Global Optimization Method Applying Chaos in a Multitrajectory Inertial System

摘要

In conventional global optimization techniques, such as the trajectory method, the usual procedure is to start the trajectory from an initial point and try an optimization. In contrast, this paper proposes a model in which the trajecto- ries are started in parallel from multiple initial points, and global optimization is achieved by providing interactions among the trajectories. It is shown by calculating the Lyapunov index that the trajectory becomes chaotic in a Conservative system unless annealing is applied.