Kernel of a Character

In the study of abstract algebra, groups provide a powerful language for describing symmetry. To make these abstract structures concrete, we use representations, which translate group elements into matrices. However, a crucial challenge in group theory is uncovering a group's internal architecture—particularly identifying its normal subgroups, a process that is often complex and ad-hoc. This article introduces a surprisingly elegant tool from character theory, the kernel of a character, that simplifies this challenge by turning structural questions into straightforward arithmetic.

This introduction sets the stage for a deeper exploration. In the following chapters, you will first delve into the ​​Principles and Mechanisms​​ of the kernel, understanding its fundamental definition and the remarkable proof that connects a character's value to the identity matrix. Following this, the ​​Applications and Interdisciplinary Connections​​ chapter will demonstrate how this concept serves as a powerful detective's tool for finding normal subgroups, testing the faithfulness of representations, and revealing deep truths about the structure of simple and quotient groups.

To make abstract groups tangible, group theory utilizes representations, which are homomorphisms from a group $G$ to a group of invertible matrices. A representation, denoted $\rho$, maps each element $g$ from a group $G$ to a matrix, $\rho(g)$. A crucial property of a representation is that it respects the group's structure, meaning the matrix of a product of two elements is the product of their individual matrices: $\rho(ab) = \rho(a)\rho(b)$.

But like any camera, a representation might have blind spots. It might fail to distinguish certain elements from the most boring element of all—the identity. The set of all elements that our camera $\rho$ maps to the identity matrix is called the ​​kernel of the representation​​, $\ker(\rho)$. These are the elements that are, from the representation's point of view, "invisible."

Now, working with matrices can be cumbersome. What if we could simplify the picture? Instead of looking at the whole matrix, let's just look at a single number derived from it: its ​​trace​​, the sum of its diagonal entries. This number is called the ​​character​​, denoted by $\chi(g) = \text{tr}(\rho(g))$. The character is like a shadow of the matrix. It's much simpler, but does it lose too much information? Specifically, can we still spot the "invisible" elements—the kernel—just by looking at their shadows?

This is where the magic begins. An element $g$ is in the kernel of the representation, $\ker(\rho)$, if $\rho(g)$ is the identity matrix, $I$. The character of the identity element, $e$, is $\chi(e) = \text{tr}(\rho(e)) = \text{tr}(I)$. For a matrix of size $d \times d$, this trace is simply $d$, the dimension of our vector space, also known as the ​​degree​​ of the character. So, our question becomes: if a group element $g$ has a character value $\chi(g)$ equal to the dimension $d$, can we be sure it's in the kernel?

At first glance, this seems unlikely. Many different matrices can have the same trace. But a representation's matrices aren't just any matrices; they are very special. For a finite group, we can always think of the matrices $\rho(g)$ as being unitary. This means their eigenvalues are all complex numbers with an absolute value of 1—they all lie on the unit circle in the complex plane.

Let's say our representation has degree $d$. The matrix $\rho(g)$ has $d$ eigenvalues, $\lambda_1, \lambda_2, \dots, \lambda_d$, all on the unit circle. Its character is their sum: $\chi(g) = \sum_{i=1}^d \lambda_i$. Now, consider the condition $\chi(g) = d$. This means $\sum_{i=1}^d \lambda_i = d$. The famous triangle inequality tells us that the magnitude of a sum of complex numbers can't be greater than the sum of their magnitudes: $|\sum \lambda_i| \le \sum |\lambda_i|$. In our case, this becomes $|d| \le \sum_{i=1}^d |\lambda_i|$. Since each $|\lambda_i| = 1$, the right-hand side is just $d$. So we have $d \le d$.

The only way for the equality to hold—the only way for these numbers on the unit circle to add up to the real number $d$—is if they are not scattered at all. They must all be identical and point in the same direction along the positive real axis. In other words, every single eigenvalue $\lambda_i$ must be exactly 1.

A matrix whose eigenvalues are all 1 and which can be diagonalized (as is the case here) must be the identity matrix. So, $\rho(g)=I$. This is a stunning conclusion! The condition $\chi(g) = \chi(e) = d$ is both necessary and sufficient for an element $g$ to be in the kernel of the representation $\rho$. We have found that the kernel of the character and the kernel of the representation are exactly the same set.

The shadow did not lie. It contains all the information we need to identify the kernel.

This discovery is more than just a mathematical party trick. The kernel of any group homomorphism (like a representation $\rho$) is not just any subgroup; it's a ​​normal subgroup​​. A normal subgroup is a very special type of subgroup that behaves well under "conjugation" ($hNh^{-1} = N$). They are the building blocks for constructing simpler "quotient" groups, which is a central theme in understanding group structure. Finding them is often a difficult, bespoke process for any given group.

But characters hand us a key. Since $\ker(\chi)$ is the same as $\ker(\rho)$, the kernel of any character is automatically a normal subgroup of $G$. Suddenly, we have a systematic way to find normal subgroups! All we need is a representation, its character, and we can immediately identify one of these crucial structural components.

How do we do this in practice? A key fact is that characters are ​​class functions​​: all elements in the same conjugacy class have the same character value. It follows that the kernel of a character must be a union of entire conjugacy classes.

Let's look at the symmetry group of a square, $D_8$. Suppose we are given a character $\chi$ with values $\chi(e)=2$, $\chi(r)=2$, $\chi(r^2)=2$, $\chi(r^3)=2$ for the rotations, and $\chi(s)=0$ for all reflections. To find $\ker(\chi)$, we simply collect all elements $g$ for which $\chi(g) = \chi(e) = 2$. This gives us the set $\{e, r, r^2, r^3\}$, the subgroup of all rotations. And just like that, we've identified a normal subgroup within $D_8$.

This technique is incredibly powerful when we have a group's ​​character table​​, a grid listing the values of all its irreducible characters for each conjugacy class. By scanning the table for a character $\chi_i$, and identifying all classes where the value is equal to the degree $\chi_i(e)$, we can construct $\ker(\chi_i)$. Sometimes, this allows us to pinpoint very specific structures, like the minimal non-trivial normal subgroups of a group, which are fundamental to its composition. Even special properties, like an element being in the group's center, can simplify the analysis thanks to powerful tools like Schur's Lemma.

The beautiful structure of representation theory means that kernels behave in very predictable and intuitive ways when we combine characters.

• ​​Direct Sums (Addition):​​ If we build a new, larger representation by taking the direct sum of two smaller ones, $\rho = \rho_1 \oplus \rho_2$, its character is simply the sum of the individual characters: $\chi_{\oplus} = \chi_1 + \chi_2$. What about the kernel? For an element $g$ to be "invisible" to the combined representation, it must be invisible to both component representations simultaneously. This intuition is correct: the kernel of the sum is the intersection of the kernels:

  $$\ker(\chi_1 + \chi_2) = \ker(\chi_1) \cap \ker(\chi_2)$$

• ​​Tensor Products (Multiplication):​​ When we combine representations via a tensor product, the new character is the product of the old ones, for instance, $\psi = \lambda\chi$, where $(\lambda\chi)(g) = \lambda(g)\chi(g)$. If an element $g$ is in both $\ker(\lambda)$ and $\ker(\chi)$, then $\lambda(g)=1$ (assuming $\lambda$ is one-dimensional) and $\chi(g)=\chi(e)$. Their product is $\psi(g) = 1 \cdot \chi(e) = \psi(e)$, so $g$ is in $\ker(\psi)$. This shows that the intersection is contained within the new kernel. However, the reverse is not always true, leading to a more subtle relationship:

  $$\ker(\lambda\chi) \supseteq \ker(\lambda) \cap \ker(\chi)$$

• ​​Restriction:​​ What if we focus our attention on a subgroup $H$ of $G$? We can simply "restrict" our character $\chi$ to only look at elements from $H$. We call this restricted character $\chi\downarrow_H$. The elements of $H$ that are invisible to this restricted view are precisely those elements that were already invisible in the larger group $G$ and also happened to be in $H$. So, the relationship is a simple intersection:

  $$\ker(\chi\downarrow_H) = \ker(\chi) \cap H$$

This elegant "calculus" shows how the concept of the kernel is woven consistently throughout the theory, behaving in logical and often intuitive ways.

A single representation may not "see" the entire group; it might have a non-trivial kernel. If $\ker(\rho)$ contains more than just the identity element, the representation is called ​​unfaithful​​ because it maps different group elements to the same matrix. It's a distorted picture of the group.

This raises a profound question: could a non-trivial group element $g$ be so stealthy that it manages to hide in the kernel of every single irreducible representation? If so, the entire machinery of character theory would have a fundamental blind spot.

Here lies the final, beautiful payoff. The answer is no. A cornerstone theorem of representation theory states that the intersection of the kernels of all irreducible characters of a finite group is the trivial subgroup $\{e\}$.

Think about what this means. While any single character $\chi_i$ might have a blind spot, the collection of all irreducible characters sees everything. No element, other than the identity itself, can be invisible to all of them. If an element $g$ presents itself, at least one irreducible character will "call it out" by having a value $\chi_i(g) \neq \chi_i(e)$.

This is a statement of the deep completeness and power of character theory. The set of irreducible characters, taken together, forms a perfectly faithful blueprint of the group. By reducing complex matrices to simple numbers, we seemed to risk losing information. Yet, in the end, we find that these character "shadows," when viewed collectively, capture the entire essence of the group's structure.

The definition of a character's kernel—the set of group elements a character maps to its value at the identity—is more than a formal abstraction. The kernel serves as a powerful analytical tool for investigating the internal structure of a group. It translates the abstract search for specific group properties, such as identifying normal subgroups, into straightforward arithmetic checks based on the group's character table. This section explores how this concept provides profound insights, connecting a group's internal architecture to its representations in a practical and effective manner.

One of the most fundamental quests in group theory is the search for normal subgroups. These are the special subgroups that are invariant under conjugation, the internal symmetries of the group itself. They are the building blocks for constructing simpler groups (quotient groups) and are essential for understanding a group's structure. But finding them can be a messy business, involving checking every element and every subgroup.

Character theory hands us an elegant shortcut. As we've learned, the kernel of any character, $\ker(\chi)$, is always a normal subgroup. This fact is a gift. It means a character table, which at first glance looks like a sterile grid of complex numbers, is actually a treasure map pointing directly to a group's normal subgroups.

Let's see this magic in action. Consider the dihedral group $D_5$, the symmetry group of a regular pentagon. It's a group of order 10. How would we find its normal subgroups? We could try to do it by hand, but let's use our new tool. We look at the character table for $D_5$ and pick a character, say $\chi_2$. The rule is simple: find all group elements $g$ where $\chi_2(g) = \chi_2(e)$. The value $\chi_2(e)$ is just the character's dimension, which is 1. We scan the $\chi_2$ row of the table, looking for all the columns where the entry is 1.

Instantly, we see that the conjugacy classes representing the identity and the rotations all have a character value of 1. The reflections, however, have a value of -1. So, the kernel of $\chi_2$ is precisely the set of all rotations, $\langle r \rangle$. Just like that, with a simple table lookup, we've discovered a non-trivial normal subgroup of order 5.

This trick isn't a one-off. It's a general and powerful method. For the much more complex symmetric group $S_4$ (the 24 symmetries of a tetrahedron), a glance at its character table can reveal its secrets. The kernel of the character $\chi_5$ immediately points to the famous Klein four-subgroup $V_4$, a crucial normal subgroup of $S_4$. We could even have found this same subgroup by a more sophisticated move: taking the intersection of the kernels of two other characters, $\ker(\chi_2) \cap \ker(\chi_3)$. The fact that different characters can reveal the same structure speaks to the deep consistency of the theory.

When we represent a group with matrices, we are creating a "picture" of the group. A natural question to ask is, how good is the picture? Does it capture every detail of the group, or does it blur some elements together? A representation is called faithful if it's a perfect picture—if every distinct element of the group is mapped to a distinct matrix. In other words, the map from the group to the matrices is one-to-one.

How can we tell if a representation is faithful? We could try to check every single element, but that's cumbersome. Once again, the kernel provides a beautifully simple litmus test. A representation is faithful if and only if its kernel is the trivial subgroup, containing only the identity element, $\{e\}$. Why? Because the kernel consists of all the elements that the representation maps to the identity matrix. If more than just the identity element gets sent to the identity matrix, the map isn't one-to-one, and the representation isn't faithful.

Consider the standard permutation representation of $S_n$, the group of permutations of $n$ objects. The character $\chi(\sigma)$ of a permutation $\sigma$ in this representation has a wonderfully intuitive meaning: it's simply the number of objects that $\sigma$ leaves in place (its fixed points). The identity permutation, of course, leaves all $n$ objects in place, so $\chi(e) = n$. For the kernel, we are looking for all permutations $\sigma$ such that $\chi(\sigma)=n$. But what permutation fixes all $n$ objects? Only the identity! Thus, the kernel is $\{e\}$, and we know instantly that this fundamental representation is faithful for any $n$.

This simple check allows us to quickly assess representations. The 3-dimensional representation of the alternating group $A_4$, for instance, has a character $\chi_4$ whose value is 3 at the identity and non-3 everywhere else. Its kernel is therefore trivial, and the representation is faithful.

This brings us to one of the most elegant applications of character theory, a result of pure and stunning beauty. In the great "classification of finite simple groups," mathematicians have identified the fundamental "atoms" from which all finite groups are built. These are the simple groups, groups whose only normal subgroups are the trivial one, $\{e\}$, and the group itself, $G$. They are "unbreakable" in a certain algebraic sense.

What can our kernel tool tell us about these fundamental particles of group theory? The kernel of any character, $\ker(\chi)$, is a normal subgroup. If our group $G$ is simple, then there are only two possibilities for $\ker(\chi)$: it's either $\{e\}$ or it's all of $G$.

If $\ker(\chi) = G$, it means $\chi(g) = \chi(e)$ for all $g \in G$. The corresponding representation maps every element to the identity matrix. This is the trivial representation.

So, here is the profound conclusion: for any non-trivial irreducible character of a simple group, the kernel must be the trivial subgroup $\{e\}$. This means that every non-trivial irreducible representation of a simple group is faithful. This is an incredibly powerful structural constraint, uncovered by a simple line of reasoning about kernels.

We've seen that when the kernel is trivial, the representation is faithful. But what if the kernel is not trivial? Is the character useless? Far from it. A non-trivial kernel is just as illuminating, for it tells us about a simplified version of the group.

The elements in the kernel are "invisible" to the character; it treats them all as the identity. This act of ignoring the structure within a normal subgroup is precisely the idea behind forming a quotient group, $G/N$, where the entirety of a normal subgroup $N$ is collapsed down to a single identity element.

There is a deep and beautiful correspondence here. The irreducible characters of the quotient group $G/N$ are, in a very real sense, the same as the irreducible characters of $G$ that contain $N$ in their kernel. A character with $N$ in its kernel doesn't see the differences between elements inside a coset of $N$, so it is effectively a character of the quotient group $G/N$.

Take the dihedral group $D_8$, the symmetries of a square. Its center, $N = \{1, r^2\}$, is a normal subgroup. By inspecting the character table of $D_8$, we can find all the characters for which $\chi(r^2) = \chi(1)$. We find there are exactly four of them, all one-dimensional. Now, if we look at the quotient group $D_8/N$, its order is $|D_8|/|N| = 8/2 = 4$. This quotient group is isomorphic to the Klein four-group, $V_4$, which happens to have exactly four irreducible characters. The four characters we identified in $D_8$ are precisely the characters of $V_4$, lifted up to $D_8$. The kernel helped us find a bridge connecting the representation theory of a large group to that of its smaller, simpler quotient.

Finally, what happens when we combine the information from multiple characters? A character corresponding to a larger representation that is a sum of smaller ones, $\chi = \psi_1 + \psi_2$, carries the combined information of both. Its kernel is simply the intersection of the individual kernels, $\ker(\chi) = \ker(\psi_1) \cap \ker(\psi_2)$.

This idea gives us a sense of completeness. If we have a set of characters whose kernels only intersect at the identity element, it means there is no non-identity element that is "invisible" to all of them simultaneously. Each character provides a viewpoint, and together, they provide a complete picture of the group.

The journey from a simple definition to these far-reaching applications showcases the spirit of mathematics. The kernel of a character is more than a formal object; it is a lens, a test, and a bridge. It elegantly connects the numeric data in a character table to the deep, underlying symmetries and structures that govern the abstract world of groups, revealing its inherent beauty and unity in a way that Richard Feynman himself would have surely appreciated.