Coincidences and fixed points of nonexpansive type multi-valued and single-valued maps (Q1901566)

The authors prove the following theorem: Let \((X,d)\) be a metric space and let \(T,f : X \to X\) satisfy, for all \(x,y \in X\), \[ \begin{multlined} d(Tx,Ty) \leq a(x,y) d(fx, fy) + b(x,y) \max \bigl\{ d(fx,Tx), d(fy,Ty) \bigr\} \\ + c(x,y) \bigl[ d(fx,Ty) + d (fy, Tx) \bigr], \end{multlined} \] where \(a(x,y) \geq 0\), \(\inf \{b(x,y) : x,y \in X\} > 0\), \(\inf \{c(x,y) : x, y \in X\} > 0\), and \(\sup \{a(x,y) + b(x,y) + 2c (x,y) : x,y \in X\} = 1\) with \(T(X) \subset f(X)\) and either (a) \(X\) is complete and \(f\) is surjective, or (b) \(X\) is complete, \(f\) is continuous and \(T\) and \(f\) are compatible (i.e. \(\lim_{n \to \infty} d(fTx_n, Tfx_n) = 0\), whenever \((x_n)\subset X\) is a sequence such that \(\lim_{n \to \infty} fx_n = \lim_{n \to \infty} Tx_n = t \in X)\), or (c) \(f(X)\) is complete, or (d) \(T(X)\) is complete. Then \(T\) and \(f\) have a coincidence point in \(X\) and \(fp = fq\), whenever \(fp = Tp\), \(fq = Tq\) \((p,q \in X)\). This theorem extends a result of \textit{L. Ćirić} [Indian J. Pure Appl. Math. 24, 145-149 (1993; Zbl 0796.47042)]. A multivalued version of this theorem is given.