0

I have been given a task to find a bijection from $[0...1] \rightarrow (0...1)$

My workbook has provided the following function:

$f(x) = 0, x=\frac{1}{2}$

$f(x)=\frac{1}{n-2},x=\frac{1}{n}, n \in \mathbb{N}, n > 2$

$f(x) = x$, for every other $x \in (0...1)$.

Now my question is: Is there a more simple (but not as elegant) way to create a bijection? Here's what I have come up with:

$f(x) = \frac{1}{2}, x=1$

$f(x)=\frac{1}{3},x=0$

$f(x) = \frac{x}{4}$, for every other $x$.

Is this a valid bijection between the two? If so, how can I prove that it is (and do I need to prove it is a bijection)?

l0ner9
  • 623
  • 4
    what is the preimage of $\frac{2}{3}$ – Mor A. Jun 03 '23 at 11:27
  • $f(0)$ is not in your codomain? – MathFail Jun 03 '23 at 11:29
  • 1
    @MathFail how so? f(0) = 1/3 = 0.333.. and $0.333.. \in (0...1)$? Unless I ma missing something – l0ner9 Jun 03 '23 at 11:48
  • 1
    Think about @MorA.'s comment. – GEdgar Jun 03 '23 at 12:11
  • The standard notation for those intervals is $[0,1]$ and $(0,1)$. Avoid using the $ ... $ notation since that's sometimes used (not really standard) for intervals of integers (so $[0 ... 1]$ would be a set with only two elements and $(0...1)$ would be the empty set). – jjagmath Jun 03 '23 at 12:11
  • 1
    Your workbook solution is wrong. With that definition $f(0) = 0 \not\in (0,1)$ and $f(1) = \frac{1}{1-2} = -1 \not\in (0,1)$. – jjagmath Jun 03 '23 at 12:15
  • @jjagmath I don't think it's wrong since I was the one who made a mistake in copying it from the workbook. Fixed it now, should be in order. – l0ner9 Jun 03 '23 at 12:21
  • Still missing something. With that definition, what is $f(0)$ and $f(1)$? – jjagmath Jun 03 '23 at 12:23
  • Also, if $f(x) = 0$, the codomain is not $(0,1)$. – Ennar Jun 03 '23 at 12:23
  • With that new definition $f(\tfrac{1}{3}) = \frac{1}{3-2} = 1 \not\in (0,1)$ – jjagmath Jun 03 '23 at 12:26
  • 1
    Oh, I see the problem. The corrected definition of your book is a bijection from $(0,1)$ to $[0,1]$, not from $[0,1]$ to $(0,1)$. – jjagmath Jun 03 '23 at 12:30
  • @jagmath

    There indeed may be an error in the text but not in the function itself. This may be a bijection from $(0, 1)$ to $[0, 1]$. I surmise this because an inverse function is also provided, which is explicitly stated to map $[0,1]$ back to $(0,1)$. Because of this, I suspect that the original function $f(x)$ maps the interval to the segment, and not the segment to the interval, as is stated in the text, which could be a transpositional error.

    – l0ner9 Jun 03 '23 at 12:31
  • just now you wrote $f(0)=0$ @l0ner9 – MathFail Jun 03 '23 at 13:07

1 Answers1

-1

The function $f$ you suggested doesn't work.

You can split the domain to the following parts: $[0,1] = (0,1)\cup \{0\} \cup \{1\}$. The interval $(0,1)$ gets sent to the interval $(0,1/4)$ since you are dividing by $4$.

Therefore, the image of your function is $(0,1/4)\cup \{ 1/2\} \cup \{1/3\} \neq (0,1)$ since you are missing almost all numbers greater than $1/4$, so your function is not surjective onto $(0,1)$.


Let me explain the idea that is at play here, but let me look for a bijection $f\colon [0,1) \to (0,1)$ instead.

It's kind of intuitive to think that these two sets differ only because the domain has one extra point so you wouldn't be wrong to think that $f$ could be chosen in such a way that it's almost always identity. The problematic part is that pesky $0$ in the domain, you can't set $f(x) = x$ on $(0,1)$ since you already used all the numbers in the codomain and you didn't send $0$ anywhere yet, so whatever you do would render $f$ non-injective.

What you do instead is to send $0$ somewhere first. Now, $f(0)\neq 0$ is used in the codomain, so you still can't define $f(x) = x$ on $(0,1)$ since both $0$ and $f(0)$ in $[0,1)$ would be sent to $f(0)$. So, you need to send $f(0)$ somewhere else, to $f(f(0)) \neq f(0)$.

Iterating the process, you get a whole sequence of different numbers $a_0 = 0$, $a_1 = f(0)$, $a_2 = f(f(0))$ and in general $a_n = f(a_{n-1})$.

Now comes the beautiful part of the above observation. Let $A = \{a_1,a_2,\ldots\}$ and $A_0 = \{a_0,a_1,a_2,\ldots\}$ defined as above. Note that $f\colon A_0 \to A$ is a bijection by construction. Also, note that $[0,1)\setminus A_0 = (0,1)\setminus A$. Therefore, you can now define $f$ to be identity on $[0,1)\setminus A_0$.

To summarize the process:

  1. Choose a sequence $\{a_1,a_2,\ldots\}$ in $(0,1)$ such that $a_n\neq a_m,\ n\neq m$.
  2. Define $a_0 = 0$.
  3. Define $f\colon [0,1)\to (0,1)$, $f(x) = \begin{cases} f(x) = a_{n+1},& x = a_n,\, n\in\mathbb N_0\\ f(x) = x,& \text{otherwise}. \end{cases}$
  4. Define $g\colon (0,1)\to [0,1)$, $g(x) = \begin{cases} g(x) = a_{n-1},& x = a_n,\, n\in\mathbb N\\ g(x) = x,& \text{otherwise}. \end{cases}$
  5. Verify that $f$ and $g$ are inverse to each other by noting that $[0,1)\setminus \{a_0,a_1,a_2,\ldots \} = (0,1)\setminus \{a_1,a_2,\ldots \}$.

I will let you work out your situation $[0,1] \to (0,1)$ where you have two extra points instead of just one.

Ennar
  • 24,364