10-StochDiffEq Page 272 Wednesday, February 4, 2004 12:51 PM
272 The Mathematics of Financial Modeling and Investment Management
t t
Zt ( ω, ) = ∫ as ( ω, ) sd + ∫ bs ( ω, ) dBs ( s ω, )
0 0
Itô processes can be written in the shorter differential form as
dZt = adt + bdBt
It should be clear that the latter formula is just a shorthand for the inte-
gral definition.
THE 1-DIMENSIONAL ITÔ FORMULA
One of the most important results concerning Itô processes is a formula
established by Itô that allows one to explicitly write down an Itô process
which is a function of another Itô process. Itô’s formula is the stochastic
equivalent of the change-of-variables formula of ordinary integration.
We will proceed in two steps. First we will introduce Itô’s formula for
functions of Brownian motion and then for functions of general Itô pro-
cesses. Suppose that the function g(t,x) is twice continuously differentia-
ble in [0,∞ ) × R and that Bt is a one-dimensional Brownian motion. The
function Yt = g(t,Bt) is a stochastic process. It can be demonstrated that
the process Yt = g(t,Bt) is an Itô process of the following form
∂ g 1 ∂^2 g ∂ g
dYt = ------( tBt) + ------------( tBt) dt + ------( tBt) dBt
∂ t
,
(^2) ∂ x^2
,
∂ x
,
The above is Itô’s formula in the case the underlying process is a Brown-
ian motion. For example, let’s suppose that g(t,x) = x^2. In this case we
can write
∂ g ∂ g ∂
2
g
------= 0 , ------ = 2 x , --------- = 2
∂ t ∂ x ∂ x^2
2
Inserting the above in Itô’s formula we see that the process Bt can be
represented as the following Itô process
dYt = dt + 2BtdBt
or, explicitly in integral form