No label defined (Q4391237)

The purpose of the paper is to develop the convergence theory for a group of dual modified Lagrangian-type methods for solving multicriteria nonlinear programming problems with equality and inequality constraints. The authors study the problem of finding \(\min_{x\in X} f(x)\), where \(X:=\{x\in{Q}\subset R^{n}:g(x)\leq 0\), \(h(x)=0\}\), \(f(x)=(f_0(x),f_1(x),\cdots f_m(x))\), \(Q\) is a nonempty convex and compact set defined by simple constraints. The coordinate functions \(f_{i}(x)\) and \(g_{j}(x)\) are assumed to be convex. The algorithms under consideration are based on the use of an aggregate function \(M_{k}(f(x),g(x),h(x))=F(f(x),\mu_{k})+G(g(x),\tau_{k})+H(h(x), \omega_{k})\) at the \(k\)-th iteration. The paper is divided into three sections. The first section is devoted to standard nonlinear programming problems having only inequality constraints. The authors propose the iterative process: \(x_{k}\in \text{Argmin}_{x\in Q}M_{k}(f_0(x),g(x))\) and derive sufficient conditions for \(\{x_k\}\) to converge to a solution. The second section deals with computational algorithms for solving nonlinear programming problems having equality constraints. Under perfectly restrictive smoothness assumptions on the sensitivity function \(F(y)=\min\{f_0(x):h(x)=y,x\in Q\}\), the authors prove the global convergence of the proposed algorithms. In the third section they consider a general multicriteria optimization problem. They prove the convergence of iterations to a weakly effective solution for the original problem. The paper ends with a brief heuristic discussion of the efficiency of the constructed computational schemes as compared with some known sequential unconstrained minimization methods.