Abstract: We present a distributed parallel SPH programming algorithm using adaptive mesh refinement background grids,in which size of a grid is decided based on maximal smoothed length of local particles. All neighboring particles of a given particle can be found in the grid the particle belongs to and in grids of same size adjoining to this grid. Searching bound is confined and as a result computational efficiency is improved. The method is validated in non-uniform smoothed length SPH simulation.

Key words: smoothed particle hydrodynamics, adaptive mesh refinement, parallel, order of magnitude in neighbor search