Evolving Fixed-parameter Tractable Algorithms

Keywords

Thumbnail Image

Issue Date

2008-06-30

Language

en

Document type

Journal Title

Journal ISSN

Volume Title

Publisher

Title

ISSN

Volume

Issue

Startpage

Endpage

DOI

Abstract

One effective means of computing NP-hard problems is provided by fixed-parameter tractable (fpt-) algorithms. An fpt-algorithm is an algorithm whose running time is polynomial in the input size and superpolynomial only as a function of an input parameter. Provided that the parameter is small enough, an fpt-algorithm runs fast even for large inputs. In this paper, we report on an investigation of the evolvability of fpt-algorithms via Genetic Programming (GP). The problem used in this investigation is the NP-hard 2D-Euclidean Traveling Salesman Problem (TSP), which is known to be fpt if the number of points not on the convex hull is taken as the parameter. The algorithm evolved in our GP study turns out to have clear characteristics of an fpt-algorithm. The results suggest GP can be utilized for generating fpt-algorithms for NP-hard problems in general, as well as for discovering input parameters that could be used to develop fpt-algorithms.

Description

Citation

Faculty

Faculteit der Sociale Wetenschappen