Wiener measure

Wiener measure
Name	Wiener measure
Subject	Probability theory
Introduced	1923
Creator	Norbert Wiener
Area	Stochastic processes

Wiener measure is a probability measure on a space of real-valued continuous functions that formalizes the distribution of classical Brownian motion. It provides the mathematical foundation for path-space analysis in probability theory, connecting functional analysis, measure theory, and partial differential equations. Wiener measure underlies many developments in modern mathematics and physics, linking figures and institutions across the 20th century.

Definition

Wiener measure is the Borel probability measure on the space C([0,T]; R) of continuous functions on a compact interval that assigns to cylinder sets probabilities determined by Gaussian transition densities; it is the unique measure making coordinate maps into a Gaussian process with stationary independent increments, mean zero, and variance equal to elapsed time. The construction is often given in terms of finite-dimensional distributions specified by the heat kernel of the Laplacian on R and verified via Kolmogorov's extension theorem and Prokhorov's criterion. Historical development ties to Norbert Wiener, Andrey Kolmogorov, Paul Lévy, André Weil, and contemporaneous institutions like Massachusetts Institute of Technology and Harvard University where key work on stochastic processes matured.

Construction

The standard construction begins by specifying for each finite set of times 0 = t0 < t1 < ... < tn the multivariate normal law with covariance matrix Cov(Bti, Btj) = min(ti, tj) and then invoking the Kolmogorov extension theorem to obtain a probability measure on R^[0,T]. Tightness arguments based on moment estimates and the Arzelà–Ascoli theorem are used to show support concentrates on C([0,T]; R). The analytical connection uses the fundamental solution to the heat equation associated with the Laplace operator and semigroup methods developed by Eugene Dynkin, Kurt Friedrichs, and researchers at the Institute for Advanced Study. Alternate constructions use Gaussian measures on Banach spaces as in work by Leonard Gross and measure-theoretic techniques traceable to Andrey Kolmogorov and Stefan Banach.

Properties

Wiener measure endows path space with properties reflecting Gaussianity, time-reversal symmetry, and Markovian structure: almost surely paths are continuous but nowhere differentiable, have quadratic variation equal to elapsed time, and satisfy the strong Markov property. Its Cameron–Martin theorem characterizes shifts by absolutely continuous paths and links to the Cameron–Martin space, while the Girsanov theorem describes change-of-measure transformations producing drifted processes; these results build on work by Norman Wiener, Cameron, Martin D. D. White, Igor Girsanov, and later formalizations at Princeton University and University of Cambridge. Connections to spectral theory appear via the Feynman–Kac formula relating Wiener integrals to Schrödinger semigroups investigated by Mark Kac and to trace class operators studied by John von Neumann and Frigyes Riesz.

Relation to Brownian motion

Under Wiener measure, the coordinate process is canonical Brownian motion (also called standard Wiener process), providing a rigorous model of the empirical observations of Robert Brown and theoretical formulations by Albert Einstein and Norbert Wiener. The measure encodes transition probabilities equal to Gaussian kernels appearing in the diffusion equation and furnishes pathwise properties exploited in stochastic calculus initiated by Kiyoshi Itō, Paul Lévy, and refined in the Itô–Stratonovich framework influential at Imperial College London and ETH Zurich. Martingale characterizations of Brownian motion and Lévy's characterization relate to the Lévy–Khintchine representation and to classical results by Joseph Doob and William Feller.

Applications

Wiener measure is foundational in mathematical physics, financial mathematics, and analysis on path space: it is used to define path integrals in Euclidean quantum field theory as in work by Richard Feynman (via analytic continuation), to price derivative securities in models developed at Bloomberg, and to study heat kernels and index theorems connected to Atiyah–Singer index theorem research groups. In engineering, stochastic filtering and control theory employ Girsanov and Cameron–Martin results central to methods advanced at Bell Labs and Stanford University. Statistical physics, polymer theory, and random media exploit Wiener–measure-based models investigated by Ludwig Boltzmann-inspired programs and contemporary research at Los Alamos National Laboratory.

Generalizations and variants

Generalizations include Gaussian measures on infinite-dimensional Banach and Hilbert spaces (Cameron–Martin theory), conditioned Wiener measures such as Brownian bridge measures linked to boundary value problems studied by Mikhail Smirnov-style analysts, and path measures for Lévy processes and fractional Brownian motion developed by Benoît Mandelbrot and Paul Lévy. Other variants arise in rough path theory pioneered by Terry Lyons and Malliavin calculus introduced by Paul Malliavin, both extending differential and integral structure on Wiener space; connections to stochastic partial differential equations tie to work by Robert Dalang and research groups at Courant Institute of Mathematical Sciences.

Category:Probability theory