Algebraic Number Theory

Algebraic number theory grew out of nineteenth-century attempts to prove Fermat’s Last Theorem by working not with ordinary integers but with richer number systems where new factorizations become possible. What emerged was a discipline that reveals deep structural truths about how integers, primes, and polynomial equations interlock — truths invisible to the elementary approach but laid bare once you enlarge the arena from $\mathbb{Z}$ to the rings of integers of arbitrary number fields. The subject stands at a crossroads: it extends the ideas of quadratic reciprocity and abstract algebra into a unified framework that today underpins the Langlands program, the proof of Fermat’s Last Theorem, and large swaths of modern cryptography.

Algebraic Numbers and Algebraic Integers

The most natural way to extend the rational numbers is to adjoin solutions of polynomial equations. A complex number $\alpha$ is called an algebraic number if it satisfies some non-zero polynomial with rational coefficients: there exist $a_0, a_1, \ldots, a_n \in \mathbb{Q}$, not all zero, such that

$a_n \alpha^n + a_{n-1} \alpha^{n-1} + \cdots + a_1 \alpha + a_0 = 0.$

Every rational number is trivially algebraic (it satisfies $x - q = 0$), but so is $\sqrt{2}$ (satisfying $x^2 - 2 = 0$), the primitive cube root of unity $\omega = e^{2\pi i/3}$ (satisfying $x^2 + x + 1 = 0$), and every root of every polynomial with rational coefficients. Numbers that are not algebraic — such as $\pi$ and $e$, as proved by Lindemann in 1882 and Hermite in 1873 respectively — are called transcendental.

Among algebraic numbers, a distinguished subclass plays the role that ordinary integers play among rationals. An algebraic number $\alpha$ is called an algebraic integer if it satisfies a monic polynomial with integer coefficients:

$\alpha^n + c_{n-1}\alpha^{n-1} + \cdots + c_1 \alpha + c_0 = 0, \quad c_i \in \mathbb{Z}.$

The requirement that the leading coefficient be $1$ is precisely what distinguishes integers from fractions: the rational number $\frac{3}{2}$ satisfies $2x - 3 = 0$ (with rational coefficients) but satisfies no monic polynomial with integer coefficients, so it is not an algebraic integer. The set of all algebraic integers forms a ring — sums and products of algebraic integers are again algebraic integers — a fact that is elegant but not obvious.

The minimal polynomial of an algebraic number $\alpha$ over $\mathbb{Q}$ is the unique monic polynomial of least degree in $\mathbb{Q}[x]$ that has $\alpha$ as a root. It is always irreducible over $\mathbb{Q}$. The degree of $\alpha$ is the degree of its minimal polynomial, which equals the dimension $[\mathbb{Q}(\alpha) : \mathbb{Q}]$ of the field extension generated by $\alpha$. The norm $N(\alpha)$ and trace $\text{Tr}(\alpha)$ are, respectively, the product and the sum of all Galois conjugates of $\alpha$ — the roots of its minimal polynomial. These quantities are always rational, and when $\alpha$ is an algebraic integer, they are ordinary integers. Norms and traces are the algebraic number theorist’s primary bridge between the abstract and the concrete.

The historical impetus came from Kummer’s 1844–1847 attempts to prove Fermat’s Last Theorem by factoring $x^p + y^p$ in the ring $\mathbb{Z}[\zeta_p]$ of integers in the cyclotomic field $\mathbb{Q}(\zeta_p)$, where $\zeta_p = e^{2\pi i/p}$ is a primitive $p$-th root of unity. Kummer discovered that unique factorization fails in these rings — an obstacle he circumvented by introducing ideal numbers, later formalized by Dedekind into the modern theory of ideals.

Number Fields and Rings of Integers

A number field is a finite-degree field extension of $\mathbb{Q}$. Every number field $K$ has the form $K = \mathbb{Q}(\alpha)$ for some algebraic number $\alpha$ by the primitive element theorem, and its degree is $[K : \mathbb{Q}]$. The simplest examples are the quadratic fields $\mathbb{Q}(\sqrt{d})$ for a squarefree integer $d$, which have degree $2$, and the cyclotomic fields $\mathbb{Q}(\zeta_n)$, which have degree $\varphi(n)$ and play a central role in the theory of reciprocity laws.

Inside any number field $K$, the ring of integers $\mathcal{O}_K$ consists of all elements of $K$ that are algebraic integers. It is always a free $\mathbb{Z}$-module of rank $[K : \mathbb{Q}]$, meaning there exists an integral basis $\{\omega_1, \ldots, \omega_n\}$ such that every element of $\mathcal{O}_K$ can be written uniquely as $c_1 \omega_1 + \cdots + c_n \omega_n$ with $c_i \in \mathbb{Z}$. For the quadratic field $K = \mathbb{Q}(\sqrt{d})$, the ring of integers is

$\mathcal{O}_K = \begin{cases} \mathbb{Z}\!\left[\tfrac{1+\sqrt{d}}{2}\right] & \text{if } d \equiv 1 \pmod{4}, \\ \mathbb{Z}[\sqrt{d}] & \text{if } d \equiv 2 \text{ or } 3 \pmod{4}. \end{cases}$

The discriminant $\Delta_K$ of $K$ is an integer that encodes the geometry of $\mathcal{O}_K$ inside $K$; it equals the determinant of the matrix $(\text{Tr}(\omega_i \omega_j))$. A prime $p$ ramifies in $K$ if and only if $p \mid \Delta_K$, a fact that connects the discriminant to the branching behavior of primes.

Rings of integers are Dedekind domains: integral domains in which every non-zero ideal factors uniquely into a product of prime ideals. This is a remarkable structural theorem proved by Dedekind in 1871 in his landmark supplement to Dirichlet’s Vorlesungen über Zahlentheorie. While $\mathcal{O}_K$ may fail to be a unique factorization domain (UFD) — in $\mathbb{Z}[\sqrt{-5}]$, for instance, $6 = 2 \cdot 3 = (1 + \sqrt{-5})(1 - \sqrt{-5})$ are two distinct factorizations into irreducibles — the failure is completely captured by the ideal structure: the ideals $(2)$, $(3)$, $(1+\sqrt{-5})$, and $(1-\sqrt{-5})$ all factor into prime ideals in a way that is unique. Dedekind’s insight was that unique factorization should be sought among ideals, not elements.

Units, Class Groups, and Ideal Theory

An element $u \in \mathcal{O}_K$ is a unit if it has a multiplicative inverse also lying in $\mathcal{O}_K$, equivalently if $N(u) = \pm 1$. The units form a group $\mathcal{O}_K^\times$ under multiplication. Dirichlet’s unit theorem, proved in 1846, gives a complete structural description of this group:

$\mathcal{O}_K^\times \cong \mu_K \times \mathbb{Z}^{r_1 + r_2 - 1},$

where $\mu_K$ is the (finite cyclic) group of roots of unity in $K$, $r_1$ is the number of real embeddings of $K$, and $r_2$ is the number of pairs of complex conjugate embeddings, with $r_1 + 2r_2 = [K:\mathbb{Q}]$. The integer $r = r_1 + r_2 - 1$ is called the unit rank. For $K = \mathbb{Q}(\sqrt{2})$, we have $r_1 = 2$, $r_2 = 0$, so the unit rank is $1$ and the units are $\pm (1+\sqrt{2})^n$ for $n \in \mathbb{Z}$ — the fundamental unit $1 + \sqrt{2}$ generates an infinite cyclic factor. For imaginary quadratic fields like $\mathbb{Q}(\sqrt{-1})$, the unit rank is $0$ and the units are purely the roots of unity: $\{\pm 1, \pm i\}$ in this case.

The ideal class group $\text{Cl}(K)$ measures precisely how far $\mathcal{O}_K$ is from being a UFD. Two non-zero fractional ideals $\mathfrak{a}$ and $\mathfrak{b}$ of $\mathcal{O}_K$ are equivalent if $\mathfrak{a} = x \mathfrak{b}$ for some $x \in K^\times$; the equivalence classes form a group under ideal multiplication, and this is the ideal class group. The order $h_K = |\text{Cl}(K)|$ is the class number of $K$. The key fact is: $\mathcal{O}_K$ is a UFD if and only if $h_K = 1$. The class number thus quantifies the failure of unique factorization.

Computing class numbers is a central algorithmic problem. Minkowski’s bound gives the decisive tool: every ideal class contains an integral ideal $\mathfrak{a}$ with $N(\mathfrak{a}) \leq M_K$, where the Minkowski bound is

$M_K = \frac{n!}{n^n} \left(\frac{4}{\pi}\right)^{r_2} \sqrt{|\Delta_K|}.$

Since there are only finitely many integral ideals of bounded norm, this shows that the class group is finite. Moreover, to compute the class group it suffices to determine the ideal classes of all prime ideals $\mathfrak{p}$ with $N(\mathfrak{p}) \leq M_K$ — a finite computation. For example, in $\mathbb{Q}(\sqrt{-5})$, one finds $M_K \approx 4.5$ and the class group is $\mathbb{Z}/2\mathbb{Z}$, reflecting the non-unique factorization of $6$ seen earlier.

The analytic class number formula, one of the crown jewels of the subject, expresses the class number through analytic means. The Dedekind zeta function of $K$ is

$\zeta_K(s) = \sum_{\mathfrak{a} \neq 0} \frac{1}{N(\mathfrak{a})^s} = \prod_{\mathfrak{p} \text{ prime}} \frac{1}{1 - N(\mathfrak{p})^{-s}},$

and it satisfies the class number formula: the residue of $\zeta_K(s)$ at $s = 1$ equals

$\lim_{s \to 1} (s-1)\zeta_K(s) = \frac{2^{r_1}(2\pi)^{r_2} h_K R_K}{w_K \sqrt{|\Delta_K|}},$

where $R_K$ is the regulator (the volume of the unit lattice) and $w_K$ is the number of roots of unity in $K$. This formula, a generalization of Dirichlet’s 1839 class number formula for quadratic fields, links algebraic invariants of $K$ to the analytic behavior of its zeta function — a connection that would eventually grow into the Langlands program.

Valuations and Local Fields

A valuation on a field $F$ is a function $|\cdot| : F \to \mathbb{R}_{\geq 0}$ satisfying $|x| = 0 \Leftrightarrow x = 0$, $|xy| = |x||y|$, and the triangle inequality $|x + y| \leq |x| + |y|$. If the stronger ultrametric inequality $|x + y| \leq \max(|x|, |y|)$ holds, the valuation is called non-archimedean. The ordinary absolute value on $\mathbb{Q}$ is archimedean; the $p$-adic absolute value $|n|_p = p^{-v_p(n)}$ — where $v_p(n)$ is the exponent of $p$ in the prime factorization of $n$ — is non-archimedean.

Ostrowski’s theorem (1916) classifies all valuations on $\mathbb{Q}$ up to equivalence: every non-trivial absolute value on $\mathbb{Q}$ is equivalent either to the ordinary absolute value or to the $p$-adic absolute value $|\cdot|_p$ for some prime $p$. This is the first indication that the primes of $\mathbb{Z}$ and the archimedean embedding $\mathbb{Q} \hookrightarrow \mathbb{R}$ stand on equal footing — they are the places of $\mathbb{Q}$, and number theory is richest when all places are considered simultaneously.

The $p$-adic numbers $\mathbb{Q}_p$ are constructed as the completion of $\mathbb{Q}$ with respect to $|\cdot|_p$, precisely as $\mathbb{R}$ is the completion with respect to the ordinary absolute value. Elements of $\mathbb{Q}_p$ have a canonical $p$-adic expansion:

$x = \sum_{k=m}^{\infty} a_k p^k, \quad a_k \in \{0, 1, \ldots, p-1\},$

which converges in the $p$-adic topology. The subring $\mathbb{Z}_p = \{x \in \mathbb{Q}_p : |x|_p \leq 1\}$ of $p$-adic integers consists of those elements with non-negative $p$-adic valuation. Within $\mathbb{Z}_p$, the unique maximal ideal is $p\mathbb{Z}_p$, and the quotient $\mathbb{Z}_p / p\mathbb{Z}_p \cong \mathbb{F}_p$ is the finite field with $p$ elements.

A local field is a locally compact, non-discrete topological field. Over characteristic zero, these are exactly $\mathbb{R}$, $\mathbb{C}$, and the finite extensions of $\mathbb{Q}_p$. The extension theory of local fields is much simpler than that of global fields: every finite extension $L/\mathbb{Q}_p$ is completely described by three invariants — the ramification index $e$, the residue degree $f$, and the relation $[L:\mathbb{Q}_p] = ef$. The prime $p$ factors in $\mathcal{O}_L$ as $p\mathcal{O}_L = \mathfrak{p}^e$, so $e > 1$ means $p$ is ramified in $L$.

Hensel’s lemma is the fundamental lifting tool for $p$-adic arithmetic. In one formulation: if $f(x) \in \mathbb{Z}_p[x]$ and $a \in \mathbb{Z}_p$ satisfies $f(a) \equiv 0 \pmod{p}$ and $f'(a) \not\equiv 0 \pmod{p}$, then there exists a unique $\hat{a} \in \mathbb{Z}_p$ with $f(\hat{a}) = 0$ and $\hat{a} \equiv a \pmod{p}$. This is a $p$-adic analogue of Newton’s method, and it means that a simple root modulo $p$ lifts uniquely to an exact root in $\mathbb{Z}_p$. Hensel’s lemma implies, for instance, that every integer that is a square modulo $p$ (with $p$ odd) is a square in $\mathbb{Z}_p$.

The study of local fields is called local number theory, and it is indispensable for understanding global number fields. The passage from local to global is formalized through the adele ring $\mathbb{A}_\mathbb{Q} = \mathbb{R} \times \prod'_p \mathbb{Q}_p$ (a restricted product), which packages all completions simultaneously and enables powerful analytic techniques such as the Poisson summation formula on the adeles.

Ramification and Decomposition

When a prime $p \in \mathbb{Z}$ is extended to a prime ideal $\mathfrak{p}$ of $\mathcal{O}_K$, the prime $p$ may behave in one of three fundamentally different ways. The ideal $p\mathcal{O}_K$ factors into prime ideals of $\mathcal{O}_K$:

$p\mathcal{O}_K = \mathfrak{p}_1^{e_1} \mathfrak{p}_2^{e_2} \cdots \mathfrak{p}_g^{e_g}.$

Each prime $\mathfrak{p}_i$ has a residue degree $f_i = [\mathcal{O}_K/\mathfrak{p}_i : \mathbb{F}_p]$, and the fundamental identity $\sum_{i=1}^g e_i f_i = n = [K:\mathbb{Q}]$ holds. The prime $p$ is ramified in $K$ if any $e_i > 1$, split if $g > 1$ and all $e_i = f_i = 1$, and inert if $g = 1$ and $f_1 = n$. Ramification is the most delicate case: it signals that $p$ divides the discriminant $\Delta_K$, and only finitely many primes ramify in any given number field.

For Galois extensions $K/\mathbb{Q}$ — those for which the Galois group $\text{Gal}(K/\mathbb{Q})$ acts transitively on the prime ideals above $p$ — the situation simplifies: all $e_i = e$ and all $f_i = f$ are equal, so $efg = n$. The decomposition group $D_\mathfrak{p} \leq \text{Gal}(K/\mathbb{Q})$ is the stabilizer of $\mathfrak{p}$, and the inertia group $I_\mathfrak{p} \unlhd D_\mathfrak{p}$ consists of those automorphisms that act trivially on $\mathcal{O}_K/\mathfrak{p}$. We have $|D_\mathfrak{p}| = ef$ and $|I_\mathfrak{p}| = e$; $p$ is unramified exactly when $I_\mathfrak{p}$ is trivial.

For unramified primes in a Galois extension, each decomposition group $D_\mathfrak{p} \cong \text{Gal}((\mathcal{O}_K/\mathfrak{p})/\mathbb{F}_p)$ is cyclic, generated by a canonical element called the Frobenius element $\text{Frob}_\mathfrak{p}$, characterized by the congruence

$\text{Frob}_\mathfrak{p}(x) \equiv x^p \pmod{\mathfrak{p}} \quad \text{for all } x \in \mathcal{O}_K.$

The Frobenius element is a cornerstone of modern number theory. Different primes $\mathfrak{p}$ lying over the same rational prime $p$ have conjugate Frobenius elements, so in an abelian Galois extension the Frobenius depends only on $p$, yielding a well-defined Artin symbol $(K/\mathbb{Q}, p) \in \text{Gal}(K/\mathbb{Q})$. The map $p \mapsto (K/\mathbb{Q}, p)$ is the Artin map of class field theory, and the fact that it is a group homomorphism from the group of fractional ideals to $\text{Gal}(K/\mathbb{Q})$ — the Artin reciprocity law — is the crowning theorem of class field theory, the branch of algebraic number theory that classifies all abelian extensions of a number field.

The distribution of Frobenius elements is governed by the Chebotarev density theorem (1922), which asserts that for any conjugacy class $C \subseteq \text{Gal}(K/\mathbb{Q})$, the proportion of primes $p$ with $\text{Frob}_p \in C$ equals $|C|/|\text{Gal}(K/\mathbb{Q})|$. This is a vast generalization of Dirichlet’s theorem on primes in arithmetic progressions — which is the special case where $K$ is a cyclotomic field and the Galois group is abelian. The Chebotarev density theorem is proved using the analytic properties of the Dedekind zeta function and its factorization into Artin $L$-functions, forging yet another link between the algebraic structure of number fields and the analytic behavior of their associated $L$-functions.

The theory of ramification and decomposition connects seamlessly to the study of quadratic reciprocity (as treated in that topic): the Legendre symbol $\left(\frac{p}{q}\right)$ records how the prime $q$ splits in $\mathbb{Q}(\sqrt{p^*})$, and the quadratic reciprocity law is a special case of the Artin reciprocity law for quadratic fields. The abstract algebra of Galois groups, normal subgroups, and field extensions (as treated in abstract algebra) provides the language in which all of these ideas are most clearly expressed. Together, these strands weave into algebraic number theory’s central ambition: to understand the absolute Galois group $\text{Gal}(\overline{\mathbb{Q}}/\mathbb{Q})$, the symmetry group of all algebraic numbers, through its representations and its action on arithmetic objects — a program that remains one of the deepest open problems in mathematics.