Defining a 3D Plane Source

I’m trying to define a plane source such that it outputs a plane wave, but I’m getting some odd-looking outputs. The following code seems to produce some pretty heavy fringes of some sort. Any thoughts? Ideally it would produce a smooth plane wave. Thanks!

import matplotlib.pyplot as plt

import fdtd
import fdtd.backend as bd
import numpy as np
import torch
import torch.autograd
import torch.nn
fdtd.set_backend("torch.cuda")


WAVELENGTH = 1550e-9
SPEED_LIGHT: float = 299_792_458.0  # [m/s] speed of light


grid = fdtd.Grid(
    (2.5e-5, 2.5e-5, 1.5e-5),
    # (161,161,97),
    grid_spacing=0.1 * WAVELENGTH,
    permittivity=1.0,
    permeability=1.0,
    # courant_number=0.3
)

grid[0, :, :] = fdtd.PeriodicBoundary(name="xbounds")
# grid[0:10, :, :] = fdtd.PML(name="pml_xlow")
# grid[-10:, :, :] = fdtd.PML(name="pml_xhigh")
grid[:, :, 0:10] = fdtd.PML(name="pml_zlow")
grid[:, :, -10:] = fdtd.PML(name="pml_zhigh")
grid[:, 0, :] = fdtd.PeriodicBoundary(name="ybounds")

grid[2:160,2:160,30:31] = fdtd.PlaneSource(
    period=WAVELENGTH / SPEED_LIGHT, name="linesource"
)
grid.run(300)
grid.visualize(z=80,show=True)

[CharlesDove]

Figured it out. A plane source at z=N outputs the wrong polarization, which doesn’t propagate. Changing it to something transverse solves the problem. Also there should probably be GaussianPlaneSource and UniformPlaneSource methods to reduce ambiguity. For line sources as well. I’ll see if I can knock together a PR. 😃

[flaport]

@CharlesDove , If you have the time, would you be willing to create a PR with your fix?