Real-world device reproductions and inverse-design audit#

Ten forward examples from the official Tidy3D notebook collection are validated as executed, downloadable FDTDX notebooks. The high-Q nanobeam tutorial independently rebuilds literal cylinders and reaches 5.33× the unchanged cavity Q locally; a matched fixed-hole-count Tidy3D pair confirms a 1.741× shape-optimization gain. The other inverse-design pages distinguish local optimizer scores from independent Tidy3D forward validation of the same final geometry: the S-matrix crossing passes the user’s 10% trust check but misses the stricter 95% external gate, while WDM, bandpass, the digital splitter, and compact grating coupler remain explicit failures.

What “reproduced” means here

For a validated entry, the benchmark retains the physical device and observable, then compares an executed local result with a frozen reference derived from the notebook or a reviewed one-time Tidy3D run. Each forward link is the actual .ipynb: ordinary Python cells followed by retained geometry, material, field, spectrum, ringdown, or analysis outputs. Inverse cases return only after their initialization, parameterization, symmetries, fabrication constraints, normalization, and FOM all match the cited problem.

Browse by device family#

Integrated photonic circuits#

Each notebook follows the same practical sequence: parameter setting, geometry setting, geometry/material plotting, local simulation, field plotting, and quantitative analysis. Solid curves/markers are FDTDX values; dashed comparison curves/markers are the pinned Tidy3D reference where applicable.

Directional coupler — broadband through/cross parity

headline error 0.0385 · budget 0.2

Executed notebook · parameters · geometry · voxelization · field · analysis

Directional coupler: through and cross spectra
Waveguide crossing — O-band through and crosstalk spectra

headline error 0.0388 · budget 0.1

Executed notebook · parameters · geometry · voxelization · field · analysis

Waveguide crossing: transmission and crosstalk
Euler-like bend — broadband 90-degree transmission

headline error 0.0078 · budget 0.05

Executed notebook · parameters · geometry · voxelization · field · analysis

Euler-like waveguide bend: broadband bend loss
Edge coupler — Gaussian beam to inverse taper

headline error 0.155 · budget 0.2

Executed notebook · parameters · geometry · voxelization · field · analysis

Edge coupler: Gaussian beam to inverse taper
1x4 MMI — four-port broadband power distribution

headline error 0.0769 · budget 0.12

Executed notebook · parameters · geometry · voxelization · field · analysis

1×4 MMI: broadband four-port splitting
Compact polarization splitter-rotator — TE/TM routing

headline error 0.00308 · budget 0.08

Executed notebook · parameters · geometry · voxelization · field · analysis

Polarization splitter-rotator: TE/TM routing

Resonators and spectral devices#

Bragg grating — transmission and reflection stopband

headline error 0.0791 · budget 0.12

Executed notebook · parameters · geometry · voxelization · field · analysis

Bragg grating: transmission and reflection stopband
Nanobeam cavity — resonant wavelength and Q

headline error 0.0193 · budget 0.0229

Executed notebook · parameters · geometry · voxelization · field · analysis

Nanobeam cavity: resonant wavelength and quality factor

Free-space and periodic optics#

Dielectric metasurface absorber — periodic THz resonator

headline error 0.00985 · budget 0.12

Executed notebook · parameters · geometry · voxelization · field · analysis

Dielectric metasurface absorber: R, T, and A
Geometric-phase metalens — focal distance and spot size

headline error 0.112 · budget 0.448

Executed notebook · parameters · geometry · voxelization · field · analysis

Geometric-phase metalens: focal distance and spot size

Inverse-designed devices#

Executed Autograd24 digital-splitter audit →

Parameters → exact shallow-hole geometry → uniform start → 150 differentiable FDTD updates → fully binary checkpoint selection → convergence and passivity → fresh five-wavelength fields. The scientifically comparable run reaches 0.4325 (91.6% of Tidy3D), below the repository’s stricter 0.4485 gate.

Digital 1×2 splitter: shallow-hole inverse design
Executed compact grating-coupler audit →

Exact random cold start → 20 nm double-filtered controls → 80 nm erosion/dilation rule → 150 differentiable 3D FDTD updates → fully binary checkpoint → fresh xy/xz fields → independent Tidy3D spectrum. FDTDX reports 0.5773; Tidy3D gives 0.5339 for that same geometry versus 0.6637 published. Scientific result: FAIL.

Compact inverse-designed grating coupler

Five former passes or scalar claims remain visible as failures when independent validation, parameterization, or memory requirements do not support equivalence. The high-Q cavity is a separate validated local objective—not a claim of parity with a published Tidy3D inverse-design result. No reduced 16×16 surrogate is presented as equivalence. See the inverse-design fidelity audit for the exact comparison and memory evidence.

GPU-native inverse design of a high-Q nanobeam cavity

best FOM 187133.234 · target 70250.601 · 1 updates

Executed notebook · parameters · geometry · voxelization · field · analysis

Inverse design of a high-Q nanobeam cavity
Inverse-designed waveguide bend — modal-transmission parity

invalid adapter · scalar claim withdrawn

Audit record · initialization · symmetry · constraints · FOM

Inverse-design fidelity audit
Inverse-designed four-channel wavelength demultiplexer

invalid adapter · scalar claim withdrawn

Audit record · initialization · symmetry · constraints · FOM

Inverse-design fidelity audit
Inverse-designed broadband photonic bandpass filter

invalid adapter · scalar claim withdrawn

Audit record · initialization · symmetry · constraints · FOM

Inverse-design fidelity audit
Inverse-designed low-loss waveguide crossing

invalid adapter · scalar claim withdrawn

Audit record · initialization · symmetry · constraints · FOM

Inverse-design fidelity audit
Inverse-designed crossing from a four-port S-matrix target

best FOM 0.969 · target 0.931 · 75 updates

Executed notebook · parameters · geometry · voxelization · field · analysis

S-matrix crossing: four-port inverse design
Frozen remote-atom shifted-LDOS optimization →

Executed parameter setting → literal seven-row geometry → shifted-pole FDTD configuration → complete accepted/rejected history → paired Ey field → frozen 40 nm analysis. The best accepted silicon geometry reaches LDOS 18.166 and conservative Q 44,667; it is explicitly pending fine-grid validation.

Shape optimization of a remote-atom silicon cavity

Results at a glance#

Device

Mode

Headline result

Status

Directional coupler — broadband through/cross parity

forward

0.03851 / 0.2 budget

PASS

Waveguide crossing — O-band through and crosstalk spectra

forward

0.03879 / 0.1 budget

PASS

Euler-like bend — broadband 90-degree transmission

forward

0.007795 / 0.05 budget

PASS

Edge coupler — Gaussian beam to inverse taper

forward

0.155 / 0.2 budget

PASS

1x4 MMI — four-port broadband power distribution

forward

0.07692 / 0.12 budget

PASS

Bragg grating — transmission and reflection stopband

forward

0.07908 / 0.12 budget

PASS

Nanobeam cavity — resonant wavelength and Q

forward

0.0193 / 0.02294 budget

PASS

GPU-native inverse design of a high-Q nanobeam cavity

invdes

187133.2341 / 70250.6014 minimum

PASS

Dielectric metasurface absorber — periodic THz resonator

forward

0.009849 / 0.12 budget

PASS

Compact polarization splitter-rotator — TE/TM routing

forward

0.003077 / 0.08 budget

PASS

Geometric-phase metalens — focal distance and spot size

forward

0.1125 / 0.448 budget

PASS

Inverse-designed waveguide bend — modal-transmission parity

invdes

not scientifically comparable

AUDIT FAIL

Inverse-designed four-channel wavelength demultiplexer

invdes

not scientifically comparable

AUDIT FAIL

Inverse-designed broadband photonic bandpass filter

invdes

not scientifically comparable

AUDIT FAIL

Inverse-designed low-loss waveguide crossing

invdes

not scientifically comparable

AUDIT FAIL

Inverse-designed crossing from a four-port S-matrix target

invdes

0.9694 / 0.9314 minimum

PASS

Compact inverse-designed grating coupler

invdes

0.5339 / 0.6637 published

FAIL

Remote-atom shifted-LDOS shape optimization

invdes research

LDOS 18.166 / Q 44,667 at 40 nm

FROZEN EXAMPLE

Automatic setup policy#

The point of this ladder is broader than making fifteen hand-tuned scenes pass. The reusable policies that emerged are now part of the benchmark infrastructure:

  • choose grid pitch from the smallest physical feature and material wavelength, with anisotropic spacing for long adiabatic devices;

  • boolean-union touching material before rasterization so geometry does not acquire artificial seams;

  • normalize modal and free-space power with an independent incident/reference solve;

  • stop pulsed simulations from a relative field-energy decay criterion, retaining a hard time-window ceiling;

  • reproduce the source notebook’s filter, projection schedule, initialization, and erosion/dilation constraints exactly;

  • retain fixed incident calibration and reject non-passive results instead of rescaling powers onto a passive simplex.

Those choices are recorded in each case and are visible in the runners. A new device should normally need geometry, materials, ports, and an objective—not a private collection of normalization factors and termination constants.

Reproduce the ladder#

Run one case directly:

uv run fdtdx-bench run --case device_directional_coupler
uv run fdtdx-bench run --case invdes_power_splitter

The exact notebook list, SHA-256 pins, and cloud-spend ceiling live in benchmarks/device_ladder.yaml. Frozen-reference provenance lives in benchmarks/golden_provenance.yaml; no Tidy3D job is submitted in the benchmark loop.