A fast algorithm for simulation of flocking behavior

This paper proposes an algorithm to enhance the performance of the spatial partitioning method for flocking behavior. Even when a moving entity (boid) in a flock changes its direction and location, its k-nearest neighbors (kNN), which influence its decision for the next direction, seldom change. Using this fact, this paper improves the performance by finding kNN of boids efficiently. A method to check that the new kNN is not changed from the previous kNN is proposed, and the correctness of the method is proven with two theorems. In order to minimize the cost of computing the new kNN, the method checks the fact that the new kNN did not change from the previous kNN. If the new kNN is not changed, the method copies the previous kNN to the new kNN instead of computing the new kNN. The proposed algorithm was implemented and its performance was compared with the original spatial partitioning method. The results of the comparison showed that the proposed algorithm outperformed the original method by about 57.7% with respect to the number of frames per second.