Cosets of a Subgroup

In the world of abstract algebra, groups provide a powerful framework for studying symmetry and structure. But a group is more than just a collection of elements; it possesses a rich internal architecture defined by its subgroups. A fundamental question then arises: how can we systematically explore this internal landscape and break down a complex, often infinite, group into simpler, more manageable pieces? The concept of a coset provides the answer, offering a method to 'slice' a group into non-overlapping partitions based on one of its subgroups. This article will guide you through this elegant idea. First, in "Principles and Mechanisms," we will explore the fundamental definition of a coset, see how cosets partition a group, and uncover their connection to the celebrated Lagrange's Theorem. We will also examine the crucial difference between left and right cosets. Following this, "Applications and Interdisciplinary Connections" will reveal how this seemingly abstract concept becomes a practical tool for organization, counting, and problem-solving in fields ranging from physical chemistry to quantum physics.

Imagine you have an infinite collection of beads, each labeled with an integer, strung out on a line from negative to positive infinity. This is our group, the integers under addition, or $(\mathbb{Z}, +)$. Now, let's say we are particularly interested in the multiples of four. This is a special club, a subgroup, let's call it $H = 4\mathbb{Z}$. It contains $\{\dots, -8, -4, 0, 4, 8, \dots\}$. It's a group in its own right: if you add any two multiples of four, you get another multiple of four.

Now, what about all the other integers? Where do they fit in? Let's take the number 1. It's not in our club. But we can use it to create a new collection. Let's add 1 to every single member of our club $H$. This gives us a new set, $1+H$, which looks like $\{\dots, -7, -3, 1, 5, 9, \dots\}$. This is a ​​coset​​. It's not a subgroup (it doesn't contain the identity element, 0), but it has a definite structure. It's a "shifted" version of our original subgroup. All the elements in this new set share a common property: they all leave a remainder of 1 when divided by 4.

We can play this game again. What about 2? It's not in $H$ or in $1+H$. Let's form its coset: $2+H = \{\dots, -6, -2, 2, 6, 10, \dots\}$. These are all the integers that leave a remainder of 2 when divided by 4. And again with 3: $3+H = \{\dots, -5, -1, 3, 7, 11, \dots\}$.

What happens if we try 4? Well, 4 is already in our original club $H$, so $4+H$ just gives us back $H$ itself. Trying 5? $5+H$ will give you the exact same set as $1+H$. You see, we've done it. We have sorted every single integer into one of four bins, or cosets.

1. Numbers with remainder 0 (the subgroup $H$ itself).
2. Numbers with remainder 1 (the coset $1+H$).
3. Numbers with remainder 2 (the coset $2+H$).
4. Numbers with remainder 3 (the coset $3+H$).

These four sets are the ​​left cosets​​ of the subgroup $4\mathbb{Z}$ in $\mathbb{Z}$. We have sliced the entire infinite group of integers into four distinct, non-overlapping infinite slices.

This idea isn't limited to infinite sets or simple addition. Consider a clock with 6 hours, numbered 0 through 5. This is the group $\mathbb{Z}_6$. Let's take the subgroup $H = \{0, 3\}$. Starting from 0 and adding 3 gets you to 3. Adding 3 again gets you back to 0 (since $3+3=6 \equiv 0 \pmod 6$). Now, let's form the cosets.

• The coset starting with 0 is just $H$ itself: $0+H = \{0, 3\}$.
• Pick an element not yet used, say 1. The coset is $1+H = \{1+0, 1+3\} = \{1, 4\}$.
• Pick another unused element, say 2. The coset is $2+H = \{2+0, 2+3\} = \{2, 5\}$.

And we're done! We've sorted all six elements of $\mathbb{Z}_6$ into three neat pairs: $\{0, 3\}, \{1, 4\},$ and $\{2, 5\}$. Each pair is a coset, a clean slice of the group.

Look closely at what happened in both examples. Every element of the original group landed in exactly one of the cosets. The cosets didn't overlap at all, and their union gave us back the entire group. When a set is divided up like this—into non-empty, disjoint subsets that cover the whole set—it's called a ​​partition​​.

This is not a lucky coincidence. It's a fundamental truth of group theory. For any group $G$ and any subgroup $H$, the set of left cosets of $H$ will always form a partition of $G$. This is a direct, beautiful consequence of the group axioms themselves. Two cosets, $aH$ and $bH$, are either completely identical or entirely disjoint. There is no halfway.

Let's see this in a more physical setting. Consider the symmetries of a square. The set of all these symmetries—rotations and reflections that leave the square looking the same—forms a group called the ​​dihedral group​​ $D_4$. It has eight elements. Four of them are rotations: by $0^\circ$ (the identity, $e$), $90^\circ$ ($r$), $180^\circ$ ($r^2$), and $270^\circ$ ($r^3$). These four rotations form a subgroup, let's call it $\mathcal{R}$.

What are the cosets of $\mathcal{R}$ in $D_4$? Well, one coset is $\mathcal{R}$ itself: the set of all rotations. Now, let's pick an element that's not a rotation, for example, a reflection $s$ across the horizontal axis. What is the coset $s\mathcal{R}$? It's the set you get by performing a rotation, and then following it up with the reflection $s$. If you take any rotation and combine it with a reflection, you don't get another rotation; you get another reflection! So, the coset $s\mathcal{R}$ is the set of all four reflections in the group.

So, the cosets have partitioned the symmetries of a square into two fundamental types:

1. The orientation-preserving transformations (the rotations, $\mathcal{R}$).
2. The orientation-reversing transformations (the reflections, $s\mathcal{R}$).

This isn't just a mathematical curiosity. It's a blueprint for the structure of the group. The existence of the subgroup of rotations forces the entire group of symmetries to be neatly cleaved into these two distinct families of transformations. This partitioning principle applies universally, from the integers to the symmetries of a square or the abstract structure of quaternions.

There is another deep pattern you might have noticed. In $\mathbb{Z}_6$, the subgroup $H$ had 2 elements, and its cosets also had 2 elements. In $D_4$, the subgroup $\mathcal{R}$ had 4 elements, and its coset also had 4 elements. This, too, is no accident. Every single coset of a subgroup $H$ has the exact same size as $H$ itself. Why? Because you can create a perfect one-to-one correspondence between them. For any element $g$, the mapping that takes an element $h$ from the subgroup $H$ to the element $gh$ in the coset $gH$ is a perfect pairing. No elements are missed, and none are counted twice.

This simple observation, when applied to finite groups, leads to one of the most elegant and powerful results in all of abstract algebra: ​​Lagrange's Theorem​​. It states that if $G$ is a finite group and $H$ is a subgroup of $G$, then the order of $H$ (the number of elements in it, denoted $|H|$) must divide the order of $G$ (denoted $|G|$).

The proof is now almost self-evident. The group $G$ is partitioned into a collection of cosets. Each coset has size $|H|$. If there are $k$ such cosets, then the total number of elements in $G$ must be $k \times |H|$. Therefore, $|G| = k|H|$, which means $|H|$ divides $|G|$. The number of cosets, $k$, is called the ​​index​​ of $H$ in $G$, written as $[G:H]$, and it is simply $\frac{|G|}{|H|}$.

This theorem is a powerful constraint on the structure of finite groups. A group with 15 elements cannot have a subgroup with 4 elements. It acts as a fundamental selection rule, telling us what kinds of substructures are possible and which are forbidden.

Let's see its power in a more complex scenario. Consider the group $G$ of all invertible $2 \times 2$ matrices with entries from a finite field $\mathbb{Z}_p$ (the integers modulo a prime $p$). This is the group $GL_2(\mathbb{Z}_p)$. Now consider the subgroup $H$ of all upper-triangular matrices within this group. How many distinct left cosets of $H$ are there in $G$? This seems like a horribly complicated question. But Lagrange's theorem gives us a clear path: we just need to count the elements in $G$ and $H$.

A bit of clever counting reveals that $|G| = (p^2-1)(p^2-p)$ and $|H| = p(p-1)^2$. The number of cosets is the ratio: $[G:H] = \frac{|G|}{|H|} = \frac{(p^2-1)(p^2-p)}{p(p-1)^2} = \frac{(p-1)(p+1)p(p-1)}{p(p-1)^2} = p+1$ Out of all that complexity emerges a number of breathtaking simplicity: $p+1$. This is the magic of the coset concept. It allows us to find simple, profound patterns in structures that seem bewilderingly complex. Whether it's a direct product group or a group of matrices over a finite field, the principle holds.

So far, we have been forming our cosets by multiplying by an element $g$ on the left: $gH$. This is a "left coset". A natural question arises: what if we multiply on the right? What about "right cosets," sets of the form $Hg$?

For many groups, like the integers or any other ​​abelian​​ (commutative) group, there is no difference. Since the operation is commutative, $g+H$ is identical to $H+g$. But what about groups where order matters?

Let's return to the group of permutations $S_3$. Consider the subgroup $H = \{e, (12)\}$, generated by swapping 1 and 2. We can compute the left cosets. For example, let's take the permutation $g=(13)$. The left coset is: $(13)H = \{(13)e, (13)(12)\} = \{(13), (123)\}$ Now let's compute the right coset with the same element: $H(13) = \{e(13), (12)(13)\} = \{(13), (132)\}$ They are not the same! $(13)H \neq H(13)$. This is a fascinating discovery. It reveals a kind of "lopsidedness" in the group's structure relative to this subgroup. The way the group elements interact with $H$ depends on which side you approach from.

This distinction is tremendously important. Subgroups for which the left and right cosets always coincide (i.e., $gH = Hg$ for all $g$ in the group) are called ​​normal subgroups​​. They have a special kind of symmetry and are the essential building blocks for constructing new groups from old ones, a topic for another day.

Even though the individual cosets might differ, there is still a beautiful symmetry at a higher level. The number of left cosets is always equal to the number of right cosets. There is a simple, elegant way to prove this by constructing a perfect one-to-one mapping between the two sets. Consider the map $\phi$ that sends a left coset $gH$ to the right coset $Hg^{-1}$. Notice the sneaky inverse! That $g^{-1}$ is the magic ingredient that makes the whole thing work perfectly. This mapping provides a formal bridge between the world of left cosets and the world of right cosets, showing that even when they look different, they are equal in number, reflecting a deep, hidden unity in the structure of the group.

You might be tempted to think that cosets, this idea of slicing a group into neat, identical pieces, are a clever but purely abstract game for mathematicians. After all, once we have proven a theorem like Lagrange's, what more is there to do? It turns out, this is just the beginning of the story. The concept of a coset is not merely a tool for partitioning; it is a powerful lens that reveals hidden structure, a building block for new mathematical objects, and, most surprisingly, a practical instrument used in the laboratories of chemists and the whiteboards of quantum physicists. It is one of those beautiful ideas that, once understood, starts appearing everywhere, unifying seemingly disparate fields of science.

Let's start with the most direct application: counting. If you have a group $G$ and a subgroup $H$, Lagrange's theorem tells you that the order of $H$ must divide the order of $G$. The reason, as we've seen, is that the cosets of $H$ all have the same size and they perfectly tile the entire group $G$. This means we can know the number of cosets—the index $[G:H]$—simply by dividing the size of the group by the size of the subgroup, without ever having to list them. For a large group like the symmetric group $S_4$ (with $4! = 24$ elements), finding the number of cosets of a small subgroup, say one with 3 elements, becomes a trivial calculation: $24/3 = 8$. We instantly know the group can be chopped into 8 distinct blocks, a piece of information that would be tedious to discover by hand.

But cosets do more than just count; they organize. Think about the symmetries of a square, the dihedral group $D_4$. Intuitively, we feel that its eight symmetries fall into two natural categories: the four rotations that preserve the square's orientation, and the four reflections that flip it over. This intuition is made precise by the language of cosets. The set of rotations forms a subgroup $H$. What about the reflections? It turns out they are not a subgroup, but they do form a single, coherent set: the other coset of $H$!. The entire group of symmetries is neatly partitioned into $\{ \text{rotations} \}$ and $\{ \text{reflections} \}$. This is a general feature for the symmetries of any regular polygon: the group $D_n$ always splits cleanly into the coset of rotations and the coset of reflections. A vague physical intuition is thus exposed as a deep mathematical structure.

This organizing principle extends far beyond geometric shapes. Consider the group of all affine functions, those of the form $f(x) = ax+b$ with $a \neq 0$. This is a vast, infinite group. Where are the cosets here? Let's look at the subgroup $T$ of pure translations, where the slope $a$ is fixed to 1. If we form the right cosets of $T$, a remarkable picture emerges. Each coset is the set of all affine functions that share the same slope. The coset represented by $g(x) = 2x$, for instance, is the collection of all lines with a slope of 2. The partition of the group into cosets is the same as partitioning all the lines on a plane into families of parallel lines. Once again, an algebraic decomposition reveals a familiar geometric reality.

So far, we have viewed cosets as static partitions of a set. But the next leap in understanding is to see them as dynamic objects in their own right. A group can act on its own set of cosets, shuffling them around like cards in a deck. This seemingly circular idea is incredibly powerful.

For one, it provides a profound new perspective on Lagrange's theorem itself. By viewing the group $G$ as acting on the cosets of its subgroup $H$, we can apply a different powerful tool, the Orbit-Stabilizer Theorem. The "stabilizer" of a coset turns out to be the subgroup $H$ itself, and the "orbit" is the entire collection of all cosets. The theorem then immediately spits out the famous formula: the number of cosets is $|G|/|H|$. Seeing the same truth emerge from two different angles is a hallmark of deep connection in physics and mathematics; it tells us our concepts are robust and unified.

More importantly, this action on cosets allows us to build new things. In physics and mathematics, we often study groups by representing them as groups of matrices. These "representations" are like shadows of the group that are easier to analyze. A fundamental question is: how can we construct these representations? The action on cosets gives us a direct method. If a group $G$ has $[G:H] = n$ cosets of a subgroup $H$, then the action of shuffling these $n$ cosets gives us a representation of $G$ as a group of $n \times n$ permutation matrices. This is called a permutation representation, and it's a standard technique for turning an abstract group into a concrete set of matrices, whose properties can then be studied with the tools of linear algebra. The study of how coset structures relate to each other under group mappings (homomorphisms) further deepens this connection, showing how the skeleton of a group, its coset lattice, transforms in predictable ways.

This is where the story leaves the realm of pure mathematics and lands squarely in the physical world. The abstract machinery of cosets has become an indispensable tool for scientists.

​​Physical Chemistry:​​ Consider a molecule like methane, $\text{CH}_4$. It has a beautiful tetrahedral symmetry, described by the point group $T_d$, which has 24 symmetry operations. A chemist wanting to understand methane's vibrational modes (which determine its infrared spectrum) or its electronic orbitals needs to use the representation theory of this group. This involves calculations that sum quantities over all 24 group elements. This is tedious, to say the least. However, a clever chemist will notice that a subset of these operations, for instance those that form the $D_{2d}$ point group (with 8 elements), form a subgroup $H$. By Lagrange's theorem, the 24 elements of $T_d$ can be partitioned into $24/8 = 3$ cosets of $H$. One coset is $H$ itself, and there are two others. It turns out that for many important calculations, such as those involving the characters of representations, the sum over all elements in a given coset is a constant, or related in a simple way to the sum over the subgroup $H$. This allows the chemist to perform a calculation on the 8 elements of the simpler subgroup $H$ and then, with minimal extra work, deduce the results for the other 16 elements. Cosets provide a systematic way to exploit symmetry to reduce a large, complex problem into smaller, manageable chunks.

​​Quantum Information and Physics:​​ The application of cosets is perhaps even more striking in the quantum world. The building blocks of quantum computers are qubits. The fundamental operations on a system of $n$ qubits are described by a structure called the Pauli group. To understand the behavior of a quantum computer, we must understand the representations of this group. Just as in the chemistry example, these groups can be very large, and we need clever ways to analyze them. One way is to build a representation from the group's action on the cosets of a chosen subgroup—for example, the subgroup $D_n$ consisting of "diagonal" Pauli operators.

Here, something truly remarkable can happen. We might try to build a representation on the cosets in hopes of finding the single most important representation—the one that describes the states of the $n$ qubits themselves. We do the math, and we find that the multiplicity of this crucial representation is exactly zero! Why? The reason lies deep in the structure of the subgroup and its cosets. In this case, the subgroup $D_n$ contains a central element $\omega$ which, in the crucial representation, must act like multiplication by $-1$. However, in the representation built from cosets, that subgroup is, by definition, the stabilizer of the base coset, meaning all its elements must act like the identity (multiplication by $+1$). These two conditions, $-v = v$, can only be satisfied if the vector $v$ is zero. The coset structure acts as a filter, and its specific nature prevents the very representation we seek from appearing. This is not a failure; it is a profoundly useful piece of information. It tells us that this particular substructure of the Pauli group is "blind" to the essential qubit states, a non-trivial insight that guides the design of quantum algorithms and error-correcting codes.

From a simple way of dividing a finite set into equal parts, the idea of a coset has grown into a far-reaching principle. It organizes symmetries, connects disparate mathematical theorems, and provides a computational backbone for chemistry and quantum physics. It is a testament to the power of abstract thought to illuminate the workings of the concrete world, revealing the hidden unity that lies beneath the surface of things.