A nonmonotone coordinate search method for bound constrained optimization

Abstract

A new coordinate search method for bound constrained optimization isintroduced. The proposed algorithm employs coordinate directions, ina suitable way, with a nonmonotone line search for accepting the newpoint, without using derivatives of the objective function. The mainglobal convergence results are strongly based on the relationship betweenthe step length and a stationarity measure. Also, a detailed benchmarkstudy comparing dierent line search strategies is presented using a wellknown set of test problems.