15
$\begingroup$

If I have $f(x)\leq f(f(x))$ for all real $x$, can I deduce $x\leq f(x)$?

Thank you.

| cite | improve this question | |
$\endgroup$
  • 2
    $\begingroup$ You can deduce $x \le f(x)$ for all $x$ in the image of $f$ (i.e., all those numbers $x$ which actually occur as the function value at some $y$, namely $x = f(y)$). You cannot deduce anything about other $x$. $\endgroup$ – ShreevatsaR Nov 4 '12 at 18:07

11 Answers 11

Active Oldest Votes
52
$\begingroup$

No. Suppose $f$ is a constant function.....

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ If $f(x)$ is a constant function then $x\le f(x)$, so it is not a counterexample. ?????? Or was this perhaps meant as a hint, that the OP should try to find a function for which $x\le f(x)$ does not hold, but $f(f(x))$ is constant? $\endgroup$ – Martin Sleziak Nov 4 '12 at 6:57
  • 10
    $\begingroup$ @Martin: If $f(x)$ is a constant, what constant could it be that is greater than all $x$? $\endgroup$ – user856 Nov 4 '12 at 7:02
  • 2
    $\begingroup$ @MartinSleziak: If $f(x)$ is a constant function, then there is a unique $x_0$ such that $f(x_0)=x_0$. Then for all $x>x_0$ we have $x>x_0=f(x_0)=f(x)$. $\endgroup$ – Cameron Buie Nov 4 '12 at 7:03
  • $\begingroup$ @Rahul, Cameron: You are correct, sorry for my blunder. $\endgroup$ – Martin Sleziak Nov 4 '12 at 7:04
15
$\begingroup$

For a slightly non-trivial example, $f(x) = - \vert x \vert$. We then have that $$f(f(x)) = - \vert f(x) \vert = -\vert-\vert x \vert \vert = - \vert x \vert = f(x)$$ However, $x \geq f(x)$ for all $x \in \mathbb{R}$ and in fact $x > f(x)$ for all $x \in \mathbb{R}^+$.

| cite | improve this answer | |
$\endgroup$
9
$\begingroup$

Let $f(x)=1$. Then $f(f(x))\leq f(1)=1$. However, $2\geq f(2)=1$ so that the conclusion does not follow.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ You example makes sense. But if I substitute $a=f(x)$, $a\leq f(a)$, which leads us $x\leq f(x)$, this seems make sense too... So what's wrong? $\endgroup$ – A. Chu Nov 4 '12 at 7:04
  • 2
    $\begingroup$ @jasoncube when you substitute $a = f(x)$ you imply that $a \in \mathrm{ Range}(f)$ so you have this inequality for those $a$ you cannot deduce it holds for every $x$ $\endgroup$ – clark Nov 4 '12 at 7:07
  • $\begingroup$ @jasoncube: Note that for the function $f(x) = 1$ (the constant function which takes value $1$ for all $x$), if you substitute $a = f(x)$, then necessarily $a = 1$. So although $a \le f(a)$ is true (for $a = 1$), for any other $x > 1$, it is not true that $x \le f(x)$, because $x > 1$ but $f(x) = 1$. $\endgroup$ – ShreevatsaR Nov 4 '12 at 18:04
9
$\begingroup$

Let $f$ map $-1$ to $-2$ and all other values to $0$. Then $f(-1)<-1$ but $f(f(x))\geq f(x)$ for all $x$.

| cite | improve this answer | |
$\endgroup$
7
$\begingroup$

For an example showing that even strict inequality $f(x)<f(f(x))$ for all $x$ does not imply $x\leq f(x)$ for all $x$, consider $$ f: x\mapsto \begin{cases} -\frac{|x|}2 &\text{if }x\neq0\\ \\-1&\text{if }x=0\\ \end{cases} $$ The discontinuity at $x=0$ is inevitable, since a continuous function $\mathbf R\to\mathbf R$ that has both points $x$ where $x<f(x)$ and other points where $x>f(x)$ must have some fixed points with $x=f(x)$.

| cite | improve this answer | |
$\endgroup$
7
$\begingroup$

No.

Suppose $f(x)=1$.

$\left\{ \begin{array}{l} f(x)=f(f(x))=1 \\ 1\leq1 \end{array} \right. \therefore f(x) \leq f(f(x)) $ for all real $x$.

If $x>1$, $f(x)<x \therefore x\not\leq f(x)$.

Plot

| cite | improve this answer | |
$\endgroup$
6
$\begingroup$
  • YES if $f$ is monotonically increasing in the smallest interval containing $x$ and $f(x)$
  • NO if $f$ is monotonically decreasing in the smallest interval containing $x$ and $f(x)$
  • If $f = k$ is a constant function, then it is TRUE for all $x\le k$ and FALSE for all $x>k$
  • NOT NECESSARILY TRUE in any other case.
| cite | improve this answer | |
$\endgroup$
5
$\begingroup$

If $f$ satisfies the above conditions, then:

  1. For all $x \in Im(f)$ we have $x \leq f(x)$. In particular, if $f(x)$ is onto then, the problem is true. To see this, let $y$ be so that $x=f(y)$, then $f(y) \leq f(f(y)) \Rightarrow x \leq f(x)$.

  2. There exists non onto functions which satiesfy the above condition for which, $x \leq f(x)$ if and only if $x \in Im(f)$. For example $f(x)=\frac{x-|x|}{2}$.

| cite | improve this answer | |
$\endgroup$
4
$\begingroup$

No, when the function which is a decreasing function did not meet the result. Such as $f(x)=-x$,only when $x\geq 0$, $f(x)\leq f(f(x))$; but at the same time, $x\geq f(x)$.

| cite | improve this answer | |
$\endgroup$
2
$\begingroup$

If $f(x)<f(f(x))$ for all $x$ and $f$ is continuous then $x<f(x)$ for all $x$:

Define a continuous function $g$ by $g(x)=x-f(x)$. Now $g(f(0))=f(0)-f(f(0))<0$, so $g$ is below the $x$-axis at some point. Since $g(x)=0$ would imply $x=f(x)$ and hence $f(x)=f(f(x))$, we get $g(x)\ne 0$ for all $x$. Thus, $g(x)<0$ for every $x$.

| cite | improve this answer | |
$\endgroup$
2
$\begingroup$

Well.. here is a sufficient condition:

If $f$ is invertible and monotonically increasing then $f^{-1}$ exits and is also monotonically increasing and therefore $f(x)\leq f(f(x)) \implies f^{-1}(f(x)) \leq f^{-1}(f(f(x))) \implies x \leq f(x)$.

| cite | improve this answer | |
$\endgroup$
  • $\begingroup$ Actually, as mentioned in my answer, onto is enough ;) $\endgroup$ – N. S. Nov 4 '12 at 18:04
  • $\begingroup$ @N.S. You're right! :) $\endgroup$ – Amatya Nov 4 '12 at 20:37

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.