Some micromagnetic corrections to numerical coercivity calculations

Previous numerical coercivity calculations used models that were essentially at the domain level in resolution. This paper attempts to assess some of the errors in such calculations without a full scale micromagnetic calculation, since such a calculation is at present prohibitively expensive. Errors as large as an order of magnitude have been calculated by analyzing the nature of the breakdown that occurs when the equilibrium is exceeded. The interaction between an inclusion, much smaller than the thickness of a domain wall, and the wall can also yield large errors, especially when a graded mesh is used to study the details of the interaction.