Bounding Expectation of the Solution to SDE

128 Views Asked by At

I have been give the following problem.

Let X be a solution to the SDE $X(t) = X(0) + \int_0^1 b(X(s))ds + \int_0^1 \sigma(X(s)) dW(s)$ where $W$ is a standard one-dimensional Brownian motion, and the functions $b : \mathbb{R} \rightarrow \mathbb{R}$ and $\sigma : \mathbb{R} \rightarrow \mathbb{R}$ have at most linear growth; that is, there is a constant $K$ such that $|b(x)| \leq K(1 + |x|), |\sigma(x)| \leq K(1 + |x|)$. Suppose that $E|X(0)|^2 < \infty$. Show that there is a constant $0 < C_2(T) < \infty$ such that $$E [\sup_{0≤t≤T} |X(t)|^2]< C_2(T).$$


The way I am trying to prove this statement is by first finding $d|X(t)|^2$ by using Ito's Lemma. Then take the expectation of $d|X(t)|^2$ and bound this expectation in a way so that I can use Gronwall's Inequality. From here, I think I would be able to arrive at the conclusion of the proof. Below is a sketch of the work I have done so far.


Sketch of my work: First, we find that $$d|X(t)|^2 = [2X(t) b(X(t)) +\sigma^2(X(t))]dt + 2X(t)\sigma(X(t)) dW(t).$$ Taking the expectation, we have $$dE|X(t)|^2 = E[2X(t) b(X(t)) +\sigma^2(X(t))]dt.$$ We can bound this expectation by using the growth conditions on $b$ and $\sigma$ to get $$E|X(t)|^2 \leq E|X(0)|^2 + \int_0^t E[2X(s)K(1+|X(s)|) +K^2(1+|X(s)|)^2]ds.$$


This is the point were I am getting stuck. I am unsure how I can manipulate the right side of the inequality to achieve something that will resemble Gronwall's Inequality. So far, does my work seem correct? Is this a valid way of trying to prove this statement? Any advice on how to proceed?

1

There are 1 best solutions below

0
On BEST ANSWER

Yes, this looks correct so far.

It looks like the parts that don't immediately fit Gronwall's are the $E[|X(s)|]$ term and the $E[(1+|X(s)|)^2]$ term. For the first, you can just use that $|a| \le 1+a^2$ for all $a \in \mathbb{R}$. For the second, you can use that $(a+b)^2 \le 2(a^2+b^2)$ for all $a,b \in \mathbb{R}$. This should be the exact form you need to apply Gronwall's inequality to the function $E[|X(t)|^2]$.