Error reduced ADI-FDTD methods

The Courant-Friedrich-Levy (CFL) stability condition makes the explicit finite-difference time-domain (FDTD) methods computationally expensive in applications where small cell sizes are needed to resolve high variations of fields. To circumvent the problem, unconditionally stable alternate-direction implicit (ADI)-FDTD method has been recently proposed. However, its uses are limited by large errors associated with larger time steps. In this letter, based on the Crank-Nicolson FDTD methods, we propose two new ADI-FDTD methods with reduced errors at large time steps. Numerical examples are presented to demonstrate the methods.