11
$\begingroup$

Define $$f_1(x)=x\\f_2(x)=x^x\\\vdots\\f_{n+1}(x)=x^{f_n(x)}$$

Let $F_n(x)=f_n^{'}(x).$ Hence $$F_1(x)=1\\F_2(x)=x^x(1+\log(x))\\\vdots$$

Prove that $F_{2n}(x)=0$ has exactly one root in the interval $x\in(0,1),$ and this root $\to 0$ when $n \to \infty.$

Here are the images of $f_{2n}(x)$ for $n=1,2,\ldots,10.$

enter image description here

$\endgroup$
  • $\begingroup$ @Jaycob Coleman How do you get this? I think $F_3(x)=0$ has no real roots. $\endgroup$ – Next Aug 10 '13 at 14:20
  • $\begingroup$ @Hecke Do you have a source for this problem? $\endgroup$ – user1337 Aug 11 '13 at 22:55
  • $\begingroup$ Since $f_n$ converges only for $x \in [e^{-e},e^{1/e}]$ I would expect that the root of $F_n$ tends to $e^{-e}$ as $n \to \infty$. $\endgroup$ – Antonio Vargas Nov 10 '15 at 11:50
0
$\begingroup$

I would suggest to study formally $f_n(x)$ :

  • look at the values in 0 and 1, and the variations in between
  • study $f_n$ recursively
  • At this point prove that there is 1 root only $r_n$
  • prove that $r_n$ decreases
  • show that the limit cannot be > 0.
$\endgroup$
  • $\begingroup$ This would be more appropriate as a comment I think. $\endgroup$ – Antonio Vargas Nov 10 '15 at 11:53

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.