抄録
In a particle method, it is necessary to search for neighboring particles within the influence radius of a particle. This search is called a neighborhood search, and, when using a uniform influence radius, a data structure based on a uniform grid is often used. However, because there are some grids where registered particles are all outside the influence radius depending on the particle position in the grid, the neighborhood search with data structures based on the uniform grid is not efficient. In this study, the efficiency of the neighborhood search with the uniform grid was improved. In this method, an axis-aligned bounding box (AABB), which includes particles registered in each grid, is acquired, and it is determined whether the AABB is within the influence radius of the particle. When the AABB exists outside the influence radius, all particles registered in this grid are outside the influence radius, and thus, it is possible to perform the pruning of the neighborhood search. This method, implemented on a central processing unit, matched the search result of the uniform grid, and the sum of the calculation times for data structure construction and the neighborhood search was reduced by 20.7%.
