Stochastic calculus

Stochastic calculus
Name	Stochastic calculus
Field	Probability theory, Mathematical analysis
Founded by	Kiyosi Itô, Ruslan Stratonovich
Key people	Paul Lévy, Norbert Wiener, Joseph Doob
Related areas	Martingale theory, Partial differential equations, Financial mathematics

Stochastic calculus. It is a branch of mathematics that operates on stochastic processes, providing a rigorous framework for modeling systems influenced by random noise. It extends the rules of classical calculus to functions that are not smooth, most famously to the paths of the Wiener process, which are continuous but nowhere differentiable. The field is foundational for the modern theory of financial derivatives and for analyzing phenomena in statistical physics and systems biology.

Introduction

The development of this field was driven by the need to give precise meaning to integrals and differentials involving random functions, a problem that traditional Newtonian calculus could not solve. Its origins are deeply intertwined with the mathematical formalization of Brownian motion, pioneered by Norbert Wiener, and the theory of martingales advanced by Joseph Doob. The seminal work of Kiyosi Itô in the mid-20th century provided a consistent and powerful calculus, now central to quantitative finance and engineering. This framework allows for the modeling of dynamic systems where uncertainty evolves continuously over time, distinguishing it from models based on discrete-time Markov chains.

Fundamental concepts

The core objects are stochastic processes, with the Wiener process (or Brownian motion) serving as the fundamental building block for continuous noise. A key property is that its paths have infinite total variation but finite quadratic variation, a concept absent in ordinary calculus. This leads to the definition of the Itô integral, a stochastic integral constructed as a limit in mean square convergence, which crucially treats the integrand as non-anticipating with respect to the driving noise. Other essential processes include the Poisson process for modeling jumps and the broader class of Lévy processes, which generalize both continuous and discontinuous noise. The foundational theorems, such as the martingale representation theorem, link these integrals to the theory of martingales.

Itô calculus

The central result is Itô's lemma, a change-of-variable formula that is the stochastic analogue of the chain rule. For a function of an Itô process, the lemma includes an extra term involving the second derivative, arising from the non-zero quadratic variation. This formula is indispensable for solving stochastic differential equations and for pricing options in the Black–Scholes model. The calculus also defines the Stratonovich integral, an alternative formulation developed by Ruslan Stratonovich that obeys the classical chain rule but requires knowledge of the future, making it less suitable for financial modeling but natural in certain areas of physics and stochastic control theory. The comparison between the Itô integral and the Stratonovich integral is a key conceptual topic.

Stochastic differential equations

These equations, often abbreviated as SDEs, are differential equations where one or more terms are stochastic processes. A prototypical example is the equation for geometric Brownian motion, which models stock prices in the Black–Scholes–Merton framework. Solutions to SDEs are stochastic processes themselves, and their existence and uniqueness are governed by theorems analogous to those in ordinary differential equations, such as conditions by Itô and Kiyosi Itô. Important classes of solutions include diffusion processes, which are Markov processes with continuous paths described by an infinitesimal generator linked to a partial differential equation like the Fokker–Planck equation. Numerical methods for solving SDEs, such as the Euler–Maruyama method, are critical in computational finance.

Applications

Its most famous application is in mathematical finance, where it underpins the theory of arbitrage-free pricing of derivatives, as formalized in the Black–Scholes model by Fischer Black, Myron Scholes, and Robert C. Merton. It is equally vital in filtering theory, as in the Kalman filter and its nonlinear extension, the Kushner equation, used for estimating the state of dynamic systems. In the natural sciences, it models phenomena like particle diffusion in statistical mechanics, population dynamics in biology, and signal propagation in neuroscience. The field is also essential in stochastic optimal control, solving problems like the Merton's portfolio problem.

Extensions and related theories

The theory has been generalized to infinite-dimensional settings, such as stochastic partial differential equations, which are used in modeling random fields in quantum field theory and fluid dynamics. The Malliavin calculus, developed by Paul Malliavin, provides a differential calculus on the Wiener space and is a powerful tool for proving the smoothness of densities of solutions to SDEs. For processes with jumps, the calculus extends to integrals with respect to compensated Poisson random measures, forming part of the theory of semimartingales. Connections to other areas include rough path theory, developed by Terry Lyons, which provides an alternative pathwise integration theory, and links to Dirichlet forms in potential theory.

Category:Stochastic processes Category:Mathematical analysis Category:Financial mathematics