dc.contributor.author | Malalanirainy Rakotoson, T | |
dc.date.accessioned | 2021-07-28T08:57:43Z | |
dc.date.issued | 2021-08-02 | |
dc.description.abstract | This thesis aims to develop a unified runtime analysis of: EA 1 with no mutation and with a standard crossover, (1 + 1) EA, and EA with both a mutation and a standard crossover. To this end, we determined for each algorithm a class of problems they efficiently solve. Polynomially quasi-concave problems on the Hamming (resp. Manhattan) space, that are already known to be easy for EA with no mutation and with a non-standard crossover, were shown to be easy for EA with no mutation and with a standard crossover. A class of problems that is determined by its balls of radius ρ, is defined for each the following algorithms: (1 + 1) EA and EA with both a mutation and a standard crossover. Each of these classes is shown to only contain easy problems for an instantiation of a gener- alization of the algorithm they correspond to. Unlike the class of quasi-concave fitness landscapes, these classes are not affected by the choice of representation. We conclude that if the definition of a class of problems is built upon particular metrics, then the runtime result is affected by the choice of representation. | en_GB |
dc.identifier.uri | http://hdl.handle.net/10871/126589 | |
dc.publisher | University of Exeter | en_GB |
dc.rights.embargoreason | I plan to publish chapters of the thesis as research papers. | en_GB |
dc.title | A runtime analysis method unifying evolutionary algorithms on discrete search spaces | en_GB |
dc.type | Thesis or dissertation | en_GB |
dc.date.available | 2021-07-28T08:57:43Z | |
dc.contributor.advisor | Moraglio, A | en_GB |
dc.publisher.department | Computer Science | en_GB |
dc.rights.uri | http://www.rioxx.net/licenses/all-rights-reserved | en_GB |
dc.type.degreetitle | PhD in Computer Science | en_GB |
dc.type.qualificationlevel | Doctoral | en_GB |
dc.type.qualificationname | Doctoral Thesis | en_GB |
rioxxterms.version | NA | en_GB |
rioxxterms.licenseref.startdate | 2021-08-02 | |
rioxxterms.type | Thesis | en_GB |
refterms.dateFOA | 2021-07-28T08:57:58Z | |