I am new to this forum as well as partial differentiation. I would like to ask the following question.

Given a geometric brownian motion:

$$dS_t=(\mu S_t)dt+(\sigma S_t)dz_t$$

I would like to apply Ito's Lemma to $dS_t$ itself such that I would expect it will still give the same $dS_t$.

To do this, We need to know $\partial S_t \over \partial S_t$, $\partial^2 S_t \over \partial S_t^2$, and $\partial S_t \over \partial t$.

I have no problem of finding the first 2 derivatives, but how to calculate $\partial S_t \over \partial t$?


  • $\begingroup$ Hint: Write the differential equation in integral form. $\endgroup$ – Chris Janjigian Jun 7 '13 at 18:38
  • $\begingroup$ What is the meaning of $dz$ in $dS_t=(\mu S_t)dt+(\sigma S_t)dz$? What do you mean by "to apply Ito's Lemma to $dS_t$ itself"? How do you define $\partial S_t \over \partial S_t$, $\partial^2 S_t \over \partial S_t^2$, and $\partial S_t \over \partial t$? $\endgroup$ – Did Jun 7 '13 at 18:58
  • $\begingroup$ @Did, we refer $dz_t$ to a Wiener's Process. In my attempt, I am learning to use Ito's Lemma, which can be used to find $df_t$ in terms of $dS_t$. Here if we let $f_t=S_t$, we should expect the Ito's Lemma should give us the original $S_t$. So this is what we mean by "applying Ito's Lemma to $dS_t$ itself". $\endgroup$ – Ric Yik Jun 8 '13 at 4:59
  • $\begingroup$ ?? Your use of Itô's lemma below is rather... idiosyncratic. :-) For example you still did not say what you mean by $\partial S_t/\partial t$. One might mention at this point that the function $t\mapsto S_t(\omega)$ is almost surely not differentiable. $\endgroup$ – Did Jun 8 '13 at 10:42
  • $\begingroup$ @Did, the $\partial f_t \over \partial t$ is one of the partial derivatives we need in order to apply Ito's Lemma to obtain $df_t$. Here I set $f_t=S_t$ and therefore we have $\partial S_t \over \partial t$. $\endgroup$ – Ric Yik Jun 10 '13 at 15:31

Thanks @chris-janjigian. Let me try to find out $\partial S_t \over \partial t$ by using the integral form of $dS_t$. I got the integral form after searching from Google: $$S_t = S_0 + \int_0^t (\mu S_u)du+\int_0^t(\sigma S_u)dz_u$$

As $t$ is just representing "t-times" in this form, ${\partial S_t \over \partial t}=0+0+0=0$. We also know that ${\partial S_t \over \partial S_t}=1$ and ${\partial^2 S_t \over \partial S_t^2}=0$. Therefore, using Ito's Lemma: $$dS_t=[{\partial S_t \over \partial S_t}(\mu S_t)+{\partial S_t \over \partial t}+{1 \over 2}{\partial^2 S_t \over \partial S_t^2}(\sigma S_t)^2]dt+[{\partial S_t \over \partial S_t}(\sigma S_t)]dz_t$$ $$dS_t=[1(\mu S_t)+0+0]dt+[1(\sigma S_t)]dz_t$$ $$dS_t=(\mu S_t)dt+(\sigma S_t)dz_t \Box$$


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.