On Kirk's asymptotic contractions (Q1765791): Difference between revisions

A selfmap \(T\) of a metric space \(X\) is an asymptotic contraction if there exist a sequence \(\phi_n: \mathbb R_+\to \mathbb R_+\) and \(\phi: \mathbb R_+\to \mathbb R_+\) such that \(d(T^nx, T^ny)\leq \phi_n(d(x, y))\) for all \(x, y \in X\) and \(n\in \mathbb N\), \(\phi_n\to\phi\) uniformly on \(\mathbb R_+\), and \(\phi(t) < t\) for all \(t > 0\). \textit{W. A. Kirk} [J. Math. Anal. Appl. 277, No. 2, 645--650 (2003; Zbl 1022.47036)] proved that an asymptotic contraction \(T\) on a complete metric space \(X\) has a unique fixed point provided that (i) each \(\phi_n\) is continuous, (ii) \(T\) is continuous, and (iii) a Picard orbit of \(T\) is bounded. The requirement (ii), viz., the continuity of \(T\) is inadvertently missing in the statement given by Kirk but the same is used in its (nonconstructive) proof. The authors show that the requirement (ii) is essential. They highlight the importance of asymptotic contractions and show that ``there exist asymptotic contractions which are not nonexpansive.'' However, the main purpose of this paper is to modify Kirk's above theorem slightly by requiring (iv) \(\lim_{t\to\infty}(t -\phi(t)) = \infty\). This requirement helps to weaken or even remove some of the assumptions from the theorem of Kirk. Indeed, the authors give a constructive proof of the following (Theorem 2): Assume that \((X, d)\) is complete and \(T\) is a uniformly continuous asymptotic \(\phi\)-contraction, where \(\phi\) satisfies the limit condition (iv). Then every Picard orbit of \(T\) converges to a unique fixed point of \(T\). Further, the authors give a complete characterization of the class of all asymptotic contractions on a compact metric space. Finally, they obtain a separation theorem for upper semicontinuous functions \(\phi\) satisfying (iv). They, besides using this result in the proof of their Theorem 2, apply a corollary of this result to compare two contractive definitions for multifunctions.