|
1 | | -********************************** |
2 | | -Fabrication-aware inverse design |
3 | | -********************************** |
4 | | - |
5 | | -This seminar (October 9, 2025) tracks a dual-layer grating coupler workflow end to end: build the baseline, seed with Bayesian optimization, refine with adjoint gradients, evaluate fabrication sensitivities, and close the loop with measurement-informed calibration. Every step uses Tidy3D so you can replicate the exact jobs or adapt the utilities to your own stack. |
6 | | - |
7 | | -Seminar recording: `YouTube link <https://www.youtube.com/watch?v=OpVBJmomzoo>`_ |
8 | | - |
9 | | -Notebook lineup |
10 | | -================ |
11 | | - |
12 | | -* ``00_setup_guide.ipynb`` - builds the nominal SiN stack, launches a reference simulation, and visualizes the initial geometry for later reuse. |
13 | | -* ``01_bayes.ipynb`` - performs a five-parameter Bayesian search to find a high-performing uniform grating without gradients. |
14 | | -* ``02_adjoint.ipynb`` - expands to per-tooth parameters and runs Adam with adjoint sensitivities to apodize the grating. |
15 | | -* ``03_sensitivity.ipynb`` - sweeps :math:`\pm 20` nm etch bias, runs Monte Carlo samples, and records adjoint-derived sensitivity units. |
16 | | -* ``04_adjoint_robust.ipynb`` - penalizes variance across nominal, over, and under etch corners to produce a fabrication-aware design. |
17 | | -* ``05_robust_comparison.ipynb`` - replays the Monte Carlo experiment for nominal and robust designs to quantify yield gains. |
18 | | -* ``06_measurement_calibration.ipynb`` - applies gradient-based calibration so simulated spectra match measurements, illustrating fabrication corner recovery. |
19 | | - |
20 | | -Supporting assets |
21 | | -================= |
22 | | - |
23 | | -* ``setup.py`` - shared simulation builders, fabrication constraints, and helper utilities. |
24 | | -* ``optim.py`` - lightweight Adam plus parameter clipping helpers. |
25 | | -* ``results/`` - JSON checkpoints (Bayes best guess, adjoint refinements, robust design) used downstream. |
26 | | - |
27 | | -Walkthrough notebooks |
28 | | -===================== |
29 | | - |
30 | | -.. toctree:: |
31 | | - :maxdepth: 1 |
32 | | - |
33 | | - ../notebooks/2025-10-09-invdes-seminar/00_setup_guide |
34 | | - ../notebooks/2025-10-09-invdes-seminar/01_bayes |
35 | | - ../notebooks/2025-10-09-invdes-seminar/02_adjoint |
36 | | - ../notebooks/2025-10-09-invdes-seminar/03_sensitivity |
37 | | - ../notebooks/2025-10-09-invdes-seminar/04_adjoint_robust |
38 | | - ../notebooks/2025-10-09-invdes-seminar/05_robust_comparison |
39 | | - ../notebooks/2025-10-09-invdes-seminar/06_measurement_calibration |
| 1 | +.. include:: /notebooks/docs/fabrication_aware_invdes.rst |
0 commit comments