Note
Go to the end to download the full example code.
Choose a mesh for detector collection#
Compare collected power, sampled detector area, and sampled illumination area over four explicit meshes. The study preserves the physical sample and instrument geometry. Its recommendation is a numerical resolution check for these observables, not validation of P1 against radiative transfer.
import os
import numpy as np
import MomentRT
from MomentRT import (
P1,
CartesianMesh,
Material,
RobinBoundary,
Simulation,
StructuredMedium,
SurfaceDetector,
SurfaceSource,
ureg,
)
show_figures = os.environ.get("MOMENTRT_SPHINX_GALLERY") != "1"
Fix the sample and optical configuration#
These coefficients describe an illustrative homogeneous phantom at 800 nm. The disk source and the separated reflection aperture lie on the top face.
material = Material(
phase_function=MomentRT.phase_function.HenyeyGreenstein(),
absorption=0.02 / ureg.mm,
reduced_scattering=1.0 / ureg.mm,
refractive_index=1.4,
)
sample = StructuredMedium(
origin=(-4, -3, 0) * ureg.mm,
size=(8, 6, 4) * ureg.mm,
)
sample.add_background(material=material)
source = SurfaceSource(
face="z_min",
position=np.array([-1.5, 0, 0]) * ureg.mm,
radius=0.8 * ureg.mm,
wavelength=800 * ureg.nm,
optical_power=1 * ureg.mW,
)
detector = SurfaceDetector(
face="z_min",
position=np.array([1.5, 0, 0]) * ureg.mm,
radius=0.75 * ureg.mm,
collection_axis=np.array([0, 0, -1]),
acceptance_angle=45 * ureg.degree,
)
mesh = CartesianMesh.from_sample(
sample=sample,
shape=(17, 13, 13),
)
boundary = RobinBoundary(effective_reflectance=0.4)
solver = P1(
boundaries={
face: boundary
for face in ("x_min", "x_max", "y_min", "y_max", "z_min", "z_max")
},
)
simulation = Simulation(
medium=sample,
source=source,
mesh=mesh,
solver=solver,
detectors={"reflection": detector},
)
Compare explicit mesh levels#
Power must change by at most 5 percent between successive levels. Both sampled footprint areas must be within 5 percent of their geometric areas. Two consecutive passing refinements are required at the fine end of the study; a recommended level must also agree with the finest supplied power. Each detector legend entry controls both its power and area curves. Node-count ticks identify the exact tested mesh levels. A missing recommendation means more refinement or different tolerances need investigation, rather than automatic acceptance of the last mesh.
study = simulation.mesh_convergence(
mesh_shapes=[
(9, 7, 7),
(13, 10, 10),
(17, 13, 13),
(25, 19, 19),
],
power_rtol=0.05,
area_rtol=0.05,
required_consecutive=2,
)
convergence_figure = study.plot(
power_unit="mW",
area_unit="mm^2",
show=show_figures,
)
for record in study.records:
measurement = record["detectors"]["reflection"]
print(f"Mesh shape: {record['mesh_shape']}")
print(f"Reflection power: {measurement['collected_power'].to('mW'):~P}")
print(f"Detector area error: {measurement['area_relative_error']:.2%}")
print(f"Source area error: {record['source_area_relative_error']:.2%}")
print(f"Recommended shape: {study.recommended_shape}")
Mesh shape: (9, 7, 7)
Reflection power: 0.006115150694112388 mW
Detector area error: 13.18%
Source area error: 0.53%
Mesh shape: (13, 10, 10)
Reflection power: 0.0043455866554655215 mW
Detector area error: 7.78%
Source area error: 0.53%
Mesh shape: (17, 13, 13)
Reflection power: 0.004653302067557733 mW
Detector area error: 3.75%
Source area error: 0.53%
Mesh shape: (25, 19, 19)
Reflection power: 0.00432318018288807 mW
Detector area error: 1.49%
Source area error: 1.31%
Recommended shape: None
Inspect the finest field with the same instrument overlays#
The study retains each result. A source footprint with no quadrature points is recorded with result=None; this example’s meshes resolve the source. The original simulation.mesh remains unchanged.
finest_result = study.results[-1]
fluence_figure = finest_result.fluence.plot(
view="volume",
scale="log",
overlays=[source, detector],
show=show_figures,
)
Inspect the modeled geometry#
This view shows the configured geometry and instrument footprints.
gallery_system_figure = simulation.plot_system(show=False)
Total running time of the script: (0 minutes 1.060 seconds)