A quick summary of the things regarding my question is this:

You have a probability space $(\Omega, \mathcal{F},P)$ and a filtration $\{\mathcal{F}_t\}$.

You have a $f(t,\omega): [0,\infty)\times\Omega \rightarrow \mathbb{R}$ satisfying.

  1. It is $\mathcal{B}\times\mathcal{F}$-measurable.

  2. $f(t,\omega)$ is $\mathcal{F}$-adapted.

  3. For two real numbers S,T $E[\int_S^Tf(t,\omega)^2dt]<\infty$.

A function $\phi$ is elementary if it satisfies the above points and is of the form :

$$\phi(t,\omega)=\Sigma_j e_j(\omega)\mathcal{X}_{[t_j,t_{j+1})}$$

Øksendal then shows that there is a sequence of elemntary functions $\phi_n$ such that $E[\int_S^T|f-\phi_n|^2dt]\rightarrow 0$.

Then he defines the integral as the $L^2$ limit of $\int_S^T\phi_n(t,\omega)dB_t(\omega)$. He says that the limit exists in $L^2$ since this is a Cauchy sequence. He refers to the Itô-isometry which he has proved for elementary functions, using this I get myself trying to show that the limit is Cauchy in $L^2$ :


But I still can't see how this can get arbitrary small as $n,m$ get big enough. I tried writing $(\phi_n-\phi_m)^2=(\phi_n-f+f-\phi_m)$ and then multiplaying out using the elementary square rules, and then trying to use Hölder's-inequality. But I wasn't able to finish it.

Can you please help me?


If $x,y$ are non-negative real numbers then $2xy\leq x^2+y^2$, hence $$ |\phi_m-\phi_n|^2\leq (|\phi_m-f|+|f-\phi_n|)^2=|\phi_m-f|^2+2|\phi_m-f||\phi_n-f|+||\phi_n-f|^2$$ $$\leq 2(|\phi_m-f|^2+|\phi_n-f|^2)$$

Now you can integrate from $S$ to $T$ and take expectation.

The big picture of what's going on is that every convergent sequence in a metric space is Cauchy.


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.