Convergence of multi-step curve search method for unconstrained optimization

Abstract

A new multi-step curve search method for unconstrained minimization problems is proposed. The convergence of the algorithm is proved under some mild conditions. The linear convergence rate is also investigated when the objective function is uniformly convex. This method uses previous multi-step iterative information and curve search rule to generate new iterative points. Using more previous iterative information and curve search rule can make the new method converge more stably than traditional descent methods and be suitable to solve large scale problems.