3D simulation 2D visualization PML bug

[hammy4815]

There appears to be in issue when running 3D simulations and calling plot2D to visualize. It shows PML everywhere, when there shouldn't be.

import meep.adjoint as mpa
from matplotlib import pyplot as plt
import numpy as np

diamond = mp.Medium(index=2.4)  # Lossless diamond
air = mp.Medium(index=1.0)  # Lossless air

# Geometry
radius = 0.88
geometry = [
    mp.Cylinder(radius=radius, height=0.22, material=diamond),  
]

# Source
fcen = 1.67969  # Center frequency
df = 0.05  # pulse frequency width
sources = [
    mp.Source(src=mp.GaussianSource(fcen, fwidth=df), component=mp.Ex, center=mp.Vector3(0.88 - 0.15, 0, 0.00))
]  # Dipole source

# Simulation
sim = mp.Simulation(
    cell_size=mp.Vector3(7*radius, 7*radius, 10),  # Size of the simulation cell
    geometry=geometry,  # Geometry
    boundary_layers=[mp.PML(1.0,direction=mp.X),mp.PML(1.0,direction=mp.Y),mp.PML(1.0,direction=mp.Z)],  # Boundary conditions
    resolution=10,  # Resolution
    sources=sources,  # Sources
)

# Plot the geometry
sim.init_sim()
if mp.am_really_master():
    plt.figure()
    sim.plot2D(output_plane=mp.Volume(center=mp.Vector3(0, 0, 0), size=mp.Vector3(7*radius, 7*radius, 0)))
    plt.savefig("geom")`

cc @smartalecH

[smartalecH]

Similar issues with 2D simulations that simply change the geometry_center.

[stevengj]

I guess it is getting confused by the PML in the z direction, and doesn't notice that those PML layers don't intersect the slice being plotted?

[oskooi]

This bug is indeed related to the use of PML since replacing mp.PML(...) with mp.Absorber(...) produces the correct output (i.e., no hatched lines).