I'm currently learning about the Ito's lemma / formula

In my textbook, a direct application of the formula is to compute quantities like that : (W is a Brownian motion)

enter image description here

While trying to prove these results I am finding that these computations are really not direct. Am I missing something and they're actually trivial with Ito's formula ?

For instance for the first one I couldn't directly compute the integral of Wt*dWt, I had to use Ito's formula on the derivative of Wt²

For the second one I managed to get the result but it was also very convoluted. So my question : using Ito's lemma, what's the direct approach to compute these formulas? thanks


Since by Ito's lemma you have for $\phi(t,x) \in \mathcal C ^{1,2}$

$$ \phi(t, X_t) = \phi(o,X_0) + \int _0 ^t (\partial_t +b\partial_x +\frac{\sigma^2}{2}\partial^2_{xx})\phi(s,X_s) ~ ds + \int_0^t\partial_x\phi(s,X_s) ~dW_s$$ if $X$ is a Ito's diffusion

$$ X_t = X_0 + \int _0 ^t b~ ds + \int_0^t\sigma ~dW_s,\ t \geq 0$$

you must search for the simples $\phi$ such that $\partial_x \phi =x $ with $X =B$ ( so $b=0$ and $\sigma =1$) for the first integral.

Indeed, we have $$ \int_0^T W_s ~dW_s=\frac{1}{2}(W^2_T-W^2_0 -T).$$

For the second formula a direct application of Ito's lemma would give the result. By effect, if you consider $\phi(t,x) = \exp(\lambda x-\frac{\lambda^2}{2}t) $ since $\partial_t\phi(t,x)= -\frac{\lambda^2}{2}\phi(t,x),\ \partial_x\phi(t,x)=\lambda \phi(t,x)$ and $\partial_{xx}^2\phi(t,x)=\lambda^2\phi(t,x)$, we have

\begin{align}\exp(\lambda W_t-\frac{\lambda^2}{2}t)=\phi(t,W_t)&=\phi(0,W_0)+ \int _0 ^t (\partial_t +\frac{1}{2}\partial^2_{xx})\phi(s,W_s) ~ ds + \int_0^t\partial_x\phi(s,W_s) ~dW_s \\&=1+\lambda\int_0^t \exp(\lambda W_s-\frac{\lambda^2}{2}s) ~dW_s\end{align}

Note that $(\phi(t,W_t))_{t \geq0}=(\exp(\lambda W_t-\frac{\lambda^2}{2}t))_{t \geq0}$ satisfy the SDE $$ L_t = 1 +\lambda \int_o^t L_s dW_s ,\ t \geq 0$$ and not $$ L_t = 1 + \int_o^t L_s dW_s ,\ t \geq 0$$ as you said in the question(probably a typo). Also, since $\mathbb E \left\{ \exp(2\lambda W_t-\lambda^2 t)\right\}< +\infty$ it is a (positive) exponential martingale

  • $\begingroup$ Nice answer, +1 $\endgroup$ – Ilya Jun 3 '13 at 12:01
  • $\begingroup$ @Ilya: Thank you Ilya! $\endgroup$ – Paul Jun 3 '13 at 13:38

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.