Stochastic differential equations · Part 4 of 10
Itô's formula
Itô’s formula is the chain rule of stochastic calculus, and it is the most used tool in the subject: almost every calculation in the rest of this collection is an application of it.
The idea
For a smooth function , Taylor’s theorem gives
In ordinary calculus the second-order term would vanish in the limit. Here has expectation , the same order as the time step, and it does not average away. Summed over a partition, the second-order terms add up to , because quadratic variation turns into . The third- and higher-order terms do vanish, because is of order .
The basic formula
For a twice continuously differentiable ,
or, in differential form, . With , this recovers from the previous post. When also depends on time,
General Itô processes
For an Itô process , the same expansion, now with , gives
The term is the drift of , and the term is its noise. A trick that recurs throughout this collection is to choose so that the drift vanishes, which makes a (local) martingale.
Several dimensions
For a vector of Itô processes and a twice continuously differentiable on ,
For a -dimensional Brownian motion, , so the second-order term becomes . Two special cases are worth memorizing:
- The product rule: .
- Harmonic functions: if , then is a local martingale. This is where the connection between Brownian motion and potential theory begins.
Example: geometric Brownian motion
The Black–Scholes model of a stock price is
Apply Itô’s formula to , using , and :
Integrating both sides,
The mean grows like , but a typical path grows at the slower rate . If , almost every path tends to zero even though the mean grows without bound: the mean is carried by rare, enormous paths.
Example: exponential martingales
Taking and in the computation above shows that
so is a martingale with mean one. More generally, for an adapted process , the stochastic exponential solves . These are the densities that change the drift of a Brownian motion in Girsanov’s theorem, later in the collection.
Example: computing moments
Itô’s formula turns moment calculations into ordinary integrals. For ,
The stochastic integral has mean zero, so , the familiar fourth moment of a Gaussian. The same idea gives a recursion for every moment.
Example: the radial part of Brownian motion
Let be a -dimensional Brownian motion with , started away from the origin, and let . Away from the origin, has gradient and Laplacian , so Itô’s formula gives
The process is a one-dimensional Brownian motion, because it is a continuous local martingale with quadratic variation (this uses Lévy’s characterization, from the post on martingales). So is the Bessel process of dimension . Its drift pushes the radius outward, more strongly in higher dimensions, which is the analytic reason Brownian motion in three or more dimensions eventually wanders off to infinity.