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

Let $K$ be a non-empty compact convex subset of a Banach space $E$, and let $f : K \longmapsto K$ be a continuous function. Fix $u_0 \in K$, and define by recurrence $u_{n+1} = \frac{1}{n+1} \sum_{j=0}^n f(u_j)$ ; or equivalently $u_{n+1} - u_n = \frac{1}{n+1} (f(u_n) -u_n)$.

Is it always true that $(u_n)$ converges to some fixed point of $f$ ?

  • When $E$ is one-dimensional, this is an easy exercise, but all proofs I know use in a crucial way the ordering of the real numbers.
  • I don't expect the two-dimensional case to be easier than the general case.
share|cite|improve this question
up vote 4 down vote accepted

A counterexample in dimension $2$: take $K$ the unit closed disk of $\mathbb{C}$, and $f:K\to K$ the map $$f(z)=e^{i\pi/4}\frac{z}{|z|}\min(2|z|,1)$$ for $z\neq0$, and $f(0)=0$, which is the only fixed point of $f$. Then, starting by $u_0\in K$ with $|u _ 0|\ge1/2$ produces a sequence with $|u_n|\ge 1/2$. (Reason: if $|u_n|\ge1/2$ then $u_ {n+1}$ lies in the segment of endpoints $u _ n$ and $f(u_n)=e^{i\pi/4}\frac{u _ n}{|u _ n|}$, a segment that is disjoint from the disk $B(0,1/2)$, by elementary geometry considerations).

share|cite|improve this answer
If we start with $u_0 = 1$ then $u_1 = i$ and $u_2 = (i-1)/2$ so that $u_2$ is in the interior of $K$. – js21 Jun 26 '13 at 19:01
More precisely, $|u_n|$ is $\sim \frac{1}{\sqrt{n}}$ in your example. – js21 Jun 26 '13 at 19:11
Sorry. I think it is fixed now. – Pietro Majer Jun 26 '13 at 20:40
Thanks ! This also yield counterexamples in any dimension. – js21 Jun 27 '13 at 8:14

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.