An Evaluation and Comparison of Three Nonlinear Programming Codes
Abstract
This study evaluates and compares the production use of three nonlinear programming codes. The three codes and their developers are: SUMT by W. C. Mylander, R. L. Holmes and G. P. McCormick, GRG by L. S. Lasdon, A. D. Waren, M. W. Ratner and A. Jain, and GRAVES by G. W. Graves. This is the first computer comparison of these three particular codes. Each code was evaluated with respect to the time and sophistication required of the user and the degree of mandatory or potential interaction between the code and the analyst. The comparison criteria were accuracy, robustness, efficiency and ease of utilization. Eight current and realistic test problems employing from 9-100 variables and 2-20 constraints were used. The results revealed that no single code was superior or inferior in all aspects. The choice of an optimal code among these three would be dependent upon the problems to be solved, the ability of the analyst and the desire of the analyst to alter the code for his own purposes.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1976
- Accession Number
- ADA024930
Entities
People
- Ralph J. Waterman
Organizations
- Naval Postgraduate School