Get A guide to experimental algorithmics PDF

By Catherine C. McGeoch

ISBN-10: 0521173019

ISBN-13: 9780521173018

ISBN-10: 1107001730

ISBN-13: 9781107001732

"Computational experiments on algorithms can complement theoretical research by way of exhibiting what algorithms, implementations, and speed-up tools paintings most sensible for particular machines or difficulties. This ebook publications the reader throughout the nuts and bolts of the key experimental questions: What should still I degree? What inputs may still I attempt? How do I research the information? Answering those questions wishes rules from set of rules design Read more...

Show description

Read Online or Download A guide to experimental algorithmics PDF

Best programming languages books

Download PDF by Mary Beth Chrissis, Mike Konrad, Sandra Shrum: CMMI - Guidelines for Process Integration and Product

CMMI® for improvement (CMMI-DEV) describes top practices for the improvement and upkeep of goods and providers throughout their lifecycle. through integrating crucial our bodies of information, CMMI-DEV offers a unmarried, complete framework for enterprises to evaluate their improvement and upkeep methods and enhance functionality.

Read e-book online Executive Guide to Speech-Driven Computer Systems PDF

A brand new new release of speech-driven pc platforms provides to remodel the enterprise use of knowledge expertise. this isn't purely a question of discarding the keyboard, yet of rethinking company methods to exploit the elevated productiveness that speech-driven platforms can deliver.

Download e-book for iPad: A guide to experimental algorithmics by Catherine C. McGeoch

"Computational experiments on algorithms can complement theoretical research via exhibiting what algorithms, implementations, and speed-up tools paintings top for particular machines or difficulties. This booklet courses the reader during the nuts and bolts of the key experimental questions: What may still I degree?

Extra info for A guide to experimental algorithmics

Example text

Question 2 should be attacked by measuring both time and solution quality, using a variety of graph classes and some state-of-the art algorithms for comparison, and problem sizes that are typical in practice. An experimental design is a plan for an experiment that targets a specific question. The design specifies what properties to measure, what input classes to incorporate, what input sizes to use, and so forth. Like battle plans, experimental designs may be small and tactical, suitable for reconnaissance missions, or large and strategic, for full-scale invasions.

And incrementing by powers of 10 n = 10, 100, 1000, . .. 9 To study trends and functions, choose design points that exploit what you already know. Making Comparisons with Factorial Designs Another common goal of algorithm research is to compare performance across several algorithm and instance factors, to discover which implementation ideas work for which inputs. These types of questions arise in horse race experiments and assessment studies. For this type of problem a full factorial design, a cornerstone of DOE, is simplest and often the best choice.

Which input parameters appear to be relevant to performance? The workhorse study comprises experiments built upon precisely stated problems: Estimate, to within 10 percent, the mean comparison costs for data structures A and B, on instances drawn randomly from input class C; bound the leading term of the (unknown) cost function F (n). Designs for workhorse experiments require some prior understanding of algorithm mechanisms and of the test environment. This understanding may be gleaned from pilot experiments; furthermore, a great deal of useful intelligence – which ideas work and do not work, which input classes are hard and easy, and what to expect from certain algorithms – may be found by consulting the experimental literature.

Download PDF sample

A guide to experimental algorithmics by Catherine C. McGeoch


by Thomas
4.5

Rated 4.58 of 5 – based on 30 votes