convex separable minimization problems with a linear constraint and bounded variables一个线性约束凸可分极小化问题和有界变量.pdfVIP

CONVEX SEPARABLE MINIMIZATION PROBLEMS WITH A LINEAR CONSTRAINT AND BOUNDED VARIABLES STEFAN M. STEFANOV Received 30 June 2004 and in revised form 19 April 2005 Consider the minimization problem with a convex separable objective function over a feasible region defined by linear equality constraint(s)/linear inequality constraint of the form “greater than or equal to” and bounds on the variables. A necessary and sufficient condition and a sufficient condition are proved for a feasible solution to be an optimal so- lution to these two problems, respectively. Iterative algorithms of polynomial complexity for solving such problems are suggested and convergence of these algorithms is proved. Some convex functions, important for problems under consideration, as well as compu- tational results are presented. 1. Introduction In many cases, we have to minimize a convex separable function over a region defined by a linear equality or inequality “ ≥” constraint with positive coefficients, and two-sided bounds on the variables. Such problems and problems related to them arise, for example, in production plan- ning and scheduling [2] and Problem 4, Section 5, in allocation of resources [2, 31] and Problem 1, Section 5, in allocation of effort resources among competing activities [16] and Problems 3, 5, and 6, Section 5, in the theory of search [6], in subgradient opti- mization [ 11], in facility location [24], and in the implementation of projection methods when the feasible region is of the considered form [28] and Problem 2, Section 5, and so forth. The problems under consideration can mathematically be formulated as follows: min c(x) = cj xj (1.1) j ∈J subject to d x =α,