Best quasi-convex uniform approximation (Q805957): Difference between revisions

Let B be the linear space of all bounded real functions f on [0,1] with the norm \(\| f\|_{\infty}=\sup_{x\in [0,1]}| f(x)|.\) The function \(g\in B\) is quasi-convex if g(x)\(\leq \max \{g(s),g(t)\}\) for all \(0\leq s\leq x\leq t\leq 1\). Let \(K\subset B\) denote the set of all quasi-convex functions on [0,1]. \textit{V. A. Ubhaya} [J. Math. Anal. Appl. 116, 439-449 (1986; Zbl 0595.90099)] proved that \(g\in B\) is quasi-convex if and only if there exists a point \(p\in [0,1]\) such that either ``g is nonincreasing on [0,p) and is nondecreasing on [p,1]'' or the same holds on [0,p] and (p,1], respectively. In both cases it is said that the point p is a knot of g. For \(f\in C[0,1]\) let \[ G(f)=\{g^*\in K:\;\| f- g^*\|_{\infty}=\inf_{g\in K}\{\| f-g\| \}\} \] be the set of best quasi-convex approximations to f and let \(p^*=\{p\in [0,1]:\) p is a knot for some \(g^*\in G\}\) be the set of optimal knots of f. The authors characterize both the best quasi-convex approximations and the optimal knots.