Metaheuristika analüüs ümberjaotamise puhul
Date
2018
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Ümberjaotamine on protsess, mille käigus paigutatakse ümber ringkonna piirid. Seda võib formuleerida kombinatoorse optimeerimise probleemina, et saavutada soovitud poliitiline eesmärk. Lahendusruum on selles olukorras hiigelsuur ja puuduvad teadaolevad meetodid järjepidevaks probleemi lahendamiseks. Metaheuristika on üks meetodeid, mis on välja pakutud probleemi lahendamiseks.Käesolevas töös luuakse teste, et analüüsida metaheuristikaid ümberjoonistamise eesmärgil. Kaks implementeeritud simuleeritud lõõmutamise versiooni, mis on tuntud metaheuristika, on järgmised: järjestikune ja paralleelne. Eelnevalt väljatoodud versioone analüüsitakse töös loodud testide põhjal. On näidatud, et simuleeritud lõõmutamist võib kasutada piirkonna ringkondade piiride ümberjoonistamiseks vähese arvuga ringkondade puhul, näiteks Iowa osariik, millel on 4 kongressiringkonda. Aga loodud simuleeritud lõõmutamise versioonid ei ole efektiivsed piirkondades, kus on suur arv ringkondi nagu New Yorki osariik, kus on 27 kongressiringkonda.
Redistricting is the process of redrawing district boundaries. It can beformulated as a combinatorial optimization problem in order to achieve a particular political objective. However, the solution space is staggering in this case, and there are no known methods to tackle the problem in a consistent manner. Metaheuristics are one of the methods that have been proposed in order to address the problem.In this thesis, test suites are created for analysis of metaheuristics for redistricting purposes. Two versions of simulated annealing, a well-known metaheuristic, are created: sequential and parallel. They are analyzed using the created test suites.It is shown that simulated annealing can probably be used for redistricting of aregion with a small number of districts such as the state of Iowa, which has 4congressional districts. However, the created versions of simulated annealing are ineffective for regions with a larger number of districts such as the state of New York, which has 27 congressional districts.
Redistricting is the process of redrawing district boundaries. It can beformulated as a combinatorial optimization problem in order to achieve a particular political objective. However, the solution space is staggering in this case, and there are no known methods to tackle the problem in a consistent manner. Metaheuristics are one of the methods that have been proposed in order to address the problem.In this thesis, test suites are created for analysis of metaheuristics for redistricting purposes. Two versions of simulated annealing, a well-known metaheuristic, are created: sequential and parallel. They are analyzed using the created test suites.It is shown that simulated annealing can probably be used for redistricting of aregion with a small number of districts such as the state of Iowa, which has 4congressional districts. However, the created versions of simulated annealing are ineffective for regions with a larger number of districts such as the state of New York, which has 27 congressional districts.