-1

If one seeks the functional solutions of $f^2(x)=f(x)+1=f(f(x))$

Are they called 'golden functions' ?

Are they always of the form of rational functions ?

  • 1
    Just curious: wouldn't it be more logical to call something like $f^{\circ2}(x)=f^{\circ1}(x)+f^{\circ0}(x)$ a "golden function"? (Which would then be $f^{\circ2}(x)=f(x)+x$ ) – Gottfried Helms Feb 15 '21 at 23:36

1 Answers1

4

At least on the (unbounded and infinite!) set $\operatorname{im}(f)$, we see that $f(t)=t+1$. Hence the only rational function solution is just $f(x)=x+1$.


In fact, the following will describe all solutions $f\colon \Bbb R\to\Bbb R$ of the functional equation $$ \forall x\in \Bbb R\colon f(f(x))=f(x)+1$$

Pick a non-empty set $A\subseteq \Bbb R$ that is closed under "$+1$". In other words, we have $A+1\subseteq A$, or $A$ can be written as $A=\bigcup_{a\in A_0}(\Bbb N+a)$. Let $g\colon \Bbb R\setminus A\to A$ be arbitrary and define $$ f(x)=\begin{cases}x+1&x\in A\\g(x)&x\notin A\end{cases}$$ Indeed, with such $f$, we have $f(x)\in A$ for all $x$ and hence $f(f(x))=f(x)+1$.

Here's a non-trivial example of a smooth such function: $$ f(x)=\begin{cases}x+1&x\ge 0\\x^2+x+1+e^{1/x}&x<0\end{cases}$$

  • I see what went wrong; I wrote a Moebius transformation $\frac{ax+b}{x + d},$ identities showed $b=ad$ so it is actually constant. But your linear map would appear as denominator constant in a Moebius function – Will Jagy Feb 14 '21 at 23:36
  • I tried rational functions but got stuck on a cubic – QuantumPotatoïd Feb 15 '21 at 07:02