Minimization of isotonic functions composed of Fractions (Q613584)

The authors consider a special class of optimization problems of the type \[ \mathrm{minimize } \mathbb I \left(\frac{f_1(x)}{g_1(x)},\frac{f_2(x)}{g_2(x)},\dots,\frac{f_n(x)}{g_n(x)}\right) \mathrm{ subject to } x \in X, \tag{P} \] where \(\mathbb I : \mathbb R ^n \to \mathbb R\) is a continuous function, which is isotonic with respect to the componentwise ordering, i.e., \(\mathbb I (v) \geq \mathbb I (v')\) for all \(v, v' \in \mathbb R^n\) such that \(v - v' \in \mathbb R^n_+\), while for \(i \in \{1,\dots,n\}\) the functions \(f_i : X \to \mathbb R\) and \(g_i : X \to \mathbb R\) are continuous on a nonempty compact set \(X \subset \mathbb R^m\) such that \(g_i(x) > 0\) for all \(x \in X\). Several well-known problems, such as the sum-of-ratios minimization or min-max fractional programming, can be recovered as particular instances of (P) for certain isotonic functions \(\mathbb I\). The principal aim of this paper is to develop a generic Dinkelbach-like algorithm for the numerical solution of problem (P).