Consider the functional equation $$\Big(\frac{1}{x}-1\Big)f(x)+\Big(\frac{1}{x^{\phi-1}}-1\Big)f(x^\phi)=1$$ where $\phi$ is the golden ratio. I’m looking for a continuous function $f:[0,1)\to \mathbb R^+$ with $f(0)=0$ satisfying this equation. I’ve shown that this function is unique, so if I can find a single elementary function satisfying it, then I’ve found the only solution meeting these requirements.

QUESTION: Can anyone find the function $f$ in closed form? I’m not interested in integral or series representations.

How I know there’s a unique solution: To see why there is a unique continuous solution with $f(0)=0$, we can do a series of repeated substitutions into the original functional equation:


$$\Big(\frac{1}{x^{\phi^2}}-1\Big)f(x^{\phi^2})+\Big(\frac{1}{x^{(\phi-1)\phi^2}}-1\Big)f(x^{\phi^3})=1$$ $$...$$

If we keep making the substitution $x\mapsto x^\phi$, we can treat this like a long system of equations in the variables $f(x),f(x^\phi),f(x^{\phi^2}),$ and so on. Through repeated substitution, we can solve for $f(x)$ in terms of $f(x^{\phi^n})$, which approaches $0$ as $n\to\infty$. The algebra is messy, but this leaves us with a different series representation for $f(x)$, showing that it is uniquely determined when we assume continuity and $f(0)=0$.

MOTIVATION: It turns out that the unique solution $f$ has the following series representation: $$f(x)=\sum_{n=1}^\infty x^{n+(\phi-1)\lfloor n (\phi-1)\rfloor}$$ and I’m trying to find a closed-form of this series (if not in terms of $x$, at least at some special values of $x$).

It’s a bit tricky to explain how I know $f$ satisfies this functional equation, but it can be proven from the following generalized identity: $$\frac{1-x}{x}\sum_{n\ge 1}x^n y^{\lfloor n\alpha\rfloor}+\frac{1-y}{y}\sum_{n\ge 1}y^n x^{\lfloor n/\alpha\rfloor}=1$$ which holds for all $x,y\in (0,1)$ and positive irrational $\alpha$. The functional equation for $f$ follows by setting $y=x^{\phi-1}$ and $\alpha=\phi-1$.

  • $\begingroup$ Is there any reason to expect it to have an elementary representation? $\endgroup$ Dec 27 '19 at 14:04
  • $\begingroup$ Can you maybe explain to me the idea you used to prove uniqueness and to obtain the series you got? $\endgroup$ Dec 27 '19 at 14:09
  • $\begingroup$ @Fimpellizieri No, no particular reason. I’m just hopeful. :) $\endgroup$ Dec 27 '19 at 14:10



$(e^{-\phi x}-e^{-x})f(e^{\phi x})=(e^{-x}-e^{-2x})f(e^x)+e^{-x}$





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.