I have long ago give up trying to find a nice formula for the $n$th iteration of functions in the form $$f_a(x)=x^2+a^2$$ However, it would be interesting to consider the asymptotic growth of the iteration of these functions. It is apparent that $f_a^{\circ n}(x)$ should behave like $\text{b}^{2^n}$, but calculating the value of the base $b$ in terms of $a$ and $x$ has been more difficult than I anticipated. We may solve for $b$ using a limit: $$b_a(x)=\lim_{n\to\infty} \big[f_a^{\circ n}(x)\big]^{2^{-n}}$$ but this is as far as I have gotten, and I can’t make this any neater. The best thing I have managed to find is the rather trivial observation $$b_a(x^2+a^2)=b_a^2(x)$$ and I have noticed that the graph of $b_a(x)$ typically looks like (but is not) a hyperbola.

Can anyone calculate any special values of $b_a(x)$ in closed form, or in terms of an infinite series or integral? Can you find any more interesting (non-trivial) functional or differential equations?

  • $\begingroup$ To get some idea of how complicated this can get, see oeis.org/A000058 and the links therein, for $f(x)=x^2-x+1$. Or oeis.org/A003095 for $f(x)=x^2+1$. $\endgroup$ – Gerry Myerson Mar 2 at 4:03
  • $\begingroup$ Had a look at those sites? Convinced? $\endgroup$ – Gerry Myerson Mar 3 at 5:33

From Wikipedia, it states that the $n$th iteration of $$ax^2+bx+\frac{b^2-2b-8}{4a}$$ is $$\frac{2c^{2^n}+2c^{-2^n}-b}{2a}$$ where $$c=\frac{2ax+b\pm\sqrt{(2ax+b)^2-16}}{4}$$

Thus, we can solve the case $f(x)=x^2-2$.

$$b_{\pm 2i}(x)=\frac{x+\sqrt{x^2-4}}2$$

  • $\begingroup$ Yes, but that's a very special case. In the general quadratic case, there's no closed-form formula. $\endgroup$ – Gerry Myerson Mar 2 at 4:01

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.