Chaotic Behavior in a New Class of Parallel Optimization Algorithms

摘要

A new class of optimization algorithms for linear and nonlinear problems is proposed. The algorithms are based on sigmoidic updatings and can be looked upon as nonlinear multidimensional maps. The bifurcation and chaotic regimes of these maps are analyzed and their possible application to optimization problems are indicated. 9 refs., 3 figs.