Sensitivity Analysis Procedures for Geometric Programs: Computational Aspects

Publisher Website
Google Scholar
Abstract
Computational procedures for sensitivity analysis in geometric programming are developed and implemented. As a result of this research computationally feasible methods for parameterizing the objective function coefficients and right-hand sides of the constraints of the primal program may be identified. These procedures, while applied to the parameters of the primal program, are based on the dual geometric program. However, the results of this paper are independent of a particular optimization technique. In addition to comparing these numerical results with the optimal solution, a measure of error for the approximation is developed which does not depend on computation of the optimal solution.