Modelos matemáticos e algoritmos para problemas combinatórios

Detalhes bibliográficos
Ano de defesa: 2011
Autor(a) principal: Ravelo, Santiago Valdes lattes
Orientador(a): Meneses, Claudio Nogueira de lattes
Banca de defesa: Meneses, Cláudio Nogueira de, Longo, Humberto José, Morabito, Reinaldo
Tipo de documento: Dissertação
Tipo de acesso: Acesso aberto
Idioma: por
Instituição de defesa: Universidade Federal de Goiás
Programa de Pós-Graduação: Programa de Pós-graduação em Ciência da Computação (INF)
Departamento: Instituto de Informática - INF (RG)
País: Brasil
Palavras-chave em Português:
Palavras-chave em Inglês:
Área do conhecimento CNPq:
Link de acesso: http://repositorio.bc.ufg.br/tede/handle/tede/5354
Resumo: This work considers three relevant NP-hard problems. The firstone is the one-dimensional cutting stock problem in which the non-used material in the cutting patterns may be used in the future. For this problem we analyze the existing mathematical models, propose new models, design a heuristic and two metaheuristic approaches, being their performances improved by using parallel programming, and solve instances, practical and randomly generated, from the literature. The computational experiments were quite good for all tested instances. The second problem we consider is the stable roommates problem (a variant of the stable matching problem). For this we give two mathematical programming models, sequential and parallel implementations of a Tabu Search, and a Branch-andBound. Also, we report computational experiments to instances of the problem. The last problem we consider is the compartmentalized knapsack problem (a generalization of the knapsack problem) for which we analyze a quadratic integer model and give a linear integer model. We design a greedy heuristic and a GRASP algorithm, that uses path-relinking, and solve randomly generated instances. All parallel implementations use Graphics Processing Units (GPUs).