MathOverflow is a question and answer site for professional mathematicians. 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

Suppose the points of two random walks in $\mathbb{R}^2$ are given the step number (or time) as a third coordinate, so that they become paths in $\mathbb{R}^3$. Here are several pairs of walks of $n=100$ steps, both starting at the origin, with steps normally distributed with $\sigma=1$:
I would like to know the expected number of times that one path winds about the other, as a function of $n$, the number of steps. I believe for the three pairs illustrated, there is zero winding by $n=100$. Experiments indicate winding becomes less likely as $n$ grows. This is a bit counterintuitive to me.

The winding number of path $b(t)$ about path $a(t)$ up to $t=T$ could be defined by counting the number of times (the normalization of) the vector $b(t)-a(t)$ turns around the origin for $t\in[0,T]$.

share|cite|improve this question
What do you mean by "turns around the origin"? -- For random walks $a(t)$ and $b(t)$, the difference $a(t)-b(t)$ will "go hence and forth" in some sense, I think -- I don't see how you get your winding number from this. – Stefan Kohl Aug 6 '13 at 12:38
Sorry for not being clear. If one normalizes $b(t)-a(t)$, it can be viewed as a point on the unit circle centered on the origin. (Assume $a(t)=b(t)$ occurs with zero probability.) This point wanders around that unit circle. – Joseph O'Rourke Aug 6 '13 at 12:49
The normalization is clear -- but the issue is that the point won't wander around the unit circle in one direction. It will rather move randomly hence and forth, in both directions -- how do you count this? -- Or have I misunderstood something? – Stefan Kohl Aug 6 '13 at 13:06
Can't you equivalently consider a single random walk's winding number about the origin? I think you must put some barrier at the origin, because (in the plane) I think it will hit the origin with non-zero probability. Once you figured out the right set-up, you can convert the discrete random walk into a diffusion equation on the Riemann surface for $\log z$, and if you can solve the diffusion equation you will have your answer. – Yoav Kallus Aug 6 '13 at 15:31
By symmetry the expectation would be zero winding, but the interesting question would be how the variance grows with time. Here is the surface on which I was suggesting to solve the diffusion equation: – Yoav Kallus Aug 6 '13 at 15:43
up vote 4 down vote accepted

Please find here three papers giving at least partial answers to your question:

The upshot is that the root mean square winding number grows logarithmically with the number of steps N for $N \to \infty$.

share|cite|improve this answer
Excellent, Andreas! Direct hits. Thanks so much! – Joseph O'Rourke Aug 6 '13 at 20:04

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.