Meshing and Adaptive Mesh Refinement: Information on meshing is given in “Initial Meshing”. For information on the AMR process, see “Using Adaptive Mesh Refinement (AMR)”. Adaptive mesh refinement is performed on the basis of the gradient of the underlying magnetic potential. This typically produces reasonable refined meshes over a sequence of several AMR steps. In some cases it may be useful to seed the mesh with more elements in certain areas (by applying mesh constraints), or globally using the element size parameters on the initial mesh simulation properties panel. This can lead to higher quality solutions early in the AMR process, possibly reducing the number of steps required for solution convergence.
Resource Usage: As with other solution types, the computer resource requirements are a strong function of the problem. As compared to other solvers, MS2p and MS3p generally take less time on a given mesh than would a frequency-domain solver, but more time than is needed for an electrostatic run because of the non-linear materials commonly used in magnetostatics. On a given problem, solution time is affected by the choice of linear solver, and also the tolerances that were chosen. The default linear solver is the conjugate gradient iterative method, and this is generally the most efficient solver for magnetostatics. For an iterative linear solver the solution time will generally increase as the Convergence Tolerance is reduced. The Nonlinear Convergence Tolerance, which is used to determine when the material permeability values are sufficiently converged, also strongly affects the solution times, with smaller values generally increasing the solve time.