A comparison of differential evolution and a genetic algorithm applied to the longest path problem

Genetic algorithms and differential evolution are two well-established types of generic algorithms that can be applied to a great numberof optimization problems. Both are subgroups of evolutionary algorithms that are inspired by nature, with many practical implementations in for instance research an...

Full description

Bibliographic Details
Main Authors: Hamilton, Marcus, Nyman, Jacob
Format: Others
Language:English
Published: KTH, Skolan för elektroteknik och datavetenskap (EECS) 2018
Subjects:
Online Access:http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-233544
id ndltd-UPSALLA1-oai-DiVA.org-kth-233544
record_format oai_dc
spelling ndltd-UPSALLA1-oai-DiVA.org-kth-2335442018-08-29T05:59:06ZA comparison of differential evolution and a genetic algorithm applied to the longest path problemengEn jämförelse av differentialevolution och en genetiskalgoritm tillämpat på längstastigproblemetHamilton, MarcusNyman, JacobKTH, Skolan för elektroteknik och datavetenskap (EECS)KTH, Skolan för elektroteknik och datavetenskap (EECS)2018Engineering and TechnologyTeknik och teknologierGenetic algorithms and differential evolution are two well-established types of generic algorithms that can be applied to a great numberof optimization problems. Both are subgroups of evolutionary algorithms that are inspired by nature, with many practical implementations in for instance research and the industry. In this paper the algorithms are applied to the NP-hard longest path problem with the purpose of comparing their perfomances. The basics of the algorithms are provided along with defining their most important components, chromosome, gene, crossover, mutationand selection. All of which are further described in detail how they were implemented for this paper. Additionally a specialized type ofthe differential evolution algorithm is brought up, namely discrete differential evolution, as this version was implemented for this paper. The results show that the differential evolution algorithm performssignificantly better than the genetic algorithm and the possible underlying causes to this are discussed, one major cause being that the differential evolution algorithm is more adapted to the specified problem, longest path problem. The conclusion is that discrete differential evolution perfoms considerably better than a generic genetic algorithmon this particular problem, but no further general assumptions can bemade regarding their perfomance for all or any other types of problems. Genetiska algoritmer och differential evolution är två väletablerade typer av generiska algoritmer som går att applicera på mängder av olika optimeringsproblem. Båda tillhör gruppen evolutionära algoritmer som är inspirerade av naturen och har många praktiska tillämpningar inom exempelvis forskning och industrin. I denna studie appliceras algoritmerna på det NP-fullständiga längsta stig problemet i syftet att jämföra deras prestanda. En grundläggande beskrivning av algoritmerna ges och dess viktigaste komponenter som kromosom, gen, crossover, mutation och selektion definieras. Sedan beskrivs deras implementationen i närmare detalj för denna studie. Även en specialiserad variant av differentialevolution, discrete differential evolution, tas upp då denna variant implementeras i denna undersökning .Resultaten visar att algoritmen för differential evolution presterarväsentligt bättre än motsvarande genetisk algoritm. En större orsak till detta är att den implementerade varianten av differential evolutionär mer problembaserad än den mer generiska genetiska algoritmen. Slutsatsen som dras är att en variant av discrete differential evolution presterar betydligt bättre än en generisk genetisk algoritm på detta problem, men inget generellt antagande kan göras om algoritmernasprestanda för alla eller andra typer av problem. Student thesisinfo:eu-repo/semantics/bachelorThesistexthttp://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-233544TRITA-EECS-EX ; 2018:214application/pdfinfo:eu-repo/semantics/openAccess
collection NDLTD
language English
format Others
sources NDLTD
topic Engineering and Technology
Teknik och teknologier
spellingShingle Engineering and Technology
Teknik och teknologier
Hamilton, Marcus
Nyman, Jacob
A comparison of differential evolution and a genetic algorithm applied to the longest path problem
description Genetic algorithms and differential evolution are two well-established types of generic algorithms that can be applied to a great numberof optimization problems. Both are subgroups of evolutionary algorithms that are inspired by nature, with many practical implementations in for instance research and the industry. In this paper the algorithms are applied to the NP-hard longest path problem with the purpose of comparing their perfomances. The basics of the algorithms are provided along with defining their most important components, chromosome, gene, crossover, mutationand selection. All of which are further described in detail how they were implemented for this paper. Additionally a specialized type ofthe differential evolution algorithm is brought up, namely discrete differential evolution, as this version was implemented for this paper. The results show that the differential evolution algorithm performssignificantly better than the genetic algorithm and the possible underlying causes to this are discussed, one major cause being that the differential evolution algorithm is more adapted to the specified problem, longest path problem. The conclusion is that discrete differential evolution perfoms considerably better than a generic genetic algorithmon this particular problem, but no further general assumptions can bemade regarding their perfomance for all or any other types of problems. === Genetiska algoritmer och differential evolution är två väletablerade typer av generiska algoritmer som går att applicera på mängder av olika optimeringsproblem. Båda tillhör gruppen evolutionära algoritmer som är inspirerade av naturen och har många praktiska tillämpningar inom exempelvis forskning och industrin. I denna studie appliceras algoritmerna på det NP-fullständiga längsta stig problemet i syftet att jämföra deras prestanda. En grundläggande beskrivning av algoritmerna ges och dess viktigaste komponenter som kromosom, gen, crossover, mutation och selektion definieras. Sedan beskrivs deras implementationen i närmare detalj för denna studie. Även en specialiserad variant av differentialevolution, discrete differential evolution, tas upp då denna variant implementeras i denna undersökning .Resultaten visar att algoritmen för differential evolution presterarväsentligt bättre än motsvarande genetisk algoritm. En större orsak till detta är att den implementerade varianten av differential evolutionär mer problembaserad än den mer generiska genetiska algoritmen. Slutsatsen som dras är att en variant av discrete differential evolution presterar betydligt bättre än en generisk genetisk algoritm på detta problem, men inget generellt antagande kan göras om algoritmernasprestanda för alla eller andra typer av problem.
author Hamilton, Marcus
Nyman, Jacob
author_facet Hamilton, Marcus
Nyman, Jacob
author_sort Hamilton, Marcus
title A comparison of differential evolution and a genetic algorithm applied to the longest path problem
title_short A comparison of differential evolution and a genetic algorithm applied to the longest path problem
title_full A comparison of differential evolution and a genetic algorithm applied to the longest path problem
title_fullStr A comparison of differential evolution and a genetic algorithm applied to the longest path problem
title_full_unstemmed A comparison of differential evolution and a genetic algorithm applied to the longest path problem
title_sort comparison of differential evolution and a genetic algorithm applied to the longest path problem
publisher KTH, Skolan för elektroteknik och datavetenskap (EECS)
publishDate 2018
url http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-233544
work_keys_str_mv AT hamiltonmarcus acomparisonofdifferentialevolutionandageneticalgorithmappliedtothelongestpathproblem
AT nymanjacob acomparisonofdifferentialevolutionandageneticalgorithmappliedtothelongestpathproblem
AT hamiltonmarcus enjamforelseavdifferentialevolutionochengenetiskalgoritmtillampatpalangstastigproblemet
AT nymanjacob enjamforelseavdifferentialevolutionochengenetiskalgoritmtillampatpalangstastigproblemet
AT hamiltonmarcus comparisonofdifferentialevolutionandageneticalgorithmappliedtothelongestpathproblem
AT nymanjacob comparisonofdifferentialevolutionandageneticalgorithmappliedtothelongestpathproblem
_version_ 1718727286954393600