1
$\begingroup$

This question was on math.stackexchange but got no answer. The link is https://math.stackexchange.com/questions/598811/kolmogorov-backward-equation-for-ito-diffusion

I want to know the answer too, so here it is

Let $(X_t)_{t\ge 0}$ be the solution of the SDE $$ X_t = X_0 + \int_0^t \mu(s,X_s) \,ds + \int_0^t \sigma(s,X_s) \,dB_s, \quad t\ge 0 $$ where $\mu(s,x)$ and $\sigma(s,x) $ are Lipschitz continuous in $x$ uniformly in $t$. My question is related to the last argument in a proof (in the book of Klebaner - Introduction to Stochastic Calculus in Section 6.2 on p.154) showing that $u(t,x):=\mathbb{E}(g(X_T)|X_t=x)$ solves $$ \frac{\partial}{\partial_t}u(t,x) + \mu(t,x)\frac{\partial}{\partial_x}u(t,x) + \frac{\sigma^2(t,x)}{2}\frac{\partial^2}{\partial x^2}u(t,x)=0.\qquad (*) $$ In the proof it is shown that $$ \int_0^t \frac{\partial}{\partial_t}u(s,X_s) + \mu(s,X_s)\frac{\partial}{\partial_x}u(s,X_s) + \frac{\sigma^2(s,X_s)}{2}\frac{\partial^2}{\partial x^2}u(s,X_s) \,ds = 0 \quad \mathbb{P}-a.s. $$ for every $t\ge 0$. Now it is concluded that the PDE $(*)$ above holds. Why is that?

$\endgroup$

1 Answer 1

2
$\begingroup$

First note the following fact: if $f : [0,T] \to \mathbb{R}$ is an integrable function and $\int_0^t f(s)\,ds = 0$ for all $0 \le t \le T$, then $f = 0$ almost everywhere. (It's immediate that $\int_a^b f(s)\,ds = 0$ for all $a,b$. Now use a monotone class argument to show that $\int_A f(s)\,ds = 0$ for every measurable $A$. Finally consider $A = \{f > 0\}$ and $A = \{f < 0\}$.)

So if we call the integrand $F(s,\omega)$, what we have shown is that for almost every $\omega$, $F(s,\omega) = 0$ for almost every $s$ (the null set of $s$ depending on $\omega$). But then Fubini's theorem implies that $F(s,\omega) = 0$ for almost every $(s,\omega)$ with respect to product measure.

$\endgroup$
0

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.