Determining $dX_t$ for stochastic equations, and which of these are $\mathcal{F} $ - martingales?

123 Views Asked by At

I want to write down an expression for $dX_t$ for both:

i. $X_t=t^2W_t^2-2\int_0^t(sW_s^2+s^2)ds$; and

ii. $X_t=W_t^2-tW_t$

What is the process I would use for differentiating these stochastic equations?

Also how can I tell which of these are $\mathcal{F} $ - martingales on [0,T] where ${\mathcal{F_t}} $ is the natural filtration generated by $W_t$ ?

I believe we can assume that $\int_o^T\mathrm{E_p}[|W_t|^n]dt<\infty$ for all $n>0$

Please let me know if you have any advice or solution, would be greatly appreciated, thanks.

1

There are 1 best solutions below

6
On BEST ANSWER

Let's start with the first one; I leave the other one to you. First of all, it follows from the definition of $dX_t$ that

$$dX_t = d(t^2 W_t^2) -2 (t W_t^2+t^2) \, dt$$

i.e. we only have to calculate $d(t^2 W_t^2)$. By applying Itô's formula to $f(t,x) := t^2 \cdot x^2$ we see that

$$d(t^2 W_t^2) = 2 t^2 W_t \, dW_t + ( t^2 + 2t W_t^2) \, dt.$$

Hence,

$$dX_t = 2t^2 W_t \, dW_t - t^2 \, dt,$$

i.e.

$$X_t - X_0 = 2 \int_0^t s^2 W_s \, dW_s - \frac{t^3}{3}.$$

In order to decide whether $(X_t)_{t \geq 0}$ is a martingale we recall that any stochastic integral of the form $\int_0^t f(s) \, dW_s$ is a martingale. This means that

$$M_t := 2\int_0^t s^2 W_s \, dW_s$$

is a martingale. Since

$$X_t-X_0 - M_t = - \frac{t^3}{3}$$

is obviously not a martingale, this already implies that $(X_t)_{t \geq 0}$ is not a martingale.

Remark Note that the integrability condition $\int_0^T \mathbb{E}(|W_t|^n) \, dt<\infty$ is nothing "we can assume", we have to check it!