Research Article
Analysis of Electromagnetic Propagation from MHz to THz with a Memory-Optimised CPML-FDTD Algorithm
Figure 21
Error induced by single-precision variables with respect to double-precision ones.
| (a) Parabolic reflector numerical experiment. L2 norm error is of 7.403291 × 10−07 (2.45 GHz, snapshot at time steps (7.09 s)) |
| (b) CNWG numerical experiment. L2 norm error is of 2.630922 × 10−06 (100 THz, snapshot at time steps (1.740295218690000 × 10−04 s)) |
| (c) Indoor propagation numerical experiment. L2 norm error is of 3.0817 × 10−11 (100 THz, snapshot at time steps). It can be clearly observed that for such long time periods, single-precision version is numerically stable |