Short Exact Sequence

The short exact sequence is one of the most powerful and elegant concepts in modern mathematics. Represented by a concise chain of objects and maps, $0 \to A \to B \to C \to 0$, it provides a profound language for understanding how complex algebraic structures are built from simpler components. Its importance lies in its ability to precisely capture the relationship between a subobject ($A$), a larger object ($B$), and the resulting quotient structure ($C$). This article addresses the fundamental question of how algebraic objects are assembled and whether they can be broken down into their constituent parts, revealing the deep connections that underpin various mathematical fields.

This article will guide you through the world of short exact sequences in two main parts. First, we will explore the core "Principles and Mechanisms," dissecting the sequence to understand what exactness means at each step, the crucial difference between sequences that split and those that do not, and how these constructions are classified. Following this, the chapter on "Applications and Interdisciplinary Connections" will reveal how this abstract tool becomes a practical lens for solving problems and forging links between algebra, topology, and geometry.

A short exact sequence, that compact chain of arrows and letters $0 \to A \to B \to C \to 0$, is one of the most elegant and powerful pieces of notation in modern mathematics. It looks like a fragment of a secret code, but it’s actually a complete, self-contained story. It tells us how three objects are related, and more specifically, how a central object, $B$, is constructed from two others, $A$ and $C$. Our mission in this chapter is to become fluent in the language of these sequences, to understand the drama they encode, and to see how they form the backbone of countless arguments in algebra and topology.

Let's break down the sequence into its three main acts. We'll imagine our objects $A$, $B$, and $C$ are abelian groups (or modules, the intuition is the same), which are sets where we can add and subtract elements.

The first part of the sequence, $0 \to A \xrightarrow{f} B$, tells us that $A$ is faithfully represented inside $B$. The map $f$ is ​​injective​​, a one-to-one mapping. This is guaranteed by the "exactness" at $A$, which states that the kernel of $f$ is the image of the map coming from the initial zero object—a map whose image is just the zero element. So, $\ker(f) = \{0\}$, which is the very definition of injectivity. You can think of $f(A)$ as a perfect, undistorted copy of $A$ living inside $B$ as a subgroup. $A$ is the "subobject".

The final part, $B \xrightarrow{g} C \to 0$, tells us that $C$ is a kind of "shadow" or "quotient" of $B$. The map $g$ is ​​surjective​​, meaning every element in $C$ is the image of some element from $B$. The exactness at $C$ ensures this: the image of $g$ must equal the kernel of the final map into the zero object, which is the entire group $C$ itself. So $g$ covers all of $C$.

The central, most crucial part is the "exactness at $B$": $\text{im}(f) = \ker(g)$. This is the link that ties the story together. It says that the subgroup of $B$ that is the copy of $A$ is precisely the set of elements in $B$ that are "crushed" to zero by the map $g$. When we form $C$ by mapping $B$ with $g$, we are effectively ignoring the structure of $A$. By the First Isomorphism Theorem, this means $C$ is what's left of $B$ after we "quotient out" by the image of $A$. In other words, we have the fundamental relationship:

This isn't just an abstract statement. If you know $B$ is the group of integers modulo 10, $\mathbb{Z}_{10}$, and $A$ is the group $\mathbb{Z}_2$ which is mapped to the subgroup $\{0, 5\}$ inside $\mathbb{Z}_{10}$, then the sequence tells you $C$ must be the quotient group $\mathbb{Z}_{10} / \{0, 5\}$, which is isomorphic to $\mathbb{Z}_5$. The sequence acts like a calculator for group structures.

So, a short exact sequence tells a story of an object $B$ containing a subobject $A$ ($f(A)$), such that when $B$ is simplified by "collapsing" $A$ to nothing, the result is $C$. $B$ is, in a precise sense, an ​​extension​​ of $C$ by $A$.

This brings us to the most natural question imaginable: if $B$ is built from $A$ and $C$, is it just their simplest possible combination, the direct sum $A \oplus C$? Sometimes, the answer is a delightful "yes". When this happens, we say the sequence ​​splits​​.

A split sequence is one where the "gluing" of $A$ and $C$ to form $B$ is trivial. Imagine a ship in a bottle. The ship is $B$. It contains the rigging, $A$, and the hull, $C$. Is the ship just a hull with rigging sitting next to it? Or is the rigging inextricably woven through the hull in a complex way? A split sequence is like the first case.

Formally, a sequence splits if $B$ is isomorphic to $A \oplus C$. But how can we tell? There are two beautiful and equivalent criteria that are often easier to check:

1. ​​Existence of a Retraction:​​ There is a map $r: B \to A$ that "pulls back" $B$ onto $A$ in a way that is the identity for the elements that were already in $A$. That is, $r \circ f = \text{id}_A$. The map $r$ acts like a principled dismantling, recognizing the $A$ part of $B$ and extracting it perfectly.

2. ​​Existence of a Section:​​ There is a map $s: C \to B$ that embeds $C$ into $B$ as a subgroup, such that when we then project $B$ back to $C$, we get $C$ back. That is, $g \circ s = \text{id}_C$. This means we can find a "cross-section" of the $B \to C$ projection—a pristine copy of $C$ sitting inside $B$.

If any of these conditions hold, $B$ falls apart neatly into its constituent pieces, $A$ and $C$.

Here is where the story gets really interesting. The power and richness of this theory come from the fact that sequences often do not split. The object $B$ can be a genuinely new structure—a "twisted" product of $A$ and $C$ that cannot be untangled.

The most famous example is the sequence of integers:

Here, $A = \mathbb{Z}$, $B = \mathbb{Z}$, and $C = \mathbb{Z}_2$. The map $f$ is multiplication by 2, embedding $\mathbb{Z}$ into itself as the even numbers. The map $g$ reduces an integer modulo 2. The image of $f$ (the even numbers) is precisely the kernel of $g$. The sequence is exact.

Does it split? If it did, it would mean $B \cong A \oplus C$, or $\mathbb{Z} \cong \mathbb{Z} \oplus \mathbb{Z}_2$. But this is impossible! The group $\mathbb{Z}$ is ​​torsion-free​​: no non-zero element, when multiplied by an integer, can become zero. But the group $\mathbb{Z} \oplus \mathbb{Z}_2$ has a torsion element, $(0, 1)$, because $2 \cdot (0, 1) = (0, 0)$. They cannot be the same. The middle $\mathbb{Z}$ is a fundamentally different object from the simple direct sum. It's a non-trivial extension.

This non-splitting phenomenon is the difference between simple assembly and true synthesis. Consider building a group of order 4. We can take two copies of $\mathbb{Z}_2$ and form the direct sum $\mathbb{Z}_2 \oplus \mathbb{Z}_2$. This corresponds to a split short exact sequence. But there is another group of order 4, the cyclic group $\mathbb{Z}_4$. It also fits into a sequence $0 \to \mathbb{Z}_2 \to \mathbb{Z}_4 \to \mathbb{Z}_2 \to 0$. This sequence does not split. The groups $\mathbb{Z}_2 \oplus \mathbb{Z}_2$ and $\mathbb{Z}_4$ are two different ways to be "built from two $\mathbb{Z}_2$s". One is trivial, one is twisted.

Interestingly, this "twisting" is a feature of modules over most rings, but not all. If we consider modules over a field—that is, vector spaces—every short exact sequence splits!. Given a subspace $U$ of a vector space $V$, one can always find a complementary subspace $W$ such that $V = U \oplus W$. This is not true for modules in general, and the failure of this property is precisely what short exact sequences are designed to measure.

A sequence doesn't just relate structures; it relates their properties. Information flows along the arrows in predictable, though sometimes surprising, ways. Let's revisit the idea of being torsion-free. Consider a general sequence $0 \to A \to B \to C \to 0$.

• If $B$ is torsion-free, then $A$ must be too, since it lives inside $B$ as a subgroup. However, $C$ might not be! Our friend $0 \to \mathbb{Z} \xrightarrow{\times 2} \mathbb{Z} \to \mathbb{Z}_2 \to 0$ is the perfect example: $B=\mathbb{Z}$ is torsion-free, but $C=\mathbb{Z}_2$ is a torsion group. This makes sense: the process of taking a quotient can introduce torsion.

• Conversely, if $A$ and $C$ are torsion-free, then $B$ must be torsion-free as well. The proof is a beautiful "diagram chase": if an element $b \in B$ has torsion, its image in $C$ must be zero (since $C$ is torsion-free). So $b$ must have come from $A$. But $A$ is also torsion-free, so $b$ must have been the zero element to begin with.

This shows the sequence as a powerful logical tool. Properties like being finite, cyclic, or free also propagate through the sequence in specific ways, allowing us to deduce facts about one object from facts about the others.

We've seen that for a given $A$ and $C$, there can be at least two ways to form $B$: the simple direct sum $A \oplus C$ (the split case) and potentially other, twisted versions like $\mathbb{Z}_4$. This begs the question: how many different extensions are there? Can we classify them?

The answer is one of the most profound results in the field. The collection of all possible extensions of $C$ by $A$ (up to a natural equivalence) is not just a messy pile; it forms an abelian group itself! This group is called the first ​​Ext group​​, denoted $\text{Ext}^1_R(C, A)$.

In this group, the "zero element" corresponds to the one trivial, split extension, $B \cong A \oplus C$. All the other, non-split, twisted extensions correspond to the non-zero elements of $\text{Ext}^1_R(C, A)$. For our example $A=C=\mathbb{Z}_2$, the group $\text{Ext}^1_{\mathbb{Z}}(\mathbb{Z}_2, \mathbb{Z}_2)$ turns out to be $\mathbb{Z}_2$. This means there is one non-zero element, corresponding to exactly one non-split extension—the $\mathbb{Z}_4$ group! The $\text{Ext}$ group perfectly classifies the ways $A$ and $C$ can be glued together.

Some modules are simply "allergic" to being part of a twisted construction. A module $P$ is called ​​projective​​ if every short exact sequence $0 \to A \to B \to P \to 0$ that ends in $P$ must split. Projective modules are so well-behaved that they refuse to form non-trivial extensions. They always guarantee the existence of a section. Free modules are the prime examples of projective modules, and it turns out that projective modules are precisely the pieces you can "chip off" of free modules—they are the direct summands of free modules. The dual notion of an ​​injective module​​ $I$ ensures that any sequence $0 \to I \to B \to C \to 0$ starting with $I$ must split.

The existence of non-zero $\text{Ext}$ groups—the fact that some sequences don't split—is the engine that drives a vast area of mathematics called homological algebra. It is a measure of how far the world of modules is from the simpler world of vector spaces. And it all begins with that simple, elegant line of arrows, a short story with a surprisingly deep plot.

We have seen that a short exact sequence is a tidy, powerful piece of algebraic machinery. But like any good tool, its true value is not in what it is, but in what it does. It's not just a definition to be memorized; it is a lens through which we can view the world, a language that reveals deep and often surprising connections between seemingly disparate fields of thought. Once you learn to speak this language, you start seeing these sequences everywhere, whispering secrets about the hidden structure of mathematics itself. Let's embark on a journey to see where this simple-looking arrow notation takes us.

At its heart, a short exact sequence $0 \to A \to B \to C \to 0$ tells a story about how a larger object $B$ is built from smaller pieces, a sub-object $A$ and a quotient object $C$. A natural first question is: can we always reconstruct $B$ by just sticking $A$ and $C$ together? The answer to this question is a fascinating "sometimes," and exploring that "sometimes" reveals the very anatomy of algebraic structures.

Let's start with the most familiar of groups: the integers modulo $n$, or $\mathbb{Z}_n$. These are the groups of clock arithmetic. If you have a short exact sequence of these cyclic groups, like $0 \to \mathbb{Z}_a \to \mathbb{Z}_b \to \mathbb{Z}_c \to 0$, it turns out this structure can only exist if $a$ is a divisor of $b$, and in that case, $c$ is forced to be $b/a$. This means the short exact sequence provides a complete census of all the ways one cyclic group can be a subgroup of another with a cyclic quotient. It's a crisp, elegant description of the internal structure of these fundamental groups.

But what happens when the groups are more complicated? Consider any finite group $G$ with a normal subgroup $H$. This setup naturally gives us the short exact sequence $1 \to H \to G \to G/H \to 1$. Now we can ask our question again: can $G$ be "decomposed" back into $H$ and $G/H$? The celebrated ​​Schur-Zassenhaus theorem​​ gives us a powerful condition for when this is possible. It states that if the orders of $H$ and $G/H$ are coprime, then the sequence splits. In our language, this means there is a way to "reverse" the projection map, embedding a copy of the quotient $G/H$ back into $G$. The consequence is that $G$ can be written as a semidirect product, $G \cong H \rtimes (G/H)$. So, the abstract question of whether a sequence splits tells us something very concrete about the group's structure: it can be built by taking the elements of $H$ and "stirring" them using the elements of $G/H$.

This is wonderful, but the most profound insights often come from when things don't work out so simply. The cases where a sequence doesn't split are where the magic happens. A prime example is the quaternion group, $Q_8$. Its center is the two-element group $Z(Q_8) = \{1, -1\}$, and the quotient group $Q_8/Z(Q_8)$ is the Klein four-group $V_4$. This gives us a central extension described by the sequence $1 \to \mathbb{Z}_2 \to Q_8 \to V_4 \to 1$. Does this sequence split? If it did, $Q_8$ would have to be a direct or semidirect product of $\mathbb{Z}_2$ and $V_4$. But it is not! The group $Q_8$ has a more intricate, "twisted" structure that cannot be so easily undone. This non-splitting nature is a precise algebraic fingerprint of that twist. Short exact sequences, therefore, don't just classify the simple cases; they precisely describe the non-trivial ways in which larger structures can be assembled.

This principle extends beautifully into the world of representation theory. A cornerstone of the subject, ​​Maschke's Theorem​​, tells us when representations of a group can be broken down into their simplest, irreducible components. It turns out this entire theorem can be rephrased in a breathtakingly simple way using our new language: under the right conditions, every representation is completely reducible if and only if every short exact sequence of modules splits. This is a moment of profound unification. An entire, seemingly complex theorem about decomposing representations is captured by a single, elegant statement about the behavior of these sequences.

If short exact sequences provide the anatomy of algebra, they form the very grammar of algebraic topology. In topology, we study shapes by associating algebraic objects—like groups—to them. The process of "homology" is a primary tool for doing this. It measures, in a sense, the number and type of "holes" in a space.

The very definition of the $i$-th homology or cohomology group, $H^i$, is encoded in a short exact sequence. The group $H^i$ is defined as the quotient of "cycles" ($Z^i$) by "boundaries" ($B^i$). A cycle is something that could be the boundary of a higher-dimensional object, but isn't necessarily one within our space. A boundary is a cycle that actually is the boundary of something. The homology group consists of cycles that are not boundaries—the "holes." This entire foundational concept, $H^i = Z^i / B^i$, is expressed perfectly by the canonical short exact sequence: $0 \to B^i \to Z^i \to H^i \to 0$. Homology, at its core, is a short exact sequence.

Building on this, homological algebra provides a powerful strategy: if you have a complicated object (like a module $M$), try to "approximate" it with simpler ones you understand well, like free modules. A "finite free resolution" does just this, giving a short exact sequence $0 \to F_1 \to F_0 \to M \to 0$, where $F_0$ and $F_1$ are free. You might think that there are many ways to do this, and you'd be right. But a remarkable fact is that certain properties are invariant. For instance, the difference in the "sizes" (ranks) of the free modules, $\text{rank}(F_0) - \text{rank}(F_1)$, is always the same, and it tells you something intrinsic about the original module $M$—its own rank. This is like measuring the weight of a oddly shaped object by balancing it against a set of standard, well-understood weights.

The pinnacle of this approach is found in the ​​Universal Coefficient Theorems (UCT)​​. Suppose you've done the hard work of computing the homology of a space $X$ with integer coefficients, $H_n(X; \mathbb{Z})$. What if you need to know the homology with coefficients in a different group, say $G$? The UCT provides the answer, and it does so, of course, with a short exact sequence. For homology, it states there is a split short exact sequence: $0 \to \left(H_n(X; \mathbb{Z}) \otimes G\right) \to H_n(X; G) \to \text{Tor}\left(H_{n-1}(X; \mathbb{Z}), G\right) \to 0$. A similar, related sequence exists for cohomology, involving the $\text{Hom}$ and $\text{Ext}$ functors. Don't worry too much about the new terms like $\otimes$, $\text{Tor}$, and $\text{Ext}$. The big idea is that the homology with new coefficients, $H_n(X; G)$, is built from two pieces. The main piece is just the old homology tensored with $G$. The second piece, the $\text{Tor}$ term, is a "correction" term that depends on the homology in the dimension below. The short exact sequence tells us precisely how these pieces fit together to give the full answer. It is a universal translation machine, and it is powered by the logic of short exact sequences.

The connections forged by short exact sequences are not confined to the abstract realms of algebra. They provide a sturdy bridge to the tangible world of geometry.

A fascinating correspondence links the algebra of groups to the topology of spaces. To any group $G$, one can associate a special "classifying space" $BG$ whose fundamental group is $G$. Now, what happens if we start with a short exact sequence of groups, $1 \to K \to G \to H \to 1$? Incredibly, this algebraic structure blossoms into a geometric one: the corresponding classifying spaces form a fiber sequence, $BK \to BG \to BH$. This sequence has profound topological consequences, chief among them being a ​​long exact sequence of homotopy groups​​. This long sequence weaves together the homotopy groups of all three spaces, linking the topological features of the fiber ($BK$), the total space ($BG$), and the base ($BH$) in an intricate, beautiful pattern. An algebraic relationship between groups becomes a deep geometric relationship between spaces.

Perhaps the most stunningly direct application comes from differential geometry. Imagine two surfaces, $S$ and $T$, inside a larger space $M$, like $\mathbb{R}^4$. Suppose they are oriented and intersect transversely at a point $p$. The intersection itself will be a lower-dimensional manifold, which we would also like to orient. For a zero-dimensional intersection (an isolated point), this just means assigning it a sign, $+1$ or $-1$. How do we decide? A short exact sequence of tangent spaces comes to the rescue: $0 \to T_p(S \cap T) \to T_pS \oplus T_pT \to T_pM \to 0$ The orientations of $S$, $T$, and $M$ provide orientations on their respective tangent spaces. The abstract algebraic properties of this sequence then uniquely determine a natural orientation for the intersection space $T_p(S \cap T)$. This allows one to rigorously define an "intersection number" by summing up these signs over all intersection points. What began as a purely algebraic notion provides the foundation for a concrete, computable geometric invariant.

From the building blocks of groups to the holes in a donut, and from the decomposition of representations to the collision of surfaces, the short exact sequence is a unifying thread. It is a simple, elegant concept that, once understood, reveals the profound and beautiful unity that underlies the vast landscape of mathematics.