Rectangular Y-junction power split vs Tidy3D#
validated · 2026-08-28
This case compares a local FDTDX result with pinned reference data generated by Tidy3D. The originating Tidy3D example is YJunction.ipynb, pinned at commit c37c785d52e9.
Integrated-photonics geometries covered by the benchmark suite.#
What is exercised#
3D strip Y at dl≈40nm (~0.2M voxels). Compare mode-power T1,T2,Tsum to Tidy3D. Rectangular junction approx. Asymmetry or low Tsum = geometry/mode bug to fix.
This case runs at its declared full benchmark resolution (resolution factor 1.0).
Recorded result#
Metric |
Observed / error |
Reference / limit |
Error |
Effective budget |
Result |
|---|---|---|---|---|---|
|
0.202924 |
0.270472 |
0.0675476 |
0.147618 |
PASS |
|
0.261742 |
0.270474 |
0.00873133 |
0.147618 |
PASS |
|
0.464667 |
0.540946 |
0.0762789 |
0.208189 |
PASS |
The table is rendered from progress.json; it is not a hand-written success claim. For metrics that report an error directly, the “observed” column repeats that error and the reference column is the acceptance threshold.
Reproduce#
uv run fdtdx-bench run --case parity_y_junction
Implementation and provenance: benchmarks/cases/parity_y_junction/case.yaml, benchmarks/cases/parity_y_junction/run.py, benchmarks/goldens/parity_y_junction.npz.