Numerical Methods for Constrained and Unconstrained Optimization.

Abstract

The main thrust of the research has been toward the development of efficient algorithms for solving the finite - dimensional constrained optimization problem. Historically, problems of this type have been solved by either penalty function methods or through linearization procedures. The fact that neither of these techniques is completely satisfactory for general nonlinear problems has lead to a concentrated research effort to find better approaches. What has so far emerged from this work is a blending of the penalty function land linearization ideas with the quadratic approximation methods associated with unconstrained optimization. While there remain many unresolved issues, it is now apparent that this synthesis has resulted in more efficient algorithms for the nonlinear constrained optimization problem. (Author)

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Dec 23, 1982
Accession Number
ADA122912

Entities

People

  • Jon W. Tolle
  • Paul T. Boggs

Organizations

  • University of North Carolina at Chapel Hill

Tags

DTIC Thesaurus Topics

  • Algorithms
  • Convergence
  • Curriculum
  • Differential Equations
  • Equations
  • Lagrangian Functions
  • Linear Programming
  • Mathematical Programming
  • Mathematics
  • Military Research
  • North Carolina
  • Numerical Analysis
  • Operations Research
  • Optimization
  • Quadratic Programming
  • Systems Analysis

Readers

  • Operations Research