MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

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

It is well known that under mild assumptions a scalar diffusion $dX_t = a(X_t) dt + \sigma(X_t) dW_t$ with invariant probability distribution $\pi$ is reversible. This is indeed not true for multidimensional diffusions. The usual proofs consists in writing down generators, speed functions etc...

I am trying to intuitively understand this result, and the only (not very satisfying) argument that I have found is the following. It is straightforward to check that any Markov chain on $\mathbb{Z}$ that has an invariant probability $\pi$ and that can only make jumps of size $+1$ or $-1$ is reversible: notice for example that $$F(k) = \pi(k)p(k,k+1)-\pi(k+1)p(k+1,k)$$ is independent of $k$ and is thus equal to $0$. If $a(\cdot)$ and $\sigma(\cdot)$ are regular enough, a diffusion can be seen as a limit of such Markov chains on $\epsilon \mathbb{Z}$ so that this makes the result plausible.

question: what are arguments/proofs/examples that could shed light on why a one dimensional ergodic diffusion is automatically reversible.

share|cite|improve this question

The identity you are referring to can be interpreted as saying that if you run your Markov chain on $\mathbb Z$ from the initial distribution $\pi$, then the probability of crossing the edge $[k,k+1]$ in the positive direction is the same as the probability of crossing it in the negative direction (as otherwise the measure $\pi$ won't be stationary).

The same idea works in the general case as well. For any nice test function $f$ $$ \lim \frac1t \int [f(\xi_t) - f(\xi_0)] d{\mathbf P}(\xi) = \int Df(x) d\pi(x) , $$ where ${\mathbf P}$ is the measure in the space of sample paths $\xi$ of the diffusion process with the initial stationary distribution $\pi$, and $D$ is the diffusion generator. The left-hand side of the above equation is 0 by stationarity of the measure $\pi$, so that $\int Df(x)d\pi(x)=0$. In the same way it is true for the reverse generator $D^{\ast}$ as well. Now, the operators $D$ and $D^{\ast}$ may differ by their first-order parts only (denote these vector fields by $v$ and $v^{\ast}$, respectively). Taking a step-like function $f$ whose derivative is non-zero only on a small set where $v\neq v^*$ (it is at this place that dim $=1$ is used) then gives a contradiction.

share|cite|improve this answer
thanks: this is exactly the kind of simple approaches that I am looking for. – Alekk Nov 8 '10 at 17:51

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.