The domain configurations on the air-bearing surface (ABS) of inductive thin-film recording heads were studied. It was found that, instead of being a single domain structure, the ABS of a thin-film head usually has multidomains. The direction of the domain walls is neither parallel nor perpendicular to the gap plane. The magnetization was found to be in the plane of the ABS along the track width, with the magnetizations on the two sides of the domain walls either "head-to-head" or '"tail-to-tail." The domain walls are slanted in order to spread the magnetic charges along the wall over a larger region, thereby reducing magnetostatic energy in this configuration. The responses of the domain walls are not all in phase, and they are generally out of phase with the rotational process along the gap edge. The magnetization configuration on the ABS and in the throat and the sloped region were investigated in one. head and correlated with the domain walls on the ABS.

The recursive least square lattice (LSL) algorithm based on the newly developed scaled tangent rotations (STAR) is derived. Similar to other recursive least square lattice algorithms for adaptive filtering, this algorithm requires only O(N) operations. This algorithm also preserves the desired properties of the STAR recursive least square (STAR-RLS) algorithm. Specifically, it can be pipelined at fine-grain level. To this end, a pipelined version of the STAR-LSL (referred to as PSTAR-LSL) is also developed. Computer simulations show that the performance of the STAR-LSL algorithm is as good as the QRD-LSL algorithm. The finite precision error properties of the STAR-LSL algorithm are also analyzed. The mean square error expressions show that the numerical error propagates from stage to stage in the lattice, and the numerical error of different quantities in the algorithm varies differently with . This suggests that different word lengths need to be assigned to different variables in the algorithm for best performance. Finally, finite word length simulations are carried out to compare the performances of different topologies.

No abstract