I cannot seem to find a contraction factor such that $$Tx = \frac{x}{2}+\frac{1}{x}$$ is a contraction on the whole set $[1,\infty)$ in the complete normed space $(\mathbb{R}, |\cdot|)$.

My argument for $x,y\in [1,\infty)$:

\begin{align} d(Tx, Ty) &= \left|\frac{x}{2}+\frac{1}{x} - (\frac{y}{2}+\frac{1}{y})\right| \newline \newline &= \left|\frac{x-y}{2}+\frac{1}{x} - \frac{1}{y} \right| \newline \newline &=\left|\frac{x-y}{2}+\frac{y-x}{xy} \right| \newline \newline &\leq\left|\frac{x-y}{2}\right| +\left|\frac{x-y}{xy} \right| \newline \newline &\leq\left|\frac{x-y}{2}\right| +\left|\frac{x-y}{2} \right|,\quad \forall x,y \geq \sqrt{2} \newline \newline &=\frac{1}{2}|x-y| +\frac{1}{2}|x-y| \newline \newline &= |x-y|. \end{align}

Hence, restricting $T$ to $[\sqrt{2},\infty)$ yields a non-expansive mapping. $T$ also has a fixed point at $x=\sqrt{2}$. However, the book in which the exercise was found states that $T$ is a contraction with a minimal contraction factor $\lambda \in [0,1)$. Can I take better estimates to show that $T$ is a contraction on $[1,\infty)$?

  • 1
    $\begingroup$ Just before you apply your inequality, have you tried putting over a common denominator? $\endgroup$ – Patrick Da Silva Jan 18 '12 at 4:57
  • 1
    $\begingroup$ The mean value theorem is nice here too. $\endgroup$ – Ragib Zaman Jan 18 '12 at 5:01
  • $\begingroup$ All great ideas, thank you. $\endgroup$ – Henry Shearman Jan 18 '12 at 5:04

When you wrote

$$\left|\frac{x-y}{2}+\frac{y-x}{xy} \right| \leq\left|\frac{x-y}{2}\right| +\left|\frac{x-y}{xy} \right|,$$

you didn't take advantage of the fact that $x-y$ and $y-x$ have opposite signs, so there is cancellation that will give you a better estimate. Instead you could note that

$$\left|\frac{x-y}{2}+\frac{y-x}{xy} \right|=|x-y|\left|\frac{1}{2}-\frac{1}{xy}\right|,$$

and that $$-\frac{1}{2}\leq \frac{1}{2}-\frac{1}{xy}<\frac{1}{2}$$

when $x,y\geq 1$.

  • $\begingroup$ Heh, nice follow-up of my comment =) But I wouldn't be surprised if you just thought about it yourself. +1 $\endgroup$ – Patrick Da Silva Jan 18 '12 at 5:03
  • $\begingroup$ Thanks Patrick. Indeed I saw your comment after I posted my answer (I can't type that fast!), but I'm not sure it would be a follow up, as I did not follow your approach of finding a common denominator. $\endgroup$ – Jonas Meyer Jan 18 '12 at 5:05

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.