# Continuous function from $(0,1)$ onto $[0,1]$

While revising, I came across this question(s):

A) Is there a continuous function from $(0,1)$ onto $[0,1]$?

B) Is there a continuous one-to-one function from $(0,1)$ onto $[0,1]$?

(clarification: one-to-one is taken as a synonym for injective)

I figured the answer to A is yes, with $\frac{1}{2}\sin(4\pi x)+\frac{1}{2}$ as an example.

The answer to part B is no, but what is the reason?

Sincere thanks for any help.

• – Rudy the Reindeer Jul 7 '12 at 9:35
• You can avoid confusion by using "surjective" and "injective" instead of "onto" and "one-to-one" respectively. – Rudy the Reindeer Jul 7 '12 at 10:09
• @MattN.: the answer to that question does not answer this question. – robjohn Jul 7 '12 at 10:15
• @Asaf: True, but the OP has already gotten to a continuous surjection, so the question is really only (B). – Cameron Buie Jul 7 '12 at 11:58
• @Asaf: Fair point. If it is closed, I will vote to reopen. – Cameron Buie Jul 7 '12 at 12:13

## 5 Answers

There isn't such a function. If $f$ is such a function, then $f$ is monotonic, and $f^{-1}$ too. But a monotonic function defined on an interval is continuous iff its range is an interval. So we have that $f$ is an homeomorphism from $(0,1)$ to $[0,1]$, which is impossible since one is compact and the other is non compact.

• 2.$$f(x) = \left\{ \begin{array}{cl} 0 & x < \frac14 \\ 2x-\frac12 & \frac14 \leq x \leq \frac34 \\ 1 & x > \frac34 \end{array}\right.$$ – Unknown x Sep 3 '18 at 1:25
• Isn't a continuous function – Unknown x Sep 3 '18 at 1:26

A slightly simpler solution, perhaps, for the first part would be as follows:

Consider the function from $\left[\frac14,\frac34\right]$ to $[0,1]$ defined by $x\mapsto 2\left(x-\frac14\right)$. This is certainly a continuous function, and it is certainly onto $[0,1]$.

Define now:

$$f(x)=\begin{cases}0 & x<\frac14\\ 2\left(x-\tfrac14\right) & x\in\left[\tfrac14,\tfrac34\right]\\ 1 & x>\frac34\end{cases}$$

Of course this is not an injective function, but it is continuous and onto, as required.

• Was not the OP asking about a $1-1$ function? – Giovanni De Gaetano Jul 7 '12 at 9:37
• @GiovanniDeGaetano: The first part reads differently. – Asaf Karagila Jul 7 '12 at 9:38
• I'm sorry, I must have misread the question then. – Giovanni De Gaetano Jul 7 '12 at 11:48

Your example for A is fine.
For B "for a continuous function on an interval, being one-to-one is equivalent to being increasing throughout or decreasing throughout" so that no such function exists. (see (2.5)-(2.6) of this paper)

Answer for (B): Continuous image of a compact set is compact. [see1]

Suppose $f: [0,1] \rightarrow (0,1)$ is a bijection and continuous. Then $f([0,1])=(0,1)$, which is compact by above theorem.

But in $\mathbb{R}$, the closed interval $[0,1]$ is compact; but the open interval $(0,1)$ is not compact.

Therefore if such $f$ does not exist, then $f^{-1}$ does not exist too, where $f^{-1}: (0,1) \rightarrow [0,1]$. (your question in (B))

If such a function exists then (0,1) and [0,1] are homeomorphic. But if you delete 0 from [0,1] you get (0,1] which is connected. Deleting the point x in (0,1) which corresponds to 0 in [0,1] leaves you with (0,x) u (x,1) which is not connected ---> contradiction.

• Any reason why you're posting a answer to a 3-year old question that already has an accepted answer? – Olorun Oct 2 '15 at 5:46
• No particular reason. – Silent Duckling Oct 10 '15 at 12:27
• Unfortunately this answer is incomplete. "Such a function" would be a continuous bijection from $(0,1)$ to $[0,1]$. The matter of the question is to show that any continuous bijection from $(0,1)$ to $[0,1]$ is an open mapping, and thus a homeomorphism. – Pete L. Clark Nov 29 '15 at 8:35

## protected by Zev ChonolesOct 2 '15 at 6:23

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?