The mapping $f:[-1,1]^2\to\mathbb R^2$ is known to be continuous. Also the image of the upper edge of the rectangle is contained in the upper half-plane, the left edge's image is contained in the left half-plane, and so on for the bottom and right edge. Formally speaking this means:

If $f_1,f_2:[-1,1]^2\to\mathbb R$ are defined to be the components of $f$, i. e. $$f(x,y)=(f_1(x,y),f_2(x,y))\,,$$ then the following conditions hold

  1. for each $x\in[-1,1]$ it holds $f_2(x,1)>0$,
  2. for each $y\in[-1,1]$ it holds $f_1(-1,y)<0$,
  3. for each $x\in[-1,1]$ it holds $f_2(x,-1)<0$ and
  4. for each $y\in[-1,1]$ it holds $f_1(1,y)>0$.

Under these assumptions prove that there exist $x_0,y_0\in[-1,1]$ such that $f(x_0,y_0)=(0,0)$. This appears to be obvious intuitively, but I'm looking for a formal and rigorous proof.

  • $\begingroup$ If $f$ has two components, shouldn't it map to $\mathbb{R}^2$? $\endgroup$ – Roland Feb 5 '14 at 13:02
  • $\begingroup$ @Roland Yes, I'll correct it. Thank you. $\endgroup$ – Ralph Tandetzky Feb 5 '14 at 13:08
  • $\begingroup$ I am not so sure this is true. For example, $f$ could map continuously the unit square onto a "square annulus", which does not include the origin. I don't see any reason why such a mapping should not exist. $\endgroup$ – Giuseppe Negro Feb 5 '14 at 13:24
  • $\begingroup$ @GiuseppeNegro: It's probably impossible because of the mean value theorem or something similar. $\endgroup$ – Najib Idrissi Feb 5 '14 at 13:26
  • $\begingroup$ @GiuseppeNegro Mind the boundary conditions. How would you map onto a square annulus and still retain them? $\endgroup$ – Ralph Tandetzky Feb 5 '14 at 13:29

Here's a really high powered proof by contradiction:

Let $g:\mathbb{R}^2 \setminus (0,0) \to [-1,1]^2$ be the mapping $$ g(x,y) = \begin{cases} \frac{1}{|x|}(x,y) & |x| > |y| \\ \frac{1}{|y|} (x,y) & |x| \leq |y| \end{cases} $$ Observe that the image of $g$ is the boundary of $[-1,1]^2$, and $g$ is continuous.

Assume $f$ is as given, such that $f^{-1}((0,0)) = \emptyset$. Let $h(x,y) = g(-f(x,y))$.

By definition,

  • $h$ is continuous, as $g$ is continuous, and $-f$ is a continuous function whose image is contained in $g$'s domain.
  • $h(x,y) \neq (x,y)$ if $\max(|x|,|y|) < 1$. This is because $(x,y)$ is in the interior of the square, while $h(x,y)$ is on the boundary.
  • $h(x,y) \neq (x,y)$ if $\max(|x|,|y|) = 1$: this is because if $(x,y)$ is on the upper half plane and on the boundary, $h(x,y)$ is in the lower half plane; similarly for the other boundaries.

Thus $h:[-1,1]^2 \to [-1,1]^2$ is a continuous self-mapping of a convex, compact region in $\mathbb{R}^2$ that has no fixed points; this contradicts Brouwer's fixed point theorem.

  • 1
    $\begingroup$ Awesome! Just for clarity it would be nice to make clear that the proof is indirect at the beginning by saying something like "Assume that $f$ has no zero". $\endgroup$ – Ralph Tandetzky Feb 5 '14 at 16:22

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.