Applied Mathematics

Applied mathematics is the part of mathematics that takes its problems from the world—from physics and engineering, from biology and medicine, from economics and industry—and builds methods to solve them.

16 frameworks, from 1687 to today · Applied Mathematics · Mathematics

still a working view ended · each leads to its entry below

Overview

Applied mathematics is the part of mathematics that takes its problems from the world—from physics and engineering, from biology and medicine, from economics and industry—and builds methods to solve them1. What distinguishes it from pure mathematics is not the difficulty of the mathematics but the source of the questions and the standard of success: an applied result must work, whether that means predicting a planet's orbit, keeping a bridge standing, routing a delivery fleet, or quantifying how much a computer model can be trusted. The field's history is therefore a history of borrowings and inventions, in which tools first made for one problem (heat flow, artillery, telephone traffic) turn out to solve many others. This article follows that history through its major turning points.

From celestial mechanics to the heat equation

Applied mathematics in the modern sense begins in the seventeenth and eighteenth centuries, when the new calculus was turned on nature. Isaac Newton's Principia (1687) and his work on fluxions gave motion a mathematical form; the inverse-square law of gravitation was itself an applied result, and the prediction of celestial positions became the first great computational industry of the field. The Bernoullis, Euler, and Lagrange developed the calculus of variations and the machinery of differential equations largely to handle mechanics—the vibrating string, the rotating body, the fluid in motion.

The decisive turn came with Joseph Fourier's study of heat, presented to the Paris Academy in 1807 and published in his Théorie analytique de la chaleur (1822). Fourier claimed that an arbitrary function could be represented as a sum of sines and cosines—a claim his contemporaries, including Lagrange, found hard to accept—and used this to solve the heat equation, a partial differential equation (an equation relating a quantity to its rates of change in space and time) describing how temperature spreads through a body. The idea that a complicated signal could be decomposed into simple waves became one of the most portable tools in all of applied mathematics: Fourier analysis now underlies signal processing, image compression, and the numerical solution of countless equations2. The same era produced the boundary-value problem as a standard form—find a function satisfying a differential equation inside a region and prescribed conditions on its edge—and the techniques (separation of variables, Green's functions, eigenfunction expansions) for solving it3.

Probability finds its applications

Probability began as a mathematics of games of chance in the seventeenth century, but its transformation into an applied science came through the study of populations and errors4. Laplace's Théorie analytique des probabilités (1812) systematized the subject and applied it to astronomy and demography; Gauss's work on the method of least squares gave observational science a way to fit models to noisy data. In the late nineteenth century, statistical physics gave probability a physical foundation: Maxwell and Boltzmann derived the behavior of gases from probabilistic assumptions about molecules, and Gibbs developed statistical mechanics as a general framework.

The twentieth century made probability a tool of engineering and management. A. K. Erlang's studies of telephone traffic at the Copenhagen telephone company around 1917 introduced the queuing model—a mathematical description of waiting lines—and with it the Poisson process, a model of random events occurring independently at a constant average rate. Andrey Kolmogorov's Grundbegriffe der Wahrscheinlichkeitsrechnung (1933) put probability on axiomatic foundations, making it a branch of measure theory and so a rigorous partner to analysis. From there, stochastic processes (families of random variables indexed by time) spread into physics (Brownian motion, described by Einstein in 1905 and given rigorous form by Norbert Wiener in the 1920s), biology (population genetics, branching processes), and finance (Bachelier's 1900 thesis on speculation, later revived and extended).

Operations research and the war-born methods

The Second World War created a new kind of applied mathematics: the systematic allocation of scarce resources5. British scientists formed teams to study radar deployment, convoy sizing, and antisubmarine tactics; the work was named "operational research" and continued after the war in industry and government. The mathematical core was optimization—finding the best choice among alternatives—and the key breakthrough was linear programming, the optimization of a linear objective subject to linear constraints6. George Dantzig formulated the simplex method in 1947, an algorithm that walks along the edges of a feasible region to the optimum, and it made large-scale planning problems tractable6. John von Neumann and Oskar Morgenstern's Theory of Games and Economic Behavior (1944) supplied a companion framework for situations of strategic interaction, and game theory grew into a major applied discipline, later extended to cooperative games (Shapley value, 1953), evolutionary games (Maynard Smith, 1970s), and the analysis of auctions and markets.

Control theory developed in parallel. The problem of steering a system—a ship, an aircraft, a chemical plant—toward a desired state was formalized by Norbert Wiener's Cybernetics (1948) and by the frequency-domain methods of Harold Black, Harry Nyquist, and Hendrik Bode in the 1930s. The state-space approach of the 1960s, with Rudolf Kalman's filter (1960) for estimating a system's state from noisy measurements, made control a general mathematical theory applicable far beyond engineering.

Numerical analysis and the computer

Approximate computation is as old as applied mathematics, but the electronic computer transformed it. Early numerical work—Newton's method for roots, Euler's method for differential equations, Gaussian elimination for linear systems—was hand calculation7. The wartime and postwar machines made it possible to solve problems no one could solve by hand, and a discipline grew up around the question of how much to trust the answer. The finite element method, developed in the 1950s and 1960s by engineers including Ray Clough and John Argyris and given mathematical foundations by Gilbert Strang and others, divides a complicated region into small simple pieces and assembles a global solution from local ones; it is now the standard tool for stress analysis, heat transfer, and structural design. The fast Fourier transform, popularized by James Cooley and John Tukey in 1965, reduced the cost of Fourier analysis from roughly the square of the problem size to roughly its logarithm, making spectral methods practical8. Stability and convergence theory—the study of when a numerical scheme approaches the true answer and how fast—became a field in its own right, and scientific computing emerged as the practice of combining numerical methods, algorithms, and hardware to simulate physical systems.

Dynamical systems and the limits of prediction

The study of how systems evolve over time took a dramatic turn with Henri Poincaré's work on the three-body problem in the 1880s, which showed that even simple gravitational systems could behave in extraordinarily complicated ways910. The mid-twentieth century brought the theory of dynamical systems into its own: Andronov and Pontryagin's work on structural stability in the 1930s, Stephen Smale's horseshoe map (1960s), and the discovery of chaos—sensitive dependence on initial conditions—by Edward Lorenz in 1963, when a truncated weather simulation diverged wildly from a nearly identical run. Chaos theory changed what applied mathematicians believed prediction could achieve and fed into turbulence, climate modeling, and the study of nonlinear oscillations. The same decades saw the growth of bifurcation theory and the geometric theory of differential equations.

Uncertainty, data, and the modern synthesis

The late twentieth century added two large concerns. The first was uncertainty quantification: real models have uncertain inputs, and the question of how input uncertainty propagates to outputs became a discipline of its own, drawing on Monte Carlo methods (developed at Los Alamos in the 1940s by Stanislaw Ulam and John von Neumann), polynomial chaos expansions (Norbert Wiener, 1938; revived by Ghanem and Spanos in 1991), and Bayesian inference. The second was the data revolution. Statistical learning—the theory and practice of fitting models to data—grew from classical statistics and pattern recognition into a major applied field, with support vector machines (Cortes and Vapnik, 1995), neural networks, and the statistical theory of generalization. The availability of massive datasets and cheap computation in the 2000s and 2010s pushed applied mathematics toward data-driven methods: sparse recovery and compressed sensing (Candès, Romberg, and Tao, 2006), which showed that signals could be reconstructed from far fewer measurements than classical theory demanded, and the modern theory of deep learning, which remains an active frontier.

The present

Applied mathematics today is a federation of methods rather than a single program. Numerical analysis and scientific computing remain central, now coupled to high-performance computing and machine learning. Optimization has expanded from linear programming to convex optimization, which has efficient general algorithms, and to stochastic and derivative-free methods for problems where gradients are unavailable. Probability and statistics supply the language of uncertainty, and stochastic processes model everything from financial markets to epidemic spread11. Control theory has merged with learning and robotics. Uncertainty quantification and data-driven modeling are among the fastest-growing areas, driven by the need to trust complex simulations and to extract structure from data. The open questions are familiar in kind: how to certify that a numerical or learned model is reliable, how to scale methods to problems of enormous dimension, and how to combine physical insight with data in a principled way. The field's history suggests that its next tools will again come from problems first posed elsewhere.

References

  1. Applied mathematics, Wikipedia
  2. Fourier analysis, Wikipedia
  3. Boundary value problem, Wikipedia
  4. Probability theory (History of probability), Wikipedia
  5. Operations research, Wikipedia
  6. Mathematical optimization (introduction, History), Wikipedia
  7. Numerical analysis (History), Wikipedia
  8. Fast Fourier transform (introduction, History), Wikipedia
  9. Dynamical system (History), Wikipedia
  10. Chaos theory (History), Wikipedia
  11. Stochastic process, Wikipedia

The frameworks of Applied Mathematics

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1687-1900
  2. 1807-Present

    Fourier analysis

    How can any function be decomposed into simple waves?

    Fly to its tower in Nousfarer

  3. 1812-Present

    Probability and Stochastic Processes

    How can randomness be modeled and predicted?

    Fly to its tower in Nousfarer

  4. 1820-Present

    Mathematical Physics

    What mathematical structures underlie physical theories?

    Fly to its tower in Nousfarer

  5. 1880-Present

    Dynamical systems theory

    How do systems evolve over time and become chaotic?

    Fly to its tower in Nousfarer

  6. 1900-Present
  7. 1930-Present

    Control Theory

    How can we steer a system to a desired state?

    Fly to its tower in Nousfarer

  8. 1940-Present

    Numerical analysis

    How can we compute reliable approximate solutions?

    Fly to its tower in Nousfarer

  9. 1940-Present

    Operations research

    How can we best allocate scarce resources?

    Fly to its tower in Nousfarer

  10. 1944-Present

    Game Theory

    How do rational agents interact strategically?

    Fly to its tower in Nousfarer

  11. 1947-Present

    Optimization

    How can we find the best solution among many?

    Fly to its tower in Nousfarer

  12. 1948-Present

    Control Theory

    How can we design feedback systems that are stable?

    Fly to its tower in Nousfarer

  13. 1955-Present

    Finite Element Method

    How can we solve equations on complex geometries?

    Fly to its tower in Nousfarer

  14. 1960-Present

    Scientific Computing

    How can computers simulate physical systems?

    Fly to its tower in Nousfarer

  15. 1990-Present
  16. 1990-Present

    Uncertainty Quantification

    How can we quantify uncertainty in predictions?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Classical Applied Analysis and Mathem…Fourier analysisProbability and Stochastic ProcessesMathematical PhysicsDynamical systems theoryProbability and Statistics for Applic…Control TheoryNumerical analysisOperations researchGame TheoryOptimizationControl TheoryFinite Element MethodScientific ComputingStatistical Learning and Data-Driven …Uncertainty Quantification

Beyond the field

Where Applied Mathematics's ideas came from in other fields, and what they led to there: the strongest 12 of 308 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Applied Mathematics is a planet in Nousfarer

Its city has a tower for each of these 16 frameworks. Land, walk in, and climb one concept at a time.

Land on Applied Mathematics

Related fields

Fields its ideas are linked to

More in Applied Mathematics

See where Applied Mathematics sits on the Map of Human Knowledge