Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

If $\sigma$ is Lipschitz, with Lipschitz constant $K$, and $(X_t)_{t\geq 0}$ solves

$$dX_t=\sigma(X_t)dB_t,$$ where $B$ is a Brownian motion, then is $X$ a martingale? I'm having difficulty getting past the self-reference here. I tried showing that, for $t\geq 0$, $\mathbb{E}[X]_t$ is finite. Perhaps Gronwall's lemma is needed?

Thank you.

share|cite|improve this question
You could try and have a look at Theorem 4.40 (b) (and the above definition 4.39) in Jacod and Shiryaev's Limit Theorems for Stochastic Processes. If we are in the scope of that Theorem then $(X_t)$ is in fact a square integrable martingale. – Stefan Hansen Apr 3 '12 at 10:57
@StefanHansen I'm away from a library at the moment. If this works, put it as an answer, and I'll find the book later. :) – Ben Derrett Apr 3 '12 at 12:32

Unfortunately, it didn't work in general (not saying that your claim is false though). Here is what I was trying to do. Maybe it can help you come up with ideas, otherwise just nevermind it.

Let $L^2(X)$ be the set of all predictable processes $H$ such that the process $(\int_0^t H^2_s d\langle X\rangle_s)_{t\geq 0}$ is integrable, where $(\langle X \rangle_t)_{t\geq 0}$ denotes the predictable quadratic variation process.

In the following $\mathcal{H}^2$ (resp. $\mathcal{H}_{\text{loc}}^2$) denotes the set of all square integrable martingales (resp. locally square integrable martingales). Then we have the following theorem from Jacod & Shiryaev:

Theorem 4.40(b). Let $(X_t)_{t\geq 0}\in \mathcal{H}_{\text{loc}}^2$. Then $(\int_0^t H_s d X_s)_{t\geq 0} \in \mathcal{H}^2$ if and only if $H\in L^2(X)$.

Obviously we are in the scope of this theorem as $(B_t)_{t\geq 0}\in\mathcal{H}^2$ with predictable quadratic variation $\langle B_t\rangle = t$, $t\geq 0$. Furthermore if $(X_t)_{t\geq 0}$ is the solution to the SDE of the original post, i.e. $$ X_t=\int_0^t \sigma(X_s) dB_s,\quad t\geq 0, $$ then $(X_t)$ is adapted and continuous and hence it is a predictable process. The theorem now yields that $(X_t)_{t\geq 0}\in \mathcal{H}^2$ if and only if $$ E\left[\int_0^t \sigma(X_s)^2 d \langle B\rangle_s\right]=E\left[\int_0^t \sigma(X_s)^2 ds\right]=\int_0^t E\left[\sigma(X_s)^2\right] ds<\infty $$ holds for all $t\geq 0$. Using the Lipschitz assumption we get a sufficient condition for $(X_t)$ being a square integrable martingale: $$ \int_0^t E[|X_s|]ds<\infty, \quad t\geq 0. $$

share|cite|improve this answer
up vote 0 down vote accepted


$$[X]_t = \int_0^t\sigma(X_u)^2du,$$


$$\begin{align} \mathbb{E}([X]_t) \le \int_0^t \mathbb{E}\left[(x_0 + K|X_u-x_0|)^2\right]du.\\ \end{align} $$

$X$ is locally bounded in $L^2$. See, for example, Karatzas and Shreve equation 5.2.15 (p. 289). So it follows easily that $\mathbb{E}([X]_t)<\infty$, for each $t$. Hence $X$ is a martingale.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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