# Notebooks Questions about the tutorial notebooks and the design notebooks. The pages are the [tutorials](../tutorials/index.md), the [design notebooks](../designs/overview.md) and the [notebook template](../designs/notebook-template.md). ## Which notebook? ### What is the difference between the tutorials and the design notebooks? The five **tutorials** (`examples/tutorials/`) teach the loop in order and are saved with all their outputs, so you can read them without running anything. The **design notebooks** (`examples/designs//`) rebuild published or example chips (IBM, Google, Qiskit Metal, KQC, ...) one segment at a time, each with a picture, then follow the plugin's steps: Check, LOM, Simulate, Results, Fab. Start with tutorial 0. See [tutorials](../tutorials/index.md) and [design notebooks](../designs/overview.md). ### What are the `.v1` notebooks? The older form of a design notebook, kept byte for byte next to its rewrite (`.v1.ipynb` beside `.ipynb`). Both write the same `.json`, so run one or the other, not both in a row expecting different files. The tests skip `.v1` notebooks. See the [notebook template](../designs/notebook-template.md). ### Where do the saved outputs and kept FEM results live? A notebook keeps the outputs of the cells it was run with (that is wanted). Solved results sit next to it in a `fem/` folder (for the tutorials `examples/tutorials/fem/`) and are not solved again. ## Running ### Do the notebooks start AEDT? Not as shipped. `RUN_FEM = False` exports the jobs and loads kept results. Set `RUN_FEM = True` (or a set such as `{"q3d_cmatrix"}`) to solve in AEDT 2024 R1 or later. `AEDT_MODE` picks `batch`, `gui` or `attach`; `RESOLVE = False` does not solve again when kept results of the same revision exist; `FEM_FAST = True` uses the fast settings. See [simulation](simulation.md) for the modes. ### The notebook does not open KLayout, or I want no window. By default a notebook is live: each `add`, `route` and `set` is sent to KLayout over KLive, and the window opens at the first build. Set `QCC_LIVE=0` before starting Jupyter for headless runs; builds then write `build/.oas` and headless renders. Without KLayout installed the notebooks fall back to headless. See [KLive](../plugin/klive.md). ### What do I run first, and in which environment? `conda activate qcd`, `set OPENBLAS_NUM_THREADS=2` on a small PC, then `jupyter lab examples/tutorials/00_quickstart_klive_and_pdk.ipynb` from the repository root. Run from the top. See the [tutorials](../tutorials/index.md). ### How long does a notebook take? The tutorials with `RUN_FEM = False` take 1 to 5 minutes. With solving on, each FEM job adds minutes: Q3D about 3, eigenmode 10 to 25, driven 20 to 40 (times on a 31 GB workstation, fast settings). The design notebooks list their timings on their own pages. ### What does "Same design as before" mean? The last Design cell saves `.json` and prints whether it is byte-identical to the file it replaces. If it says "differs", the notebook changed the design; compare with the `.v1` and fix the notebook, not the JSON. ### Can I run only part of a chip? Yes. Before the FEM cell, render and solve a subset (the feedline plus the pairs you choose) as a convergence and sanity check; afterwards render the whole chip at fast settings. The LOM itself always uses the whole chip. See [simulation](simulation.md) and [the LOM](lom.md). ### Can I watch a notebook in the dashboard? Yes. Put `track.start("name", dashboard=True)` at the top or set `QCC_TRACK=1`, then run `qcc dashboard`. See [the dashboard](../overview/dashboard.md). ## Design notebooks and their pages ### Does every design notebook build? Each v2 design notebook was run up to the whole-chip render and checked with "Same design as before"; the LOM, FEM and feedback cells were not run. The "Builds today" row on a design's page says what happened. See [design notebooks](../designs/overview.md). ### Where are the LOM estimates of a design? On the design's page, under "LOM estimates", from a separate pass (the notebook's LOM cells, or the `.v1` output for big chips). They are lumped estimates, not FEM; see [the LOM](lom.md).