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

I've seen some estimates for the decay in $d$ of the probability a SRW makes a distance $d$ in time $n$, but is there any reference for the "speed" of a random walk in a group? I'm interested mostly for nilpotent groups (transient ones), though subexponential or polycyclic could be also nice.

Let $X_n$ be the random variable of a simple random walk starting at the neutral element $e$ and $|\cdot|$ be the word length (for some generating set). Let $l_n$ be the expected value of $|X_n|$. To put the question in precise form:

$\mathbf{Question}:$ For a nilpotent group $G$, does there exists $\alpha \in ]0,1[ $ so that for some constant $C>1$, $l_n \leq C n^\alpha$?

Is there any estimates on the possible values of $\alpha$?

share|cite|improve this question
up vote 4 down vote accepted

For nilpotent groups the speed exponent is 0.5. See this paper for the general picture .

share|cite|improve this answer
Very nice reference... is it obvious that there are amenable groups where the speed exponent is 1? – Antoine Jul 20 '13 at 13:47
You can have a look at Theorem 9.5 in these nice lecture notes – Dan Sălăjan Jul 20 '13 at 14:39
Thanks again for this other great reference. Reading through these notes, I read that Kaimanovich showed there are generators for which Thompson's group has speed exponent 1 (after question 15.3). I could not find any reference to this...would you happen to know of it? – Antoine Oct 1 '13 at 8:35
You re welcome. Very interesting, I don´t know any reference or proof... I might think about it, maybe there is a simple reason for it. I let you know if I succeed. Dan – Dan Sălăjan Oct 1 '13 at 8:40

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.