Memory reduction in MLFMA through target rotation [scattering computation]

Using a parallel implementation of the multilevel fast multipole algorithm (MLFMA) called LSSP, large-scale scattering problems can be solved. However, as the problem sizes in terms of number of unknowns are increased, the memory requirements increase. This paper illustrates how the memory required by some practical problems can be reduced. The memory reduction can be realized by constructing the bounding box around the target such that the overall dimension of the box enclosing the target is minimized. In the case of a sphere, the rotational symmetry gives rise to a single box size; however, with a missile-shaped target or aircraft, the box size containing the target is reduced when the target extends from corner to corner instead of side to side. This can be realized by a simple coordinate transformation or rotation of the target geometry.