Algoritmos para construção de quadtrees com base em partição binária de espaço

Abstract

Space subdivision techniques, such as quadtrees and octrees, are widely used as auxiliary structures for various applications. In particular, they are used in the generation of finite element meshes. For generation of large refined meshes, it is necessary to use parallel programming. Thus, this work presents a technique for parallel construction of quadtrees based on Binary Space Partitioning (BSP). Given a discrete model in vertices and edges, it is possible to extractinformation to perform a load estimate that serves as a guide for partitioning the model between processes using BSP. This partitioning aims to balance the work of generation and storage of the quadtree performed by each process. This technique was applied in three models for 2 to 16 processes, presenting good results for uniformly distributed models