A nonmonotone descent direction-based method for multi-objective optimizationproblems

In this work, a nonmonotone inexact Armijo line search isapplied for step length of descent direction-basedalgorithms in solving unconstrained multi-objectiveoptimization problems. The result leads to nonmonotoneversions of gradient or steepest descent, Newton andquasi-Newton methods. Two types of monotone andnonmonotone approaches are compared to investigate theefficiency of the algorithms, using some numericalexperiments, with respect to CPU time and norm ofgradient of objective functions.

کلیدواژه ها:

Multi-objective optimization problems ، Newton ، Nonmonotone ، Quasi-Newton.