On convex iterative roots of non-monotonic mappings (Q2407706)

Let \(I\) be an interval. A non-monotonic convex function \(f: I \to I\) is considered. The author proves that: {\parindent=6mm \begin{itemize}\item[(a)] If \(f^2\) is convex, then \(f^n\) is convex for \(n \geq 2\). \item[(b)] \(f^2\) is convex if and only if there exists \(x_0 \in I\), where \(f\) attains the local minimum so that \(f(x_0) \geq x_0\) or \(f\) is strictly decreasing. \end{itemize}} The function \(f\) is iteratively convex if \(f\) has convex iterative roots of \(n-th\) order for \(n \geq 2\). The function \(f\) is dyadically convex if for every \(n \geq 1\) there exists a convex iterative root \(f^{\frac{1}{2^n}}\) of \(2^n-th\) order such that \(f^{\frac{1}{2^n}} \circ f^{\frac{1}{2^n}} = f^{\frac{1}{2^{n-1}}}\) for \(n \geq 1\). The function \(f\) is flowly convex if there exists an iteration semigroup \(\{f^t: t > 0\}\) of \(f\) such that functions \(f^t\) are convex. The relations among these three types of convexity are studied.