Experimental Analysis of Algorithms.
Abstract
This thesis examines the application of experimental, statistical, and data analysis tools to problems in algorithm analysis. Note that algorithms, not programs, are studied: results in algorithm analysis generally refer to abstract cost functions, are independent of particular machines or implementation strategies, and express functional relationships between input parameters and measures of algorithmic performance. The study of algorithms presents special problems and opportunities for experimental research. The following research goals are set: (1) To demonstrate that simulation can provide a useful, general tool for developing new understanding of algorithms; (2) To identify common problems and to assess the applicability of this approach; (3) To develop principles for successful experimental research in this domain; and (4) To promote more general use of this approach by giving a handbook of useful tools and techniques.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1987
- Accession Number
- ADA188528
Entities
People
- Catherine C. Mcgeoch
Organizations
- Carnegie Mellon University