scientific article; zbMATH DE number 3852799

From MaRDI portal
Publication:3321844





[For the entire collection see Zbl 0523.00019.] The problem considered in this paper is a strictly convex separable quadratic program with capacitated transportation constraints. Problems of this sort arise in Economics, Statistics, Telecommunications, Transportation/Regional Science, among others. The object is to estimate (in a least-squares sense) the matrix with prescribed row and column sums and entries within prescribed intervals that lies closest to a given matrix. The algorithm studied in this paper uses the structure of the problem and the (necessary and sufficient) first-order optimality conditions to reduce the quadratic program to a linear complementarity problem in the dual variables. (This problem is equivalent to a particular dual quadratic program.) The algorithm is iterative - as distinct from finite - and uses the method of Block Successive Overrelaxation (BSOR). Computational results on a set of problems related to the German National Economy are reported. Note: The convergence method is treated in a separate paper by the author and J.-S. Pang.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3321844)