Eliminating boundary-induced errors in the pseudo-current method (Q1208871)

The author considers the numerical solution of electromagnetic problems by the use of the equation \(\partial E/\partial t = \text{curl }B - J + d\text{ grad}(\text{div }E-p)\). He discusses a method wherein \(\text{div }E-p\) vanishes on the boundaries, and on the first cell in from the boundaries. This generally results in significantly lower values for \(| \text{div }E - p|\) within the region under consideration.