By Ito's representation theorem because the integral of Brownian Motion is mean zero we should have some random function $\phi$ such that

$$\int_0^T W_t dt = \int_0^T\phi(\omega,t)dW_t$$

I'm a little uncomfortable about the left hand side being finite variation and the right a local martingale even for fixed $T$, I just can't get any closer to the form above than the obvious Ito formula result

$$\int_0^T W_t dt = TW_T-\int_0^TtdW_t$$



The identity

$$\int_0^T W_t \, dt = T W_T - \int_0^T t \, dW_t$$

is a good start. Now note that

$$T W_T = \int_0^T T \, dW_t$$

which implies that

$$\int_0^T W_t \, dt = \int_0^T (T-t) \, dW_t.$$

This is exactly the representation you are looking for.

  • 1
    $\begingroup$ Haha wow how do I even get stuck with these easy problems, thanks $\endgroup$ – John Fernley Oct 24 '16 at 13:57
  • $\begingroup$ @JohnFernley You are welcome. $\endgroup$ – saz Oct 24 '16 at 13:57
  • $\begingroup$ Actually also wondering in $\int_0^T W_t \, dt = \int_0^T (T-t) \, dW_t$, am I right that here the right hand side is a local martingale in $T$ and the left hand side is not? $\endgroup$ – John Fernley Oct 24 '16 at 14:00
  • $\begingroup$ @JohnFernley Why do you think so? Since $\int_0^T W_t \, dt = \int_0^T (T-t) \, dW_t$ the left-hand side is a (local) martingale if and only if the right-hand side is a (local) martingale. The stochastic integral $\int_0^t f(s) \, dW_s$ is a (local) martingale, but in our setting we have a (more general) expression of the form $\int_0^t f(s,\color{red}{t}) \, dW_t$. $\endgroup$ – saz Oct 24 '16 at 14:18
  • 1
    $\begingroup$ @JohnFernley No, it's not a martingale. $\endgroup$ – saz Oct 24 '16 at 14:43

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.