerford. Algebra. Springer, 1974.
[7] Joseph J. Rotman. An Introduction to Homological Algebra. Springer, 2009.
[8] Rachel Thomas. Maths in a Minute: Simplices – the atoms of topology. Plus
Magazine, 2021.
[9] Charles A. Weibel. An Introduction to Homological Algebra. Cambridge
University Press, 2011.