Curve search method for constrained global optimization

In this paper, we propose a revised method of "line search" for optimization problems with nonlinear constraints - "curve search" method. The discrete dynamical system based on a nonlinearly constrained gradient dynamics let us search along with a curve, which can implement diverse search without invading the original constraints. We also propose an algorithm which applies the curve search to global optimization. The performance of our proposing method is examined through some simulation results.