Local martingale property for stochastic integrals.

828 Views Asked by At

First we start with a Brownian Motion $B(t)$, $a\leq t\leq b$ and an admissible filtration for the BM $\{\mathcal F_t\}$.

Let $f(t,\omega)$ be a stochastic process satisfying:

  1. $f(t)$ is adapted to the filtration $\{\mathcal F_t\}$

  2. $\int_a^b |f(t)|^2 dt <\infty $ a.s.

Denote the space of all stochastic processes satisfying the latter $\mathcal L_{ad}(\Omega,L^2[a,b])$.

Then we have that

Let $f\in \mathcal L_{ad}(\Omega,L^2[a,b])$ then $$X_t=\int_a^t f(s)dB(s), a\leq t\leq b$$ is a local martingale w.r.t to the filtration $\{\mathcal F_t\}$.

Nothing is further added about this theorem in my textbook , and I would like to understand the proof of the later, in particular I don't get how to find a localising sequence of stopping times.

There's one example that claims:

Let $f(t)=e^{B(t)^2}\in\mathcal L_{ad}(\Omega,L^2[a,b]) $,

Then $$X_t=\int_a^t e^{B(s)^2} dB(s)$$ is a local martingale.

This is compatible with the theorem above but I am not sure how to properly prove this statement.

Any hint will be greatly appreciated.

1

There are 1 best solutions below

0
On BEST ANSWER

This is not exactly an elementary proof. Assuming that you are well versed in the properties of the Ito integral for the class of integrands $\{f: E[\int_a^b f^2(s)\, ds] < \infty \}, $ and know how to prove the existence of a continuous martingale modification for the stochastic integral for this class, then the localizing sequence (do you know the exact definition of a localizing sequence here?) can be taken to be $$\tau_n =\inf\bigl\{s: \int_a^s f^2(\omega, t)\, dt > n\} \hskip 4pt \text{ and } \hskip 4pt \tau_n = b \hskip 4pt \text{ if } \hskip 4pt \int_a^b f^2(\omega, t)\, dt \le n $$

Check Steele's Stochastic Calculus and Financial Applications, instead.Or some other textbook on stochastic calculus that is still at a reasonable introductory level.