A DUAL DECOMPOSITION PRINCIPLE

Abstract

A decomposition principle for linear programming is presented. The technique may be viewed as a dual of the Dantzig - Wolfe decomposition principle for linear programs. The program matrix in what we may call the basic problem is considered as having many (an infinite number of) columns. One visualizes a basic problem, in primal form, where, the set of permissible columns is not finite, as in the usual primal form, but is a given convex polyhedron. The basic problem is solved by the modified simplex method, but at each iteration the column to enter the current basis emerges as the solution to an auxiliary linear program and is, in fact, an extreme point of the given convex polyhedron.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 04, 1961
Accession Number
AD0269699

Entities

People

  • C. E. Lemke
  • T. J. Powers

Organizations

  • Rensselaer Polytechnic Institute

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Algorithms
  • Computational Science
  • Computations
  • Computer Programming
  • Computer Programs
  • Computers
  • Convex Sets
  • Decomposition
  • Digital Computers
  • Government Procurement
  • Iterations
  • Linear Programming
  • Military Research
  • Operations Research
  • Simplex Method

Fields of Study

  • Mathematics

Readers

  • Operations Research