tmalc said:
Winn said:
You should use the CPML from Gedney, since this kind of Perfectly matched layer is independent of the material properties at the interface.
I think PML and UPML can handle inhomogeneous medium as well, by normalizing epsilon by the relative permittivity [ 2nd-FDTD book Toflove, page304, eq.7.52]
PML is just the topic keyword for nearly perfect absorption. UPML is a specific technic to get a nearly perfect absorption. Achieving nearly perfect absorption for inhomoheneous media is not difficult to handle for every implemented absorption method (Split-Field PML, Uniaxial PML, Convolutional PML), but as far as you go into dispersive materials you should avoid Split-Field PML and Uniaxial PML, because you have to derive and implement for every dispersive media a special suited algorithm for that media. Choosing the Convolutional PML (3rd Editition Taflove) from Gedney, you got one algorithm for every dispersive media.