Nonmonotone Coordinate Search Method for Bound Constrained Optimization
Frau, J. A. and Pilotta, E. A.
Corresponding Email: elvio.pilotta@unc.edu.ar
Received date: 22 May 2020
Accepted date: 12 August 2020
Abstract:
A new coordinate search method for bound constrained optimization is introduced. The proposed algorithm employs coordinate directions, in a suitable way, with a nonmonotone line search for accepting the new point, without using derivatives of the objective function. The main
global convergence results are strongly based on the relationship between the step length and a stationarity measure. Also, a detailed benchmark study comparing different line search strategies is presented using a well-known set of test problems.
Keywords: Pattern search methods, bound constrained optimization, global convergence, nonmonotone line search, numerical experiments