2D Visualisation in 3D simulations

[Srinivas-Ganti]

Hi everyone,

I am wondering if there is a standard way of quickly visualizing a slice of the simulation using meep.plot2D(). I tried to do this but i am getting errors, I show a minimal example of a 3D geometry of a THz absorber on fused quartz. Is there a way to run plot2D without the simulation starting? Thanks for your suggestions.

import meep as mp
from meep.materials import fused_quartz
import math
import cmath
import matplotlib.pyplot as plt
import gdspy

def makeUnitCell(L,g,w,l,layer=0):
    a1 = (-L/2, -L/2)            #A 
    a2 = (L/2, -L/2)             #B  
    a3 = (L/2, L/2)              #C  
    a4 = (-L/2, L/2)             #D  
    a10 = (g/2+w, -l/2)          #I 
    a12 = (g/2, l/2)             #K
    a21 = (-g/2, l/2)            #T
    a23 = (-g/2-w, -l/2)         #V
   
    unit_cell = gdspy.Cell("unit", exclude_from_current=True)
   
    rect1 = gdspy.Rectangle(a1, (a2[0],a2[1]+w), layer = layer)
    rect2 = gdspy.Rectangle(a2, (a3[0]-w,a3[1]), layer = layer)
    rect3 = gdspy.Rectangle((a3[0],a3[1]-w),a4, layer= layer)
    rect4 = gdspy.Rectangle(a1,(a4[0]+w,a4[1]), layer= layer)
    rect5 = gdspy.Rectangle((-L/2,-w/2),(-g/2,w/2), layer= layer)
    rect6 = gdspy.Rectangle((g/2,w/2),(L/2,-w/2), layer= layer)
    rect7 = gdspy.Rectangle(a21,a23, layer= layer)
    rect8 = gdspy.Rectangle(a10,a12, layer= layer)
   
    unit_cell.add(gdspy.boolean(rect1,[rect2,rect3,rect4,rect5,rect6,rect7,rect8],operation = "or",layer = 1))
    return unit_cell

um_scale = 1.0


eV_um_scale = um_scale/1.23984193

Pt_plasma_frq = 9.59*eV_um_scale
Pt_f0 = 0.333
Pt_frq0 = 1e-10
Pt_gam0 = 0.080*eV_um_scale
Pt_sig0 = Pt_f0*Pt_plasma_frq**2/Pt_frq0**2
Pt_f1 = 0.191
Pt_frq1 = 0.780*eV_um_scale      # 1.590 um
Pt_gam1 = 0.517*eV_um_scale
Pt_sig1 = Pt_f1*Pt_plasma_frq**2/Pt_frq1**2
Pt_f2 = 0.659
Pt_frq2 = 1.314*eV_um_scale      # 0.944 um
Pt_gam2 = 1.838*eV_um_scale
Pt_sig2 = Pt_f2*Pt_plasma_frq**2/Pt_frq2**2

Pt_susc = [ mp.DrudeSusceptibility(frequency=Pt_frq0, gamma=Pt_gam0, sigma=Pt_sig0),
            mp.LorentzianSusceptibility(frequency=Pt_frq1, gamma=Pt_gam1, sigma=Pt_sig1),
            mp.LorentzianSusceptibility(frequency=Pt_frq2, gamma=Pt_gam2, sigma=Pt_sig2) ]

Pt = mp.Medium(epsilon=1.0, E_susceptibilities=Pt_susc)    

resolution = 40     # pixels/um
L = 36 # outer metamaterial dimensions
g = 2 # gap size
w = 4 # track width
l = 10 # len of capacitive section
p = 50 # periodicity in x and y directions
s_l = 50  # substrate die length
s_w = 50  # substrate die width 
tmet = 0.2          # metal layer thickness
tsub = 2.0          # substrate thickness
tabs = 5.0          # PML thickness
tair = 8.0          # air thickness
dpml = 1 
tstack = tmet + tsub + tair+dpml+tabs
cell_size = mp.Vector3(p,p,tstack)
unit_cell = makeUnitCell(L,g,w,l)
polyset = unit_cell.get_polygons(by_spec=True)
vertices = polyset[(1,0)][0]
v = [mp.Vector3(vertices[i][0],vertices[i][1]) for i in range(len(vertices))]
metam = mp.Prism(v, height = tmet, material = Pt)
subs = mp.Block(material = fused_quartz, size = mp.Vector3(p,p,tsub),center = mp.Vector3(0,0,-tsub/2))
geom = [metam, subs]

sim = mp.Simulation(resolution = 5, cell_size = cell_size, geometry = geom)
sim.run(mp.at_beginning(mp.output_epsilon),until = 1)
sim.plot2D(output_plane=mp.Volume(mp.Vector3(p,0,tstack)))
plt.show()

[smartalecH]

First off your simulation is enormous... (2000x2000x648) pixels...

Second, as the error indicates, "The specified user volume is completely outside of the simulation domain." You think you are setting the size of your volume, but in reality, you are setting the center.

Third, you can indeed initialize the 2D slice without initializing the entire 3D stack using the : eps_parameters={'resolution':<resolution_goes_here>} parameter within plot2D().

Putting all of that together:

sim = mp.Simulation(resolution = 5, cell_size = cell_size, geometry = geom)
sim.plot2D(output_plane=mp.Volume(size=mp.Vector3(p,0,tstack)),
eps_parameters={'resolution':50},frequency=1
)

gets you this:

Note that I specified a random frequency (1) so that the metals would evaluate a point along the Lorentzian (if you have sources, meep should be smart enough to use the center frequency from that).