The Tenth Seminar on the
Analysis of Algorithms

MSRI, California, USA
June 14 - June 18, 2004


Design by Mariola Szpankowska


This is the tenth seminar specifically dedicated to the Analysis of Algorithms. The previous meeting were held in Schloss Dagstuhl (Germany) in 1993, 1995, and 1997, in Princeton (USA) in 1998, in Barcelona (Spain) in 1999, Krynica Morska, Poland (2000), Tatihou, France (2001), Strobl, Austria (2002), and last year in San Miniato, Italy. This year the seminar will held at the MSRI. The meeting is sponsored by MSRI.

Scope . Predicting the performance of algorithms is a likely outgrowth of ongoing research in analytic combinatorics and the analysis of random discrete structures. This workshop will bring together leading researchers in this field to focus on such problems. Probabilistic considerations on inputs and the random combinatorial structures underlying algorithmic analysis have provided an active area of modern research. One assumes some reasonable probability distribution on input instances to an algorithm as a way of understanding the inner workings of the algorithm and its "typical behavior." Experience in the field shows that it is often unwieldy to work with exact models, where on the other hand one can say something meaningful and precise on the typical "asymptotic" behavior of the algorithm, when either the underlying combinatorial structure becomes very large or when the algorithm is challenged by massive data sets. In these cases one sometimes gets simplified but exact expressions dealing with first (or higher) order expansions of averages, moments or distributions, as some parameters of the algorithmic problem grow to be very large. The focus of this workshop is the average case analysis of algorithms, and its relation to the wider areas of analytic combinatorics, exact and limiting distributions, formal techniques, probability theory, combinatorics and computer science. We identify the following areas as being of particular interest:

Atmosphere. Following the tradition of the first four seminars, this seminar intends to bring together leading researchers in the Analysis of Algorithms and provide them with a relaxed atmosphere for interaction and discussion. Therefore, the talks will generally be brief and somewhat sparse. Long lunch breaks and one free afternoon will be purposely planned. A problem session will also be planned.

Keynote Speakers

Persi Diaconis, Philippe Flajolet, Don Knuth, and Dick Karp.

Program Committee

Philipe Jacquet, Helmut Prodinger, Gadiel Seroussi, Robert Sedgewick, Wojtek Szpankowski (chair), Marcelo Weinberger, and Brigitte Vallee