Stochastic process

Lore & Background

Stochastic processes are defined as collections of random variables indexed by a set, often interpreted as time. The index set can be discrete (e.g., integers) or continuous (e.g., an interval of the real line), leading to discrete-time or continuous-time processes. The state space—the set of possible values—can be integers, the real line, or higher-dimensional Euclidean space, among others. An increment is the change between two index values, and a single outcome is called a sample function or realization. Notably, the term 'stochastic process' appeared in English earlier than the early 1930s, for instance in a 1931 paper by R. A. Fisher, with translations of Khinchin's work following later.

Reader's Guide

Two classic examples are the Wiener process (Brownian motion), used by Louis Bachelier to model price changes on the Paris Bourse, and the Poisson process, used by A. K. Erlang to model phone call counts. These processes were invented repeatedly and independently in different settings and countries. Stochastic processes are classified by state space, index set, and dependence among variables, and include categories such as random walks, martingales, Markov processes, Lévy processes, Gaussian processes, random fields, renewal processes, and branching processes. The study of stochastic processes draws on probability, calculus, linear algebra, set theory, topology, and branches of analysis like measure theory and Fourier analysis. The term 'stochastic process' first appeared in English in a 1934 paper by Joseph Doob, citing Aleksandr Khinchin's German term 'stochastischer Prozeß' from the same year, though the German term had been used earlier by Andrei Kolmogorov in 1931. The word 'stochastic' derives from Greek meaning 'to aim at a mark, guess', and was used by Jakob Bernoulli in 1713.

Did You Know?

• The Wiener process and Poisson process are historically credited to specific individuals (Norbert Wiener and Siméon Denis Poisson), without widely accepted evidence of repeated independent invention of these exact named
• The term 'stochastic process' was used in English as early as 1931 by R. A. Fisher, with translations of Khinchin's work appearing later.
• The word 'stochastic' stems from a Greek word meaning 'to aim at a mark, guess'.

Etymology and the Birth of a Term

The word 'stochastic' traces back to a Greek root meaning 'to aim at a mark' or 'guess.' In English, it first appeared as an adjective meaning 'pertaining to conjecturing,' with the Oxford English Dictionary pinpointing 1662 as its earliest recorded use. The mathematical flavor deepened when Jakob Bernoulli, in his 1713 Latin treatise Ars Conjectandi, paired the phrase with 'Stochastice,' effectively coining a label for the art of conjecturing. Decades later, in 1917, the German statistician Ladislaus Bortkiewicz wrote the word Stochastik with a specifically random connotation, explicitly referencing Bernoulli's legacy. The full compound 'stochastic process' entered English-language literature in 1934 through a paper by Joseph L. Doob, who credited a contemporaneous German paper by Aleksandr Khinchin for the term stochastischer Prozeß. Yet the German phrasing had actually been used a few years earlier, in 1931, by Andrey Kolmogorov, making the etymological trail a layered, multilingual story spanning nearly three centuries.

The Heroic Period and Mathematical Architecture

The 1930s produced what the statistician Harald Cramér later called the 'heroic period of mathematical probability theory.' During that decade, Aleksandr Khinchin formulated the first formal mathematical definition of a stochastic process as a family of random variables indexed along the real line. He was joined by a remarkable cohort of thinkers—Andrey Kolmogorov, Joseph Doob, William Feller, Maurice Fréchet, Paul Lévy, Wolfgang Doeblin, and Cramér himself—who collectively laid the theoretical groundwork that still underpins the field today. The theory of stochastic processes remains a vital branch of probability theory and an active area of research. Beyond the core definition, the stochastic label extends to related mathematical objects: a stochastic matrix encodes the transition structure of a Markov process, while stochastic calculus builds differential equations and integrals around processes such as the Wiener process, more commonly known as Brownian motion. These tools have become indispensable across pure and applied mathematics.

Monte Carlo and the Physics Revolution

The Monte Carlo method, a stochastic simulation technique popularized by Stanisław Ulam, Enrico Fermi, John von Neumann, and Nicholas Metropolis, drew its name from the repetitive, chance-driven games of a casino. Fermi's pioneering use in the 1930s—applying random sampling to compute properties of the newly discovered neutron—marked one of the earliest celebrated applications. The method became central to the Manhattan Project's simulations, though the era's limited computational tools constrained its scope. Only after electronic computers emerged from 1945 onward could Monte Carlo techniques be explored in depth. In the 1950s, Los Alamos harnessed them for early hydrogen-bomb research, and the RAND Corporation together with the U.S. Air Force played major roles in funding and spreading the methodology across physics, physical chemistry, and operations research. A practical side effect was the demand for vast quantities of random numbers, which in turn drove the development of pseudorandom number generators, replacing the slower tables of random digits that had previously served statistical sampling.

From Pollen Grains to Creative Sparks

Stochastic thinking reaches far beyond the mathematics classroom. In biology, the botanist Robert Brown first observed what we now call Brownian motion while examining pollen grains suspended in water through a microscope—one of the simplest continuous-time stochastic processes. In medicine, a technique called stochastic resonance deliberately introduces random 'noise' into vestibular feedback loops, and it has shown promise in helping diabetic and stroke patients regain balance control. At the molecular level, gene expression carries a stochastic signature: the random collisions of molecules, including RNA polymerase binding and unbinding at gene promoters, produce bursts of transcription and cell-to-cell variability that follow Brownian-motion dynamics. In computer science, stochastic ray tracing applies Monte Carlo sampling to 3D graphics, while stochastic programs power simulated annealing, genetic algorithms, and neural networks in artificial intelligence. Even in the humanities, the psychologist Dean Simonton argued in 2003 that scientific creativity itself behaves as a constrained stochastic process, suggesting that new theories across all disciplines emerge partly through random combinatorial steps.