Let $(f_n)_n$ be a sequence of functions $f_n: [0,1] \rightarrow [0,1]$ that (pointwise) converges to a function $f$. Suppose that all $f_n$ are contractions, prove that $f$ is a contraction as well.

Here's what I have:

$$ \begin{array}{rl} | f(x) - f(y) | &= | \lim_{n \rightarrow \infty}f_n(x) - \lim_{n \rightarrow \infty}f_n(y) |\\ &= | \lim_{n \rightarrow \infty}(f_n(x) - f_n(y)) |\\ &= \lim_{n \rightarrow \infty}|f_n(x) - f_n(y) |\\ &< \lim_{n\rightarrow \infty}|x-y| &= |x-y| \end{array} $$

That's a bunch of symbols. I fear that I might have waved my hands too much at the details. I can't point to any theorems that tell my that I'm alowed to proceed with the third equality.

Is this correct and why? Also, why does it matter that the $f_n$ are defined as $f_n: [0,1] \rightarrow [0,1]$?

(The part of the question is: "prove that the convergence is uniform")

  • $\begingroup$ Haven't you missed to include a Lipschitz constant $k<1$ in the definition of contraction? In that case the constants can be different for each $f_n$ and might converge to $1$. Then the conclusion does not hold. Perhaps you have a different definition of contraction than I'm used to.The one I'm used to is the same as stated on Wikipedia: en.wikipedia.org/wiki/Contraction_mapping $\endgroup$ – dioid Jun 3 '15 at 13:47
  • $\begingroup$ I'm using this one: (It seems equivalent) A function $f$ is a contraction if this holds: $\forall x,y : |f(x)-f(y)| < |x-y|$ $\endgroup$ – Syd Kerckhove Jun 3 '15 at 13:49
  • 1
    $\begingroup$ They're not equivalent. According to the definition on Wikipedia the function $f(x) = x$ is not a contraction. According to your definition no function is a contraction (let $x=y$). If you add $x \neq y$ to your definition they are still not equivalent since the definition on Wikipedia says you have to stay away by some fixed distance from $|x-y|$ while your (modified) definition allows this distance to depend on $x, y$ and become arbitrarily small. $\endgroup$ – dioid Jun 3 '15 at 13:53
  • $\begingroup$ Oh, I made a mistake, the definition in our notes is $\forall x,y:\ |f(x)-f(y)| \le |x-y|$ (so $\le$ instead of $<$). Also the notes define a strict contraction as $\exists c\in [0,1[, \forall x,y:\ |f(x)-f(y)| \le c|x-y|$ $\endgroup$ – Syd Kerckhove Jun 3 '15 at 13:55
  • 1
    $\begingroup$ Ok, you're all good then. The terminology I'm used to is to call it non-expansion and contraction instead of contraction and strict contraction. $\endgroup$ – dioid Jun 3 '15 at 13:57

You can swap $|\cdot|$ and $\lim_{n \to +\infty}$ because $|\cdot|$ is a continuous function. But when you take the limit, $<$ becomes $\leq$. It is not essential that the functions are from $[0,1]$ to itself here - however, fixed point theorems are stated with $[0,1]$ at first for simplicity. You can look at Brouwer's, Banach's fixed point theorems, etc.

  • $\begingroup$ Wait a minute, are you saying the proof is correct? (If I add the necessary explanation?) $\endgroup$ – Syd Kerckhove Jun 3 '15 at 13:29
  • $\begingroup$ Yes, although I would have stated it like this: $$|f_n(x)-f_n(y)|< |x-y|,~\forall\,n \implies \lim_{n \to +\infty}|f_n(x)-f_n(y)| \leq |x-y| \implies |f(x)-f(y)| \leq |x-y|.$$ $\endgroup$ – Ivo Terek Jun 3 '15 at 13:31

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.