Statistical coarse graining as an approach to multiscale problems in magnetism (invited)

Multiscale phenomena which include several processes occurring simultaneously at different length scales, and exchanging energy with each other, are widespread in magnetism (e.g., nucleation of magnetization reversal at an atomic-scale defect or grain boundary, a domain wall breaking through an atomic-scale defect, etc.). We demonstrate that in such multiscale processes, a role of short-wavelength excitations dynamically generated near the defects is crucial. As a result, standard micromagnetic theory is not always sufficient for description of these phenomena. We present a novel coarse-graining approach to length scales coupling in dynamical magnetic modeling, which allows accurate treatment of microscopic defects in nanomagnets. The results of simulations performed on simplified model systems, where an atomic-scale variation of anisotropy and exchange represent defects in real systems, show that the coarse-graining approach achieves practically the precision of exact atomic simulations, while describing the system with a much smaller number of degrees of freedom. (C) 2003 American Institute of Physics.