Problem: Let $I=[0,1]$ be the closed unit interval. Suppose $f$ is a continuous mapping of $I$ into $I$. Prove that $f(x)=x$ for at least one $x \in I$.


We have a known result:

Let $g$ be a continuous real valued function on a metric space $X$. Then $Z(g)=\{p \in X: g(p)=0 \}$, i.e. the zero set of $g$ is closed. [Proof is simple using "inverse image of a closed set is closed under continuous map"]

Now, we construct $F:I \to I$, such that $F(x) =f(x)-x$, which is definitely continuous, being a linear combination of two continuous functions.
[It is assumed that $F(x)=f(x)-x>0$, $\forall x\in I$. [ If $x>f(x)$, consider the function $F_1(x)=x-f(x)$ ]. Otherwise, if the function $F(x)$ changes sign somewhere within the interval, it must attain the value $0$, giving us $f(x)=x$ ].

But we know that $Z(F)$ is closed, hence it cannot be $\phi$. We are done.

Is this at all a valid proof?

Edit: Being doubtful, I write up another approach:

The function $f$ maps into $I$, i.e. itself. Hence, to be $f(x)>x$ for every value of $x$, its range set would have to exceed $I$, and the best case scenario would be the identity map, for which $f(x)=x$ for all $x$ . Otherwise, $F(x)$ must change sign.


No, it is not at all valid. The empty set is also closed.

Note that your "proof" would also apply to $F(x) = f(x) - x/2$, but there is not necessarily a solution to $f(x) = x/2$.

  • $\begingroup$ Could you please check my "proof" now? $\endgroup$ – Subhasis Biswas Jan 16 '19 at 20:47
  • $\begingroup$ Not only that, using my argument would be so disastrous, that we can "Prove" every continuous function attains zero, which is ridiculous. $\endgroup$ – Subhasis Biswas Jan 16 '19 at 20:51

No, the proof is not valid. It is not clear (nor is it necessarily true) that $F$ is a mapping from $I$ to $I$.

I would approach it this way: Letting $F$ be as you specified, it follows that $F(0) \geq 0$ while $F(1) \le 0$. Furthermore, $F$ is also continuous. Therefore by the intermediate value theorem there is an $x \in [0,1] = I$ s.t. $F(x) = 0$, or equvialently, $f(x) = x$.

  • $\begingroup$ can you check mine now? $\endgroup$ – Subhasis Biswas Jan 16 '19 at 20:45
  • $\begingroup$ Well, the writing really does need to be improved. And, the inequality $f(x) > x$ could hold for every point in $I \setminus \{1\}$. So I would advise you to try again. $\endgroup$ – Mike Jan 16 '19 at 20:50
  • $\begingroup$ The domain is compact!? So we cannot leave out the boundary point. $\endgroup$ – Subhasis Biswas Jan 16 '19 at 20:53
  • $\begingroup$ Does the result hold good in $(0,1)$? Can you help me out? $\endgroup$ – Subhasis Biswas Jan 16 '19 at 20:56
  • $\begingroup$ To answer your question, the result does not necessarily hold in $(0,1)$. For example, consider the function $f(x)= 1/2 + x/2$. The idea for showing that it does hold in $[0,1]$ is this: The function $F$ is nonnegative at $x=0$ and it is nonpositive at $x=1$. so as $F$ is also continuous, somewhere in $[0,1]$ (even if it is at precisely one endpoint $x=0$ or the other endpoint $x=1$ or at both endpoints of $I$, it must take on the value 0. $\endgroup$ – Mike Jan 16 '19 at 20:59

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.