Let $f: \mathbb R \to \mathbb R$ be a differentiable function such that $f(0)=1$ and $|f'(x)|<\dfrac{1}{2}$.

$i)$ Prove that there exists $x_0 \in [0,2] :f(x_0)=x_0$.

$ii)$ Let $\phi:C[\frac{1}{2},2] \to C[\frac{1}{2},2]$ defined as $\phi(g)(x)=\dfrac{(g(x))^2}{2x}$. Prove that if $x_0 \in [\frac{1}{2},2]$, then $x_0$ is also a fixed point of $D(\phi)(f)(f)$.

My attempt at a solution:

I could do almost anything of this exercise. The only thing I could get to is:

Given any two $x,y \in \mathbb R$, by the mean value theorem, we have that there is $c \in (x,y)$ $f(x)-f(y)=f'(c)(x-y)$. Then $|f(x)-f(y)|=|f'(c)||x-y|$. But $|f'(c)|<\dfrac{1}{2}$, so $|f(x)-f(y)|=|f'(c)||x-y|<\dfrac{1}{2}|x-y|$, this proves $f$ is a contraction and as $\mathbb R$ is complete, then, by the fixed point theorem, there is a unique $x \in \mathbb R: f(x)=x$. Now, if I want to prove there is a fixed point in $[0,2]$ and $x$ is the only fixed point, I must prove that $x \in [0,2]$. I don't see at all why the fact that $f(0)=1$, among the other hypothesis, would imply this.

For point $ii)$, I don't even understand well what is $D(\phi)(f)(f)$, is it the differential the function that must have the fixed point? What it means $D(\phi)(f)(f)$?, if it is the differential of the function $\phi$ evaluated at $f$, wouldn't a fixed point be in $C[\dfrac{1}{2},2]$ instead of $\mathbb R$?. Two last question: why $D(\phi)(f)(f)$ has a double $(f)(f)$? and wouldn't I have to prove first that the linear operator $\phi$ is differentiable?

  • $\begingroup$ I'm afraid most of us won't know what question ii) means. $\endgroup$ – Stefan Smith Nov 28 '13 at 4:16
  • $\begingroup$ Neither do I, I'll ask my classmates and if someone clarifies it, I'll write it here. $\endgroup$ – user100106 Nov 28 '13 at 20:01

Regarding part i, here is one approach: consider the function $g(x) = f(x) - x$. We note that $g(0)>0$, and (using the mean value theorem) you can show that $g(2) < 0$. We now know that there is a $c \in (0,2)$ for which $g(c) = f(c) - c = 0$. This is your (unique) fixed point.

As for the second part, would $D(\phi)(f)(f)(x)$ be $[D\phi](f(f(x)),$ or $D[\phi \circ f \circ f](x),$ or is it perhaps $D[\phi \circ f](f(x))$? Honestly, I'm not sure how to make sense of the notation you've used here.

  • $\begingroup$ Thanks, you didn't even have to use fixed point theorem, your proof (sketch of proof) is much more simpler that what I was trying to do. $\endgroup$ – user100106 Nov 28 '13 at 16:12
  • $\begingroup$ If you do use my proof, you would have to prove that this fixed point is unique; you can do so using the mean value theorem to show that if there are two distinct fixed points, then there is a $c$ between them for which $f'(c) = 1>\frac 12$. $\endgroup$ – Omnomnomnom Nov 28 '13 at 18:26
  • $\begingroup$ Ok, but the problem says to find a fixed point, it's sufficient to prove there is one, it doesn't ask for unicity. $\endgroup$ – user100106 Nov 28 '13 at 20:00

Brouwer's fixed-point theorem http://en.wikipedia.org/wiki/Brouwer_fixed-point_theorem tells you that any continuous map from a compact, convex subset of Euclidean plane has a fixed point. Now, consider the restriction of $f$ to the interval $[0,2]$. We show that $f([0,2] \subset [0,2]$, and then the existence of the fixed point follows. By the mean-value theorem:

$-1/2 <\frac {f(2)-f(0)}{2-0}=\frac {f(2)-1}{2} < 1/2 $ , so that $f(x)<2 $ on $[0,2]$; for any $x$ in $[0,2]$use: $$ -1/2<\frac{f(x)-1}{x}<1/2$$ , so that $$\frac{3-x}{2}<f(x)<\frac {1+x}{2} $$ . This guarantees that the restriction of $f(x)$ to $[0,2]$ is a map from the compact space $[0,2]$ to itself, and so Brouwer's fixed point theorem guarantees the existence of a fixed point.


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.