Multidimensional bisection applied to global optimisation (Q1174537): Difference between revisions

The paper may be viewed as a continuation of another paper of the author [The bisection method in higher dimensions (to appear)]. Relevant results from that paper are adapted and applied to the global optimization problem for Lipschitz continuous functions in many variables; consequently, a family of deterministic algorithms is established. Every member of the family begins with bracketing all global optima over a compact subset \(K\) in \(R^ n\) in a simplex, and, at each iteration, strips away regions from this initial simplex by using two processes, reduction and elimination, in such a way that at the end of an iteration the current bracket is a union of all its predecessors. Under certain conditions, this decreasing sequence of brackets will converge to the set of global optima of the function over the compact set. In addition to convergence and optimality questions, acceleration techniques are also considered.