Loading field map
From the early nineteenth century onward, mathematicians who worked with complex numbers faced a persistent tension: what was the most fundamental way to define and study analytic functions? Should one start from differential equations, from geometric surfaces, or from convergent power series? Each choice led to a different vision of the subject, and the history of complex analysis is largely the story of how these three classical frameworks first competed, then converged for one-variable theory, and finally inspired a series of modern extensions that continue to shape research today.
The first systematic framework for complex analysis was built around the Cauchy–Riemann equations. Augustin-Louis Cauchy and Bernhard Riemann, working in the 1820s–1850s, treated analyticity as a local differential condition: a complex function is analytic precisely when its real and imaginary parts satisfy a pair of partial differential equations. This framework made complex analysis a branch of differential equation theory. It gave powerful tools for evaluating real integrals via contour integration and for understanding harmonic functions, but it said little about the global shape of the function's domain.
Riemann himself soon pushed beyond the differential-equation starting point. In his 1851 doctoral dissertation, he introduced a geometric framework that treated analytic functions as conformal (angle-preserving) maps between surfaces. For Riemann, the key object was not the function alone but the Riemann surface—a multi-sheeted domain on which a multivalued function becomes single-valued. This geometric vision connected complex analysis to topology and to the theory of algebraic curves. It allowed Riemann to prove the mapping theorem that any simply connected proper open subset of the complex plane is conformally equivalent to the unit disk, a result that the differential-equation framework could not easily reach.
At nearly the same time, Karl Weierstrass developed a third framework that began from convergent power series. For Weierstrass, an analytic function was defined by a power series together with all its analytic continuations. This approach emphasized the algebraic and combinatorial properties of series expansions. It made the theory of singularities, especially essential singularities and the classification of entire functions, particularly transparent. Weierstrass's framework also provided the first rigorous foundation for complex analysis, avoiding the geometric intuition that Riemann had relied on and the formal manipulation of differentials that Cauchy had sometimes taken for granted.
By the 1880s, mathematicians recognized that these three frameworks were equivalent for functions of one complex variable. A function that satisfies the Cauchy–Riemann equations locally can be expanded in a convergent power series; a function defined by a power series can be continued to a Riemann surface; and a conformal map between surfaces satisfies the Cauchy–Riemann equations. The equivalence was a triumph, but it did not erase the differences in emphasis. The Cauchy–Riemann framework remained the most efficient for local calculations and for integral formulas. The Riemannian geometric framework became the natural language for global problems, especially uniformization and moduli. The Weierstrassian power-series approach continued to dominate the study of entire and meromorphic functions, where series expansions are the most direct description. For the rest of the nineteenth century, practitioners moved freely among the three viewpoints, choosing whichever best fit the problem at hand.
When mathematicians turned to functions of more than one complex variable around 1900, the classical triangle of frameworks proved insufficient. The Cauchy–Riemann equations generalize naturally to several variables, but the resulting system of partial differential equations is overdetermined: a function of several complex variables must satisfy a Cauchy–Riemann equation for each variable. This overdeterminacy forces strong rigidity—for instance, a function that is analytic in a connected domain and vanishes on an open subset must vanish everywhere. The geometric framework of Riemann surfaces also does not extend straightforwardly, because the zero sets of analytic functions in several variables are no longer isolated points but complex hypersurfaces. The power-series approach remains valid locally, but analytic continuation in several variables behaves very differently: the natural domains of holomorphy are not arbitrary open sets but pseudoconvex domains, and the boundary geometry of these domains became a central object of study.
The framework of several complex variables therefore required entirely new tools. In the 1930s–1950s, Henri Cartan, Kiyoshi Oka, and others imported sheaf theory and cohomology from algebraic topology to handle the global obstructions that arise when trying to construct analytic functions with prescribed zeros or poles. The Cousin problems, which ask whether local meromorphic data can be patched into a global meromorphic function, turned out to be solvable precisely when certain cohomology groups vanish. This sheaf-theoretic approach gave several complex variables a distinctive character that set it apart from one-variable theory. Today, several complex variables remains an active research area, closely linked to algebraic geometry and complex differential geometry, and it continues to develop methods that the classical frameworks could not have supplied.
In the 1920s, Rolf Nevanlinna introduced a framework that shifted the focus of complex analysis from isolated exceptional values to the asymptotic distribution of values taken by an entire or meromorphic function. Earlier work by Émile Picard and others had shown that a nonconstant entire function omits at most one finite value (Picard's little theorem), but these results said nothing about how often the function attains each value. Nevanlinna's theory replaced the study of exceptional points with a quantitative apparatus: the characteristic function T(r,f) measures the growth of the function, the proximity function m(r,a) measures how close the function comes to a value a on a large circle, and the counting function N(r,a) counts how many times the function actually attains a. The first main theorem states that T(r,f) = m(r,a) + N(r,a) + O(1), so that the growth of the function is distributed between near-misses and actual hits. The defect relation then bounds the total deficiency of all values, generalizing Picard's theorem in a precise quantitative way.
This framework represented a genuine methodological break. Instead of asking which values are exceptional, Nevanlinna asked how the function's values are distributed on average as the radius grows. The theory proved remarkably robust: it extended to holomorphic curves in projective space and to several complex variables, and it remains a living tradition today, with applications to Diophantine approximation and complex dynamics. Nevanlinna theory coexists with the classical frameworks, providing a statistical complement to the pointwise and geometric approaches that preceded it.
Complex dynamics, the study of iterated analytic functions, began around 1910 with the independent work of Pierre Fatou and Gaston Julia. They used the classical frameworks—especially the Weierstrassian theory of normal families and the Riemannian geometry of the Riemann sphere—to classify the behavior of rational maps under iteration. The Julia set, where the dynamics is chaotic, and the Fatou set, where the dynamics is stable, became the central objects. Fatou and Julia established the basic properties of these sets, but the subject then entered a long period of relative quiet.
The revival of complex dynamics in the 1980s was driven by two forces: computer graphics that revealed the stunning fractal structure of Julia sets and the Mandelbrot set, and the injection of quasiconformal mapping techniques. Quasiconformal mappings, which allow bounded distortion of angles rather than preserving them exactly, gave a flexible tool for deforming dynamical systems while preserving their essential properties. Adrien Douady and John Hubbard used quasiconformal surgery to prove the connectivity of the Mandelbrot set and to classify the dynamics of quadratic polynomials. This feedback loop—quasiconformal methods enabling new dynamical results, and dynamical problems driving the development of quasiconformal theory—transformed both fields. Today, complex dynamics is a thriving area that draws on the geometric and power-series traditions while adding its own distinctive questions about iteration, renormalization, and the structure of parameter spaces.
Quasiconformal mappings emerged in the 1930s from the work of Herbert Grötzsch and Lars Ahlfors. A quasiconformal map distorts angles by at most a bounded factor, unlike a conformal map which preserves angles exactly. This relaxation made it possible to study deformations of Riemann surfaces that are not conformally equivalent but are still geometrically controlled. The Teichmüller space of a Riemann surface—the space of all complex structures on the surface up to a natural equivalence—became the central object. Oswald Teichmüller in the 1940s proved that each point in Teichmüller space corresponds to a unique extremal quasiconformal map, establishing a deep link between the distortion of mappings and the geometry of moduli.
Quasiconformal mappings and Teichmüller theory have since become a major framework in their own right, with connections to hyperbolic geometry, mapping class groups, and low-dimensional topology. They also fed back into complex dynamics, as noted above, and into several complex variables through the study of holomorphic motions. The framework remains active today, with researchers using quasiconformal methods to study the boundary of Teichmüller space, the geometry of Kleinian groups, and the dynamics of rational maps.
Beginning around 1950, a new subarea emerged at the intersection of complex analysis and operator theory. The central idea was to use complex-variable techniques to study linear operators on Hilbert spaces, and conversely to use operator-theoretic methods to solve problems in complex analysis. The Hardy space H², consisting of analytic functions on the unit disk with square-integrable boundary values, became a Hilbert space whose structure could be analyzed via complex analysis. Toeplitz operators, which compress multiplication by a bounded analytic function to the Hardy space, linked function theory to spectral theory and index theory. Subnormal operators, which extend to normal operators on a larger Hilbert space, turned out to be intimately connected to the theory of analytic functions and to the geometry of the unit disk.
This operator-theory subarea shares infrastructure with the Hilbert-space frameworks of functional analysis, but it retains a distinctively complex-analytic character. It has applications to mathematical physics, signal processing, and control theory. The journal Complex Analysis and Operator Theory, founded in 2007, reflects the continued vitality of this cross-fertilization. Today, researchers in this area work on problems ranging from the structure of Toeplitz algebras to the function-theoretic properties of reproducing kernels.
Among the frameworks that remain active today, several complex variables, complex dynamics, Nevanlinna theory, quasiconformal mappings and Teichmüller theory, and complex analysis and operator theory all have substantial research communities. They agree on the foundational equivalence of the three classical frameworks for one-variable theory, and they all draw on the Cauchy–Riemann, geometric, and power-series traditions as needed. Where they disagree is on method and emphasis. Several complex variables relies heavily on sheaf theory and partial differential equations, while complex dynamics uses quasiconformal geometry and computer-assisted experimentation. Nevanlinna theory is primarily asymptotic and statistical, whereas quasiconformal theory is geometric and deformational. The operator-theory subarea bridges complex analysis with functional analysis, a connection that the other frameworks do not emphasize. This division of labor is productive: each framework addresses questions that the others cannot easily reach, and the cross-fertilization between them—especially between quasiconformal theory and dynamics—continues to generate new results.