I'm not convinced this is possible, as soon as you have $2$ distinct elements mapping to the same number, the function is no longer $1$-$1$ and therefore not a bijection.
Asked
Active
Viewed 427 times
4
-
yes, it is possible, but it cannot be continuous. – Masacroso Nov 27 '17 at 17:15
-
Non monotone does not imply that two elements must map to the same target in general; you are using your intuition about continuous functions. A bijection here will not be continuous. – Sort of Damocles Nov 27 '17 at 17:15
3 Answers
12
Let $f(x) = x$ for rational $x$ and $f(x) = 1-x$ for irrational $x$. This function is not only non-monotone, it is nowhere monotone (meaning that it is non-monotone on any subinterval of $[0,1]$).
User8128
- 15,485
- 1
- 18
- 31
-
that's interesting, for some reason I was stuck on thinking it has to be continuous. thanks – Nov 27 '17 at 17:21
-
Oh I see. If the function is continuous, then I believe it must be monotone as a consequence of the intermediate value theorem. – User8128 Nov 27 '17 at 17:22
3
How about the bijection $f:[0,1]\to [0,1]$ with $f(x)=\begin{cases} x, \text{if}~~ 0<x<1\\ 1,\text{if}~~ x=0\\ 0,\text{if}~~ x=1\end{cases}$
This is not strictly monoton.
Cornman
- 11,065
- 4
- 30
- 57
-
-
@user480172 Perhaps you should check carefully the meaning of "monotone", strictly or whatever. – DonAntonio Nov 27 '17 at 17:19
-
It is strictly monotone on $(0,1)$ but not on the whole intervall $[0,1]$, since $0<1$ but $f(0)>f(1)$. – Cornman Nov 27 '17 at 17:19
-
1
Remark that if a bijection is monotonous on $[0,1]$ you can build piecewise non-monotonous function based on the restriction of this bijection scaled to smaller intervals.
For instance $f:\begin{cases} x\in[0,\frac 14[ & 4x^2\\ x\in[\frac 14,\frac 12[ & \dfrac 34-x\\ x\in[\frac 12,\frac 34[ & x\\ x\in[\frac 34,1] & \dfrac{\cos(4\pi x-3\pi)+7}8\end{cases}$
zwim
- 28,563