Competitive Algorithms Abstracts Page Replacement for General Caching Problems
Decision making can be considered in two different contexts: making decisions with complete information, and making decisions based on partial information. A major reason for the study of algorithms is to try to answer the question: 'Which is the better algorithm?' The study of the computational complexity of algorithms is useful for distinguishing the quality of algorithms based on the computational resources used and the quality of the solution they compute. However, the computational
... mputational complexity of algorithms may be irrelevant or a secondary issue when dealing with algorithms that operate in a state of uncertainty. 'Competitive' analysis of algorithms has been developed in the study of such algorithms. Competitive analysis is useful in the analysis of systems that have some notion of a time progression, that have an environment, that respond in some way to changes in the environment, and that have a memory state. Competitive analysis is used for so-called 'on-line algorithms' that have to respond to events over time. Competitive analysis is used whenever the nature of the problem is such that decisions have to be made with incomplete information. The Dagstuhl meeting on Competitive Algorithms brought together researchers with affiliations in and the USA. 36 presentations were given. The abstracts of most of these presentations are contained in this seminar report. Moreover, there is a list with some open problems that were mentioned in the open problem session on Wednesday evening. As always, we enjoyed the social aspects of a Dagstuhl meeting. As always, we had a wine-and-cheese party every evening. And as always, we are looking forward to the next Dagstuhl workshop.