Second-Order Parametric Sensitivity Analysis in NLP and Estimates by Penalty Function Methods,
Abstract
Pursuing a number of theoretical results recently obtained by Fiacco, this paper continues the development of a basis for calculating first-order changes in a Kuhn-Tucker triple and second-order changes in the optimal value function of a class of general parametric nonlinear programming problems, with respect to a perturbation of the problem parameters. Exploiting problem structure, specific formulas are derived for calculating the first partial derivatives of a Kuhn-Tucker triple. Approximations to these quantities are obtained in parallel throughout, by way of an associated logarithmic-quadratic penalty function. Applications are indicated.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 04, 1975
- Accession Number
- ADA022629
Entities
People
- Anthony V. Fiacco
- Robert L. Armacost
Organizations
- George Washington University