Search

Convolutional perfectly matched layer (CPML)’s constitutive parameters assignment methods are performed and compared in this paper. One method is to assign the PML's parameters at the centers of the cells and the other is to assign parameters at the electric and magnetic field sampling points. The former method requires less memory than the latter and the constitutive parameters in calculation are achieved by averaging those of the adjacent cells. While using the latter method, more memory will be required since the electric and magnetic sampling points are not at the same position, but the constitutive parameters are more accurately assigned. Numerical experiments were performed and the absorbing performances of the two methods were compared. The results show that assigning the constitutive parameter at the field sampling points has much better absorbing effect than the cell center assignment. [ABSTRACT FROM AUTHOR]