Loading field map
Modular forms and automorphic forms sit at a crossroads in mathematics, where number theory, complex analysis, algebra, and geometry meet. At their core, they are highly symmetric functions—objects whose structure is so rigid that studying them reveals deep arithmetic information, such as the number of ways to represent an integer as a sum of squares, or the distribution of prime numbers. The field is unified by a single question: What functions are invariant, or transform predictably, under the action of a large symmetry group, and what do those functions tell us about the underlying arithmetic?
A modular form is a complex-analytic function defined on the upper half-plane, the set of complex numbers with positive imaginary part. This space is a natural model for hyperbolic geometry, and it carries a rich group of symmetries: the modular group, consisting of 2×2 integer matrices with determinant 1, acting by fractional linear transformations. A modular form of weight k (an integer) is a holomorphic function that transforms under this action by a specific factor, roughly scaling by the denominator of the transformation raised to the power k. It must also be "holomorphic at infinity," meaning it has a controlled growth as the imaginary part tends to infinity.
The simplest examples are Eisenstein series, which are built by summing over all lattice points in the plane. These series encode arithmetic information directly: the values of the Riemann zeta function at even integers appear as their Fourier coefficients. Another central example is the discriminant function Δ, a cusp form (a modular form vanishing at infinity) whose Fourier coefficients are the Ramanujan tau function, a sequence of integers with deep multiplicative properties.
The key structural fact is that modular forms of a fixed weight form a finite-dimensional vector space. This finiteness is what makes them so powerful: a modular form is determined by finitely many initial Fourier coefficients, yet it encodes infinitely many arithmetic facts. The space of modular forms decomposes into Eisenstein series (which are built from simpler arithmetic data) and cusp forms (which are more mysterious and carry the deeper arithmetic content). The theory of Hecke operators—linear maps on these spaces that correspond to multiplying Fourier coefficients in a compatible way—reveals that the cusp forms can be further decomposed into simultaneous eigenfunctions, each associated with a multiplicative function on the integers. This decomposition is the bridge to arithmetic: the Fourier coefficients of these eigenforms satisfy relations that mirror the factorization of integers.
Automorphic forms generalize modular forms in two directions. First, the symmetry group is enlarged: instead of the modular group, one considers more general arithmetic groups, such as congruence subgroups of the modular group, or groups of matrices over number fields. Second, the domain is generalized: instead of the upper half-plane, one works on symmetric spaces associated with reductive Lie groups, such as higher-dimensional hyperbolic spaces or the space of positive-definite matrices.
The defining property remains: an automorphic form is a smooth function on a symmetric space that is invariant (up to a character) under the action of an arithmetic group, satisfies certain differential equations (being an eigenfunction of the Laplacian or more general invariant differential operators), and has controlled growth. The theory naturally splits into two flavors. Holomorphic automorphic forms, which include modular forms, are complex-analytic and are studied through algebraic geometry and complex analysis. Real-analytic automorphic forms, such as Maass forms, are not holomorphic but are eigenfunctions of the hyperbolic Laplacian; they are studied through harmonic analysis and spectral theory.
This broader framework is not a mere generalization for its own sake. It is necessary to capture symmetries that the classical modular group misses. For instance, the theory of quadratic forms in many variables naturally leads to automorphic forms on symplectic or orthogonal groups, where the symmetry group reflects the structure of the quadratic form itself. The Langlands program, a vast web of conjectures, posits that all automorphic forms—across all groups and number fields—are organized into a single coherent structure, with deep connections to Galois representations and the arithmetic of elliptic curves.
The central reason automorphic forms matter for number theory is that their Fourier coefficients are arithmetic objects. For a modular form, the Fourier expansion at infinity has coefficients that are algebraic numbers, often integers, and these coefficients frequently count something: the number of representations of an integer by a quadratic form, the number of points on an elliptic curve modulo a prime, or the number of ideals of a given norm in a number field.
The deepest link is through L-functions. To each automorphic form, one attaches a Dirichlet series—an infinite sum over the Fourier coefficients—that converges in a half-plane and extends to a meromorphic function on the whole complex plane. These L-functions satisfy functional equations, relating their values at s to their values at 1−s, and they are conjectured to satisfy the Riemann hypothesis in a suitable sense. The celebrated modularity theorem, formerly the Taniyama–Shimura conjecture, states that the L-function of every elliptic curve over the rational numbers is the L-function of a modular form of weight 2. This theorem, proved in the 1990s, was the key ingredient in Andrew Wiles's proof of Fermat's Last Theorem, and it remains the archetypal example of the Langlands correspondence: a bridge between Galois representations (attached to elliptic curves) and automorphic forms.
The field is not organized into rival schools but rather into complementary approaches that emphasize different aspects of the same objects. These approaches have developed in layers, each building on and refining the others.
The classical analytic approach, originating in the nineteenth century with the study of elliptic functions and theta series, treats modular forms as functions to be manipulated with complex analysis. Its tools are Fourier expansions, contour integrals, and explicit formulas. This tradition produced the first deep results: the proof that the number of representations of an integer as a sum of four squares is eight times the sum of its divisors, and the discovery of the Ramanujan conjectures on the size of Fourier coefficients. The analytic approach remains essential for explicit computations and for the theory of L-functions, where the analytic continuation and functional equations are proved by integral representations.
The algebraic and geometric approach, developed from the mid-twentieth century, interprets modular forms as sections of line bundles on modular curves—algebraic curves that parametrize elliptic curves with extra structure. This perspective brings the full machinery of algebraic geometry to bear: one can reduce modular forms modulo primes, study their behavior at the boundary of the moduli space, and construct Galois representations attached to them. This approach proved the Ramanujan conjectures for modular forms (through the work of Pierre Deligne, using the Weil conjectures) and made possible the modularity theorem. It also revealed that modular forms are not isolated objects but fit into families—Hida families and eigenvarieties—that vary p-adically, leading to the theory of p-adic modular forms and the construction of p-adic L-functions.
The representation-theoretic approach, initiated by Robert Langlands in the 1960s, reframes automorphic forms as vectors in representations of adelic groups. Instead of functions on a symmetric space, one considers functions on the adelic points of a reductive group, where the arithmetic group appears as a discrete subgroup. This perspective unifies the classical theory: a single automorphic form on the adelic group corresponds to an entire family of classical forms, one for each level structure. The representation-theoretic language is essential for the Langlands program, because it allows one to decompose the space of automorphic forms into irreducible representations and to compare these representations across different groups (functoriality). This approach also connects to the local theory: at each prime, the representation factors into a local component, and the study of these local components is the theory of representations of p-adic groups, a rich subject in its own right.
The spectral and harmonic analysis approach, developed in parallel, treats automorphic forms as eigenfunctions of differential operators on symmetric spaces. The central problem here is the decomposition of the space of square-integrable automorphic forms into irreducible components, a continuous analogue of the Fourier decomposition on a circle. This theory, due largely to Atle Selberg and Israel Gelfand and their schools, produces both discrete spectra (cusp forms) and continuous spectra (Eisenstein series), and it yields the trace formula—a powerful tool that relates sums over eigenvalues of the Laplacian to sums over geometric data. The trace formula has become one of the most important technical instruments in the field, used to prove comparisons between automorphic forms on different groups and to establish cases of functoriality.
These approaches are not in competition. A single result often requires all of them: the modularity theorem, for instance, used the representation-theoretic framework to set up the comparison, the analytic theory of L-functions to establish identities, and the algebraic geometry of modular curves to construct the necessary Galois representations. The field's progress has come from the interplay of these perspectives, each revealing a different face of the same underlying structure.
The current shape of the field is dominated by the Langlands program, which has grown from a collection of conjectures into a research program with a coherent architecture. The central objects are now automorphic representations—irreducible representations of adelic groups occurring in the space of automorphic forms—and the central problem is the functoriality conjecture, which predicts that homomorphisms between Langlands dual groups should induce transfers of automorphic representations. This conjecture generalizes almost every deep result in the field, from the modularity theorem to the base change and lifting results proved in the 1980s and 1990s.
A major recent development is the proof of the fundamental lemma, a technical statement about orbital integrals that was the main obstruction to many comparisons of trace formulas. Its proof, completed by Ngô Bảo Châu in the late 2000s, has opened the way to systematic applications of the trace formula and has led to a wave of results on the stable trace formula and on endoscopy, the theory that handles the subtle discrepancies between groups that are not isomorphic but have related representations.
Another active frontier is the p-adic Langlands correspondence, which seeks to relate p-adic Galois representations to p-adic automorphic forms. This theory, still under construction, promises to unify the classical Langlands correspondence with the arithmetic of modular forms modulo primes and with the theory of p-adic families. It has already produced striking results, including new proofs of local-global compatibility and new constructions of Galois representations.
The field also continues to develop through its connections to other areas. The theory of mock modular forms and harmonic Maass forms has revealed that the classical theory of modular forms is only the holomorphic tip of a larger structure, with deep connections to partitions, indefinite quadratic forms, and the arithmetic of elliptic curves. The geometric Langlands program, which replaces number fields with function fields of curves over the complex numbers, has grown into a separate but closely related subject, with connections to conformal field theory and the geometric representation theory of affine Lie algebras.
For the educated newcomer, the field is best understood not as a collection of theorems but as a web of correspondences. The central intuition is that automorphic forms are a bridge: they live in the analytic world of functions and symmetries, but their Fourier coefficients live in the arithmetic world of integers and primes. The Langlands program is the systematic attempt to map this bridge in both directions—from automorphic forms to Galois representations and back—and the field's vitality comes from the fact that each new correspondence reveals unexpected connections between seemingly distant parts of mathematics. The subject rewards patience: its objects are intricate, its proofs are long, but its structure is remarkably coherent, and its results have repeatedly transformed the landscape of number theory.