Two problems with GGP generalized geometric programming algorithm

✍ Scribed by Yeh, Alexander

• Publisher: John Wiley and Sons
• Year: 1993
• Tongue: English
• Weight: 386 KB
• Volume: 9
• Category: Article
• ISSN: 1069-8299
• DOI: 10.1002/cnm.1640090907

✦ Synopsis

The generalized geometric programming algorithm GGP has been found to be one of the better algorithms for optimizing algebraic functions subject to algebraic constraints. The paper discusses two problems with the GGP algorithm presented by Avriel et al. (1980). The first problem is that, because of the way that constraint tolerances are checked. the checks may pass even when a constraint is violated beyond the acceptable tolerance. This problem can affect an acceleration technique suggested for the algorithm. This paper presents two possible solutions to this problem. The second problem is the algorithm's weakness in handling equality constraints. This problem seems to especially affect the 'Phase I' part of the algorithm. which finds a feasible solution from a random starting point. While extensions have been made to get around this limitation, no discussion on why the original algorithm has problems with equality constraints has been made. This paper discusses why the problems exist.