COMPUTING EXPERIENCE WITH LINEAR PROGRAMMING AND ITS VARIANTS

Abstract

The application of linear programming to complex problems gives rise to the handling of large matrix problems. The simplex method proved to be the most expedient of any tried but, since this is essentially Gaussian elimination, numerical techniques for handling it have definite limitations, both with regard to running time and memory size. RAND's simplex codes for the IBM 701 were designed, through revisions to the method and special compiling routines, to maintain great precision and facilitate checking, to be adaptable to many variations, and to be as automatic in operation as possible. Several quite large models of considerable variety are discussed along with the prospects for future improvements, both in theory and in coding techniques.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Aug 08, 1955
Accession Number
AD0604717

Entities

People

  • William Orchard-hays

Organizations

  • RAND Corporation

Tags

Communities of Interest

  • Materials and Manufacturing Processes

DTIC Thesaurus Topics

  • Algorithms
  • Automatic
  • Computations
  • Computer Programming
  • Elimination
  • Equations
  • Error Analysis
  • Errors
  • Floating Point Operations
  • Heuristic Methods
  • Iterations
  • Linear Programming
  • Linear Systems
  • Mathematical Models
  • Models
  • Petroleum
  • Simplex Method

Readers

  • Approximation Theory.
  • Computer Science.
  • Systems Analysis and Design