stochastic processes

stochastic processes
Name	Stochastic process
Field	Probability theory
Notable	Andrey Kolmogorov; Norbert Wiener; Paul Lévy; Joseph Doob; William Feller
Introduced	20th century
Related	Markov process; martingale; Brownian motion; Poisson process

stochastic processes A stochastic process is a collection of random variables indexed by time or space that models systems evolving under uncertainty. Developed through work by Andrey Kolmogorov, Norbert Wiener, Paul Lévy, Joseph Doob, and William Feller, the theory unites concepts from Émile Borel-era measure theory, David Hilbert-style functional analysis, and applied problems from Bell Labs, RAND Corporation, and Harvard University. Foundations are formalized in texts associated with institutions such as Princeton University, Cambridge University, and University of Chicago.

Definition and basic concepts

A stochastic process is formally a family {X_t}_{t∈T} of random variables on a probability space (Ω, F, P) where the index set T is often time; seminal axioms trace to work at Moscow State University and University of Göttingen. Essential constructs include state space, filtration, sample path, and measurability conditions developed in seminars at Institute for Advanced Study and École Normale Supérieure. Core examples that shaped definitions include processes studied by Albert Einstein in the context of diffusion, analyses by Wiener of Gaussian trajectories, and Poisson models used in studies at Bell Labs.

Classification and types

Processes divide by index set (discrete-time vs continuous-time) and state space (discrete-state vs continuous-state), categories influenced by research at Stanford University and Massachusetts Institute of Technology. Major classes include Markov chains rooted in work by Andrey Markov and generalized by A. A. Kolmogorov; martingales formalized by Joseph Doob; Gaussian processes analyzed by Norbert Wiener and S. N. Bernstein; Lévy processes named after Paul Lévy; and point processes like the Poisson process linked to investigations at Bell Labs and Royal Society. Other important distinctions involve stationarity, ergodicity (studied in Albert Einstein-adjacent diffusion literature), and strong vs weak continuity properties explored in seminars at Courant Institute.

Key properties and theorems

Fundamental results include Kolmogorov’s existence theorem, Doob’s martingale convergence theorems, Lévy–Khintchine formula, and the strong Markov property—developments tied to lectures at Princeton University and conferences at International Congress of Mathematicians. Limit theorems such as the central limit theorem and functional central limit theorems underpin convergence of rescaled processes; invariance principles were advanced by researchers affiliated with University of Cambridge and University of Paris. Ergodic theorems with roots in George Birkhoff and John von Neumann provide long-run averages for stationary processes, while spectral representations for stationary Gaussian processes trace to work in Bell Labs and Harvard University.

Common stochastic process models

Prominent models include Brownian motion (Wiener process) central to diffusion models studied by Norbert Wiener and used in physical theories by Albert Einstein; Poisson processes used in telecommunication models at Bell Labs; birth–death processes explored in population studies at University of Oxford; autoregressive and moving-average models developed in time series literature at University of Chicago and Princeton University; and continuous-time Markov chains applied in queueing theory from AT&T-era research. Other canonical models are Ornstein–Uhlenbeck processes linked to G. E. Uhlenbeck and L. S. Ornstein, branching processes with origins in work by Galton and Watson, and jump processes characterized via Lévy measures studied by Paul Lévy.

Applications

Applications span physics, finance, biology, engineering, and beyond, with influential developments at Los Alamos National Laboratory, Goldman Sachs, CERN, World Health Organization, and NASA. In finance, stochastic calculus and models such as Black–Scholes (linked to research in University of Chicago and Princeton University) rely on diffusion processes; in epidemiology, branching and contact processes inform outbreak models used by Centers for Disease Control and Prevention and Johns Hopkins University; in telecommunications, queueing models from Bell Labs and AT&T underpin network design; and in physics, random fields and Brownian motion relate to studies at Max Planck Society and Los Alamos National Laboratory.

Simulation and estimation methods

Monte Carlo methods, developed and popularized through computational projects at Los Alamos National Laboratory and RAND Corporation, provide simulation frameworks; Markov chain Monte Carlo algorithms received impetus from work at Harvard University and University of Toronto. Parameter estimation techniques include maximum likelihood developed in statistical schools at Princeton University and Columbia University, method of moments with roots in studies at University of Cambridge, and Bayesian estimation promoted via research at Stanford University and University of California, Berkeley. Numerical schemes for stochastic differential equations, such as the Euler–Maruyama and Milstein methods, are applied in engineering research at MIT and Caltech.

Category:Probability theory