https://hgpu.org/?p=6465
Comparison of CPML Implementations for the GPU-Accelerated FDTD Solver