Applications of interval arithmetic in non-smooth global optimization (Q1405071)

An expansion scheme for nonlinear, at least continuous functions \(f: D\subset\mathbb{R}^n\to \mathbb{R}\) is presented with which one can provide enclosures of the range of \(f\). The scheme uses real compact intervals and slopes. It is applied in a branch-and-bound algorithm in global optimization in order to enclose the global minimum \(f^*= \min_{x\in X^0} f(x)\) on a given interval box \(X^0\subseteq D\) and to enclose the corresponding minimum points. The algorithm employs also a special pruning test to cut away obsolute subboxes which have been proved not to contain a minimum point. A variety of examples illustrates the efficiency of the algorithm followed by some concluding remarks.