Lancelot

This book, which is concerned with algorithms for solving large-scale non-linear optimization problems, is the only complete source of documentation for the software package Lancelot and will mainly be used as a manual in conjunction with the software package. However, it is not only a reference to the input format but also to the underlying algorithms. It is the first publication of this kind for large-scale general non-linear problems. Numerous figures and tables enhance its usefulness to both novices and experts. In addition, it also gives valuable guidance to future designers of related software.
Lancelot ist ein Softwarepaket zur Lösung von umfangreichen nichtlinearen Optimierungsaufgaben. Das Buch gibt einen umfassenden Überblick über das Paket und seine Anwendungen. Dabei wird im Detail beschrieben, wie Beispiele zur Bearbeitung mit dem Paket dargestellt werden können und wie der Algorithmus versucht, diese zu lösen. Auch werden verschiedene technische Fragen behandelt, die für die Implementierung der Software relevant sind. Das Buch wird für Forscher und Anwender im Gebiet der nichtlinearen Programmierung von großem Nutzen sein. Zwar wird in erster Linie ein spezielles Optimierungspaket behandelt, doch sind die diskutierten Fragestellungen von allgemeinerer Bedeutung für die Entwicklung und Implementierung von umfangreichen Optimierungsalgorithmen. Insbesondere wird ein standardisiertes Eingabeformat für Optimierungsaufgaben und das Lancelot-Paket vorgestellt.

---

LANCELOT is a software package for solving large-scale nonlinear optimization problems. This book is our attempt to provide a coherent overview of the package and its use. This includes details of how one might present examples to the package, how the algorithm tries to solve these examples and various technical issues which may be useful to implementors of the software. We hope this book will be of use to both researchers and practitioners in nonlinear programming. Although the book is primarily concerned with a specific optimization package, the issues discussed have much wider implications for the design and im plementation of large-scale optimization algorithms. In particular, the book contains a proposal for a standard input format for large-scale optimization problems. This proposal is at the heart of the interface between a user's problem and the LANCE LOT optimization package. Furthermore, a large collection of over five hundred test ex amples has already been written in this format and will shortly be available to those who wish to use them. We would like to thank the many people and organizations who supported us in our enterprise. We first acknowledge the support provided by our employers, namely the the Facultes Universitaires Notre-Dame de la Paix (Namur, Belgium), Harwell Laboratory (UK), IBM Corporation (USA), Rutherford Appleton Laboratory (UK) and the University of Waterloo (Canada). We are grateful for the support we obtained from NSERC (Canada), NATO and AMOCO (UK).