bug: get_dft_array with symmetry objects

[oskooi]

There is a bug in the get_dft_array routine when used with symmetry objects. This is demonstrated in the following example of a ring resonator which is adapted from the frequency-domain solver tutorial.

import meep as mp
import numpy as np
import matplotlib.pyplot as plt

n = 3.4
w = 1
r = 1
pad = 4
dpml = 2

sxy = 2*(r+w+pad+dpml)
vol = mp.Volume(mp.Vector3(), size=mp.Vector3(sxy-2*dpml,sxy-2*dpml))

c1 = mp.Cylinder(radius=r+w, material=mp.Medium(index=n))
c2 = mp.Cylinder(radius=r)

fcen = 0.118
df = 0.08
src = mp.Source(mp.GaussianSource(fcen,fwidth=df), mp.Ez, mp.Vector3(r+0.1))

sim = mp.Simulation(cell_size=mp.Vector3(sxy,sxy),
                    geometry=[c1,c2],
                    sources=[src],
                    resolution=10,
                    symmetries=[mp.Mirror(mp.Y)],
                    boundary_layers=[mp.PML(dpml)])

dfts = sim.add_dft_fields([mp.Ez], fcen, fcen, 1, where=vol)
sim.run(until_after_sources=100)
ez_dat = np.absolute(sim.get_dft_array(dfts, mp.Ez, 0))

plt.figure(dpi=100)
plt.imshow(ez_dat.transpose(), interpolation='spline36', cmap='Reds')
plt.axis('off')
plt.show()

This simulation script produces the following image of the ring-resonator mode:

When the mirror-symmetry object is removed, the resulting image is correct: