The function $f : \mathbb{R} \to \mathbb{R}$ is continuous . Also $0\lt x_1 \lt x_2 \lt x_3 \lt x_4 \ $ and $f(x_1) = x_2 \ , \ f(x_2) = x_3 \ , \ f(x_3) = x_4 \ , \ f(x_4) = x_1 $ . Prove that there is a $m \in \mathbb{R}$ so that $f(f(m)) = m$ .

My Try : Let $g(x) = f(x) - x $ and it is continuous over $\mathbb{R} $ . Furthermore , we have $g(x_1) \ , \ g(x_2) \ , \ g(x_3) \gt 0 $ and $g(x_4) \lt 0$ . The result is $f(c) = c$ for some $c$ between $x_3$ and $x_4$ . Applying $f$ to both sides yields to $f(f(c)) = f(c) = c$ .

Is my solution correct ? Also write other solutions . Thanks in advance !

  • $\begingroup$ I think you've got the cleanest solution (my idea was the same; it boils down to using the intermediate value theorem). You actually prove a stronger result: there exists $m\in\Bbb R$ such that $f(m)=m$. This automatically means $m$ is a fixed point for any number of compositions of $f$. $\endgroup$
    – Clayton
    Oct 2 '18 at 20:06
  • $\begingroup$ @Clayton Thanks for your reply . If we have three $x_i$'s instead of four , it will work again . So the number of $x_i$'s is irrelevant to question ? $\endgroup$
    – S.H.W
    Oct 2 '18 at 20:14
  • $\begingroup$ Correct. All you really need are two points, $x_1<x_2$ with the conditions $f(x_1)=x_2$ and $f(x_2)=x_1$. This will cause $g(x)=f(x)-x$ to have the properties $g(x_2)<0<g(x_1)$, so there is a root of $g(x)$ in this interval. $\endgroup$
    – Clayton
    Oct 2 '18 at 20:19
  • $\begingroup$ Okay , thanks a lot . $\endgroup$
    – S.H.W
    Oct 2 '18 at 20:21

Yes, the solution is correct.

In fact, you show something stronger, using a weaker hypothesis: all you need is a continuous function on an interval $[x_1,x_2]$ such that $f(x_1)\geq x_1$ and $f(x_2)\leq x_2$ to conclude that $f$ has a fixed point somewhere in $[x_1,x_2]$.

  • $\begingroup$ Also it can be done, by finding such $m,n$ that $f(m)=n,f(n)=m$ since, $f$ oscillates and continuous, we must have such $m,n$. $\endgroup$
    – A learner
    Jul 29 '20 at 15:27
  • $\begingroup$ @Subhajit: I'm not sure how you would prove that without proving that $f$ has a fixed point. $\endgroup$
    – tomasz
    Jul 29 '20 at 21:33

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.