1
$\begingroup$

Let $X$ be a Banach space and $K$ its non-empty subset. Let $S:K\to X$ be a single-valued mapping. Then $S$ satisfies the condition $(C_\lambda)$ if for $x,y\in K$ and $\lambda\in(0,1)$, $\lambda\|x-Sx\|\leq \|x-y\|\implies \|Sx-Sy\|\leq \|x-y\|$

I understand the second condition to mean $S$ is a nonexpansive mapping. But I don't understand this whole condition.

From the first condition, we can extrapolate and get $\lambda\|y-Sy\|\leq \|x-y\|$. Adding, we get $\lambda(\|x-Sx\|+\|y-Sy\|)\leq 2\|x-y\|$, or $\frac{\lambda}{2}(\|x-Sx\|+\|y-Sy\|)\leq \|x-y\|$. If this condition implies that $\|Sx-Sy\|\leq \|x-y\|$, then is the actual statement of this condition this: $$\|Sx-Sy\|\leq \frac{\lambda}{2}(\|x-Sx\|+\|y-Sy\|)$$

And if the actual meaning of the condition $C_\lambda$ is indeed this, why don't we write it this way?

Thanks in advance!

$\endgroup$
3
  • $\begingroup$ I do not really get the question. What on earth should be the "actual meaning" of a condition? To me it sounds like the definition you gave says something like "if any point $x$ does not move too far under $S$, then $S$ is a non-expansive mapping". This sounds reasonable to me. I you have a good interpretation or your "derived inequality" that is also fine. I can imagine that both formulations may be very helpful in certain situations. $\endgroup$
    – Dirk
    Sep 2, 2013 at 6:34
  • $\begingroup$ It's just that I'm not sure whether my interpretation is implied by the given condition. $\endgroup$
    – user92664
    Sep 2, 2013 at 6:39
  • $\begingroup$ @92664: Maybe you could add some information: who introduced this condition $C_\lambda$, and what's the use of it? $\endgroup$ Sep 2, 2013 at 7:37

0

Browse other questions tagged or ask your own question.