2

The function $f(x)=\sqrt{x\sqrt{x\sqrt{x\sqrt{x\sqrt{x...}}}}}\qquad$ quickly approaches $f(x) =x$.

$$f(x)=\sqrt{x\sqrt{x\sqrt{x\sqrt{x\sqrt{x...}}}}}\\=\sqrt{xf(x)}\\ \left[f(x)\right]^2=xf(x)\\\left[f(x)\right]^2-xf(x)=0\\f(x)[f(x)-x]=0\\ f(x)=0\\ f(x)=x$$

To say $2=\sqrt{2\sqrt{2\sqrt{2\sqrt{2\sqrt{2...}}}}}\qquad$ makes intuitive sense. What do I do with $0=\sqrt{2\sqrt{2\sqrt{2\sqrt{2\sqrt{2...}}}}}\qquad$?

MathAdam
  • 3,309
  • I often find in physics problems that I can make sense of an extraneous root. "The t=-3 s solution shows that the trajectory of a ball along the same path, but started 3 s earlier." – MathAdam Mar 30 '21 at 14:39
  • Oh. I did not actually know that. Let me delete my comment, then. Also, thank you for giving an example as to why extraneous roots make sense. – soupless Mar 30 '21 at 14:40
  • 1
    A discussion of extranous roots is here: https://math.stackexchange.com/questions/55445/when-do-we-get-extraneous-roots – MathAdam Mar 30 '21 at 14:52

2 Answers2

2

You can also talk about stable and unstable fixed points, so in this setting, $x$ is a stable fixed point of the dynamical system defined by $f:z\mapsto \sqrt{xz}$ : there is a neighbourhood $V$ of $x$ such that for $y$ in $V$, the sequence defined by $a_0 = y$ and $a_{n+1} = f(a_n)$ for any $n\in \mathbb{N}$, converges to $x$. In contrast, $0$ is an unstable fixed point, for any neighbourhood $W$ of $0$, all the sequences starting at a point of $W - \{0\}$ will not have $0$ as an accumulation point.

jeanmfischer
  • 1,779
  • +1 though I might say it is only half a neighbourhood of $0$ (bad things happen with negative $z$ and $x \not=0$) – Henry Mar 30 '21 at 15:02
  • $f$ being defined on $\mathbb{R}_{\ge 0}$ it is a full neighbourhood of 0 in that space that I am considering – jeanmfischer Mar 30 '21 at 15:08
1

For binary $\circ$, the usual convention is to define $a\circ(a\circ(a\circ\cdots))$ as $\lim_{n\to\infty}a_n$ for $a_0:=a,\,a_{n+1}:=a\circ a_n$, not simply any fixed point of $x\mapsto a\circ x$. So you don't need to "do anything with" fixed points other than the resulting limit, which is why @soupless called such apparent alternatives extraneous. (Finally, if the limit doesn't exist, the expression is undefined.)

J.G.
  • 115,835