I. Introduction
Computing the interaction fields between magnetic elements is a key step in solving micromagnetic problems. To avoid a large time-consuming convolution, the computation may be handled in the frequency domain [1]. The magnetization in an entire layer is taken into the frequency domain, multiplied by the 2D transfer-function, and then back-transformed to give the fields in another layer. Although the magnetization may evolve with time or as a solution converges, the transfer function from magnetization to field is fixed throughout the process. The transfer function may be computed in spatial domain and then transformed to frequency domain or it may be computed directly in the frequency domain. The latter approach, discussed here, will be shown to conveniently allow soft layers (with linear B-H characteristics) to be included above and below the interacting layers.