37

If I have the function $f(x)$ and I want to apply it $n$ times, what is the notation to use?

For example, would $f(f(x))$ be $f_2(x)$, $f^2(x)$, or anything less cumbersome than $f(f(x))$? This is important especially since I am trying to couple this with a limit toward infinity.

JShoe
  • 739

8 Answers8

58

In the course I took on bifurcation theory we used the notation $$f^{\circ n}(x).$$

asmeurer
  • 9,774
  • 3
    This is the notation I prefer, no matter what the commonest usage is. – Lubin Nov 30 '12 at 03:07
  • 2
    I wonder if my professor invented this notation (he also wrote the textbook for the class, so it's hard to say). It makes sense, as $\circ$ is the notation for function composition. I personally prefer just $f^n(x)$, though, when the context is clear. Writing all those $\circ$s gets a little annoying after a while :) – asmeurer Nov 30 '12 at 03:26
  • 2
    I like this a lot I have to say +1 – Simon Hayward Nov 30 '12 at 09:31
  • 4
    Oh, nice! Unconventional, but unambiguous and nearly self-explanatory to anyone familiar with ∘ and the standard superscript notation. – camccann Nov 30 '12 at 15:27
  • 2
    Fine for some people, but not entirely unambigous; the $\circ$ could be mistaken for zero, especially by those unfamiliar with it; and it's not something that's easy to look up. You would still need to define the notation unless you're sure of your audience. – LarsH Nov 30 '12 at 16:13
  • I have never seen this before. That being said, it is very lovely notation. – Nick Dec 01 '12 at 00:01
30

You can use the notation $f^n$ to denote the composition of the function with itself $n$ times, though this may also mean the product of $f$ with itself $n$ times. Just make sure you define your notation at the start.

29

You could define the notation recursively as a sequence of functions.

Let $f_{n+1}(x) = f(f_n(x))$ for $n \geq 2$ with $f_1(x) = f(x)$.

Sequence notation of this type is so generic that the reader will be forced to consult your definition, which will avoid any possible misinterpretation.

Austin Mohr
  • 25,662
23

If you take function iteration as a fold of self-composition, you can use a sum-like notation:

$\bigcirc^nf = \underbrace{f \circ \dots \circ f}_{n\:\text{times}}$

Where:

$\left({\bigcirc^0f}\right)(x) = x$

Granted, this is not very compact, and I would prefer to typeset the limit directly above the circle. That aside, it does combine tolerably with limit notation:

$\lim_{n\rightarrow\infty}\bigcirc^nf$

Jon Purdy
  • 527
16

You can use $f^n(x)$ BUT be sure to tell the reader that you mean functional iteration, not $(f(x))^n$.

Gerry Myerson
  • 179,216
6

There are two common notations for this, in two different communities:

  1. People in quantum physics, functional analysis and similar topics write everything without parentheses and for them, naturally $B^2x=BBx=B(B(x))$.

  2. Some people in albegra treat functions as "modifiers" and when $\alpha$ is a function, then they write the image of $x$ by $\alpha$ as $x\alpha$. For them, $x\alpha^2$ is what I would write $\alpha(\alpha(x))$.

At any case, as others mention, you can use whatever you like, as long as you clarify your notations at the beginning. If you need it a lot, I would go for one of $$f^n(x),\qquad f_n(x),\qquad f^{\circ n}(x),\qquad f^{[n]}(x).$$

yo'
  • 4,506
5

You may also use Lagrange's notation of derivative $\ f^{(n)}(x)\ $ instead of more commonly used notations $\ f^n(x)\ $ or $\ f_n(x)\ $.

EDIT:
Or you can use left indices: $\ ^n f(x)\ $ or $\ _n f(x)\ $ or $\ ^{(n)} f(x)\ $ or $\ _{(n)} f(x)\ $.

Egor Skriptunoff
  • 1,309
  • 9
    Wouldn't using Lagrange's notation be confusing since it already has a purpose? I don't want to make something analogous to $f^{-1}(f(x))=x$. That's a nightmare. – JShoe Nov 30 '12 at 02:52
1

There is a notation proposal for iteration of functions. Check if the following can be useful for your purposes: Salov, Valerii. "Notation for Iteration of Functions, Iteral", arXiv:1207.0152, June 30, 2012, https://arxiv.org/abs/1207.0152; Salov, Valerii. "Inevitable Dottie Number. Iterals of cosine and sine", arXiv:1212.1027, December 1, 2012, https://arxiv.org/abs/1212.1027.

Valerii
  • 11
  • 1