Skip to Main content Skip to Navigation
New interface
Journal articles

Automatic Average-case Analysis of Algorithms

Philippe Flajolet 1 B. Salvy P. Zimmermann 1 
1 ALGO - Algorithms
Inria Paris-Rocquencourt
Abstract : Many probabilistic properties of elementary discrete combinatorial structures of interest for the average-case analysis of algorithms prove to be decidable. This paper presents a general framework in which such decision procedures can be developed. It is based on a combination of generating function techniques for counting, and complex analysis techniques for asymptotic estimations. We expose here the theory of exact analysis in terms of generating functions for four different domains: the iterative/recursive and unlabelled/labelled data type domains. We then present some major components of the associated asymptotic theory and exhibit a class of naturally arising functions that can be automatically analyzed. A fair fragment of this theory is also incorporated into a system called Lambda-Upsilon-Omega. In this way, using computer algebra, one can produce automatically non-trivial average-case analyses of algorithms operating over a variety of "decomposable" combinatorial structures. At a fundamental level, this paper is part of a global attempt at understanding why so many elementary combinatorial problems tend to have elementary asymptotic solutions. In several cases, it proves possible to relate entire classes of elementary combinatorial problems whose structure is well defined with classes of elementary "special" functions and classes of asymptotic forms relative to counting, probabilities, or average-case complexity.
Document type :
Journal articles
Complete list of metadata
Contributor : Paul Zimmermann Connect in order to contact the contributor
Submitted on : Thursday, December 12, 2013 - 12:42:26 PM
Last modification on : Friday, February 4, 2022 - 3:10:23 AM

Links full text




Philippe Flajolet, B. Salvy, P. Zimmermann. Automatic Average-case Analysis of Algorithms. Theoretical Computer Science, 1991, 79 (1), pp.37--109. ⟨10.1016/0304-3975(91)90145-R⟩. ⟨hal-00917732⟩



Record views