$$\cfrac{z}{1+z} \cdot \cfrac{z}{1+z-\cfrac{z}{1+z}} \cdot \cfrac{z}{1+z-\cfrac{z}{1+z-\cfrac{z}{1+z}}} \cdots= 1-\frac{1}{z}$$

I propose that this works for any $z \in C$ if and only if $|z|>1$.

That much I got by experimenting and by the identity I already proved using Euler's continued fraction formula:

$$\cfrac{z}{1+z-\cfrac{z}{1+z-\cfrac{z}{1+z-\cfrac{z}{1+z-\cdots}}}}=\begin{cases}1 & |z| > 1\\z & |z| \leq 1\end{cases}$$

But I don't know how to prove the above product. Any ideas?


Well, let's start by calling the terms of your product $a_n$, that is, $a_1=\frac{z}{1+z}, a_{n+1}=\frac{z}{1+z-a_n}$. Some experimenting suggests the formula:


and this can be verified fairly easily by induction. Now, let's consider the partial products:

$$a_1a_2\cdots a_n=\frac{z(1-z^1)}{1-z^{2}}\cdot \frac{z(1-z^2)}{1-z^{3}}\cdots \frac{z(1-z^n)}{1-z^{n+1}}=z^n\frac{1-z}{1-z^{n+1}}=\frac{z-1}{z-z^{-n}}$$

Considering the cases $|z|<1, |z|>1$ separately, we see that:

  • $|z|<1$ causes the product to "diverge to 0"
  • $|z|>1$ lets the product converge to $\frac{z-1}{z}$
  • $|z|=1$ has varied behaviour - if the $z$ is a root of unity, the product will contain terms of both $0$ and $\infty$, and if not, it will vary without converging.
  • $\begingroup$ But numerically I get $1-\frac{1}{z^2}$, not $1-\frac{1}{z}$, $\endgroup$ – Yuriy S Feb 22 '16 at 15:41
  • $\begingroup$ Perhaps because $1-\frac{1}{z^2}=\frac{1-\frac{1}{z}}{\left(\frac{z}{1+z}\right)}$ - might be missing the first term? $\endgroup$ – πr8 Feb 22 '16 at 15:43
  • $\begingroup$ You are right, my algorithm was wrong, thank you $\endgroup$ – Yuriy S Feb 22 '16 at 15:45

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.