You are not logged in.

- Topics: Active | Unanswered

Pages: **1**

**George,Y****Member**- Registered: 2006-03-12
- Posts: 1,306

However something needs to be clarified about *B[sub]t[/sub]* and *dc*

B[sub]t[/sub] follows a Brownian motion in time t

You can cut equal time intervals and each interval B changes by a normal random variable with 0 mean and time length variance; each B change in a time interval has nothing to do with another B change in a different interval.

And

hence

As Δt gets small, mathematicians argue the square of ΔB has a quite steady mean Δt with a negligible (higher order) variance 2*Δt[sup]2[/sup]. Hence here comes

**Ito's lemma**

**dB*dB=dt;dt*dt=o(dt);dB*dt=o(dt)**

Thus, to approach a difference of a function which involves a brownian motion B in it directly or indirectly, you have to use Taylor expansion with order 2 to capture the innegligible dB[sup]2[/sup].

thus here come the dc

where

o(dt) is higher order term of dt, negligible the same way we do our normal calculus. And the final formual for dc is:

*Last edited by George,Y (2008-04-19 05:09:29)*

**X'(y-Xβ)=0**

Offline

Pages: **1**