Stochastic Differential Equations — Mathematics Knowledge System

Overview

A stochastic differential equation (SDE) is a differential equation driven by a stochastic process, typically Brownian motion. The Ito SDE dX_t = mu(t, X_t) dt + sigma(t, X_t) dW_t models systems subject to random perturbations. SDEs are the continuous-time counterpart of difference equations with noise and arise in finance (Black-Scholes), physics (Langevin equation), and biology (population dynamics). Key results include existence and uniqueness of strong solutions under Lipschitz conditions, Ito's lemma for changing variables, and the Feynman-Kac formula connecting SDEs to PDEs.

Intuition

An SDE dX_t = mu dt + sigma dW_t says: over a tiny time dt, X changes by a deterministic drift mu*dt plus a random noise sigma*dW_t. The noise term is proportional to sqrt(dt) (since Var(W(dt)) = dt), making it much larger than the drift for very small dt -- this is why stochastic systems differ qualitatively from deterministic ones. Ito's lemma is the stochastic chain rule: for f(X_t), the usual df = f'dX has an extra term f''(dX)^2/2 = f''sigma^2 dt/2, the Ito correction from the quadratic variation of Brownian motion.

Formal Definition

Definition

An Ito SDE for a process X_t on [0, T] with drift mu and diffusion coefficient sigma is defined in integral form.

dX_t = \mu(t, X_t)\,dt + \sigma(t, X_t)\,dW_t

Ito SDE (differential form)

X_t = X_0 + \int_0^t \mu(s,X_s)\,ds + \int_0^t \sigma(s,X_s)\,dW_s

Integral form (well-defined via Ito integral)

df(t,X_t) = \frac{\partial f}{\partial t}dt + \frac{\partial f}{\partial x}dX_t + \frac{1}{2}\frac{\partial^2 f}{\partial x^2}\sigma^2\,dt

Ito's lemma (stochastic chain rule)

dS_t = \mu S_t\,dt + \sigma S_t\,dW_t \implies S_t = S_0\exp\!\left((\mu-\tfrac{\sigma^2}{2})t + \sigma W_t\right)

Geometric Brownian motion (GBM)

Notation

Notation	Meaning
$dW_t$	Ito differential of Brownian motion; formal notation for white noise
$\mu(t,x)$	Drift coefficient of the SDE
$\sigma(t,x)$	Diffusion coefficient of the SDE
$\mathcal{L}$	Generator of the diffusion: Lf = mu f' + (sigma^2/2) f''

Theorems

Theorem 1: Existence and Uniqueness of Strong Solutions

\text{If } \mu \text{ and } \sigma \text{ satisfy the Lipschitz and linear growth conditions:} \quad |\mu(t,x)-\mu(t,y)|+|\sigma(t,x)-\sigma(t,y)| \le L|x-y|, \quad |\mu(t,x)|+|\sigma(t,x)| \le C(1+|x|), \text{ then for any } X_0 \in L^2 \text{ the SDE has a unique strong solution.}

Theorem 2: Ito's Lemma

\text{Let } X_t \text{ satisfy } dX_t = \mu_t\,dt + \sigma_t\,dW_t \text{ and } f \in C^{1,2}([0,T]\times\mathbb{R}). \text{ Then:} \quad df(t,X_t) = \!\left(f_t + \mu_t f_x + \tfrac{\sigma_t^2}{2} f_{xx}\right)\!dt + \sigma_t f_x\,dW_t.

Theorem 3: Feynman-Kac Formula

\text{The solution to the PDE } u_t + \mu u_x + \tfrac{\sigma^2}{2}u_{xx} = 0 \text{ with terminal condition } u(T,x)=g(x) \text{ is given by:} \quad u(t,x) = \mathbb{E}[g(X_T) \mid X_t = x], \text{ where } X \text{ solves } dX_s = \mu(s,X_s)ds + \sigma(s,X_s)dW_s.

Worked Examples

1
Apply the integrating factor e^{alpha t} and use Ito's lemma on f(t, X_t) = e^{alpha t} X_t.
$d(e^{\alpha t} X_t) = \alpha e^{\alpha t} X_t\,dt + e^{\alpha t}\,dX_t = e^{\alpha t}\sigma\,dW_t$
2
Integrate from 0 to t.
$e^{\alpha t} X_t = X_0 + \sigma \int_0^t e^{\alpha s}\,dW_s$
3
Solve for X_t.
$X_t = X_0 e^{-\alpha t} + \sigma \int_0^t e^{-\alpha(t-s)}\,dW_s$

✓ Answer

X_t = X_0 e^{-alpha t} + sigma * integral_0^t e^{-alpha(t-s)} dW_s. This is a Gaussian process with mean X_0 e^{-alpha t} and variance sigma^2(1-e^{-2 alpha t})/(2 alpha).

Practice Problems

Mediumapplication

Use Ito's lemma to find the SDE satisfied by Y_t = ln(S_t) where dS_t = mu S_t dt + sigma S_t dW_t.

Hardfree response

Explain the Ito-Stratonovich conversion: if dX_t = a(X_t) dW_t in the Ito sense, what is the equivalent Stratonovich SDE?

Common Mistakes

Common Mistake

The classical chain rule applies to functions of Brownian motion

Ito's lemma adds the extra (1/2) f'' sigma^2 dt term due to the quadratic variation of Brownian motion; omitting this leads to incorrect results.

Common Mistake

Geometric Brownian motion satisfies S_t = S_0 exp(mu t + sigma W_t)

The correct formula is S_t = S_0 exp((mu - sigma^2/2)t + sigma W_t); the -sigma^2/2 correction comes from Ito's lemma applied to ln(S_t).

Quiz

Historical Background

Kiyosi Ito introduced the stochastic integral and the Ito calculus in 1944, providing the mathematical framework for SDEs. His central tool, Ito's lemma (the stochastic chain rule), corrects the classical chain rule to account for the quadratic variation of Brownian motion. The theory was applied to finance by Merton and Black-Scholes in 1973, who derived the option pricing formula via geometric Brownian motion. Stratonovich later proposed an alternative stochastic integral that preserves the classical chain rule, leading to the Ito-Stratonovich debate.

1944
Ito introduces the stochastic integral and Ito's lemma
Kiyosi Ito
1966
Stratonovich proposes the Stratonovich stochastic integral
Ruslan Stratonovich
1973
Black-Scholes derive option pricing formula via geometric Brownian motion SDE
Fischer Black, Myron Scholes, Robert Merton

Summary

An Ito SDE dX_t = mu(t,X_t)dt + sigma(t,X_t)dW_t models drift plus random perturbations; existence and uniqueness hold under Lipschitz and growth conditions.
Ito's lemma: df(t,X_t) = (f_t + mu f_x + (sigma^2/2) f_{xx})dt + sigma f_x dW_t; the extra (sigma^2/2)f_{xx} term is the Ito correction.
Geometric Brownian motion is the solution to dS = mu S dt + sigma S dW: S_t = S_0 exp((mu - sigma^2/2)t + sigma W_t).
The Feynman-Kac formula connects solutions of parabolic PDEs to expected values of diffusion functionals.

References

BookKaratzas, I. and Shreve, S. -- Brownian Motion and Stochastic Calculus, Chapters 3-5
BookOksendal, B. -- Stochastic Differential Equations
WebsiteWikipedia -- Stochastic differential equation