Let $(X,d)$ be a complete metric space, and let $f:X\to X$ be eventually contractive (there exists $p\in\mathbb{N}$ such that $g:=f^{\circ p}$ is contractive). Then $f$ has a unique fixed point.

Is there a way to prove this without using the Banach contraction theorem? Would appreciate a hint.

  • $\begingroup$ You have to explain what "eventually contracttive" means here $\endgroup$ – Orest Bucicovschi Oct 13 '17 at 3:12
  • $\begingroup$ Please see my updated question. $\endgroup$ – sequence Oct 13 '17 at 3:31
  • $\begingroup$ So you know how to prove it using Banach f p theorem, and want a different proof? $\endgroup$ – Orest Bucicovschi Oct 13 '17 at 3:34
  • $\begingroup$ Using the Banach theorem is actually just applying it. So I think this problem likely asks for more. $\endgroup$ – sequence Oct 13 '17 at 3:35
  • $\begingroup$ I don't think it is immediate. The contraction mapping theorem would imply that there is a unique fixed point for the power $f^{\circ p}$ only, not for $f$ itself. $\endgroup$ – timur Oct 13 '17 at 3:42

I suppose $f$ is continuous...

If $f^p$ is a contraction, then $(f^{np}(x))_n$ converges to the fixed point $x_0$ of $f^p$ for every $x\in X$. But then $f^{np+1}(x)$ converges to $f(x_0)$. But it also has to converge to $x_0$ (take $x'= f(x)$ as initial point). We conclude that $f(x_0) = x_0$.

Note that the fixed point will be the limit of $f^n(x)$ for every initial point $x$. The convergence rate is of the order $q^{\frac{n}{p}}$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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