The dashboard

The dashboard is a web page that follows a design run while it happens. It shows each number of the chip (qubit frequency, anharmonicity, resonator frequency, coupling), where the number comes from (the design, the lumped model or a solver) and whether it meets its goal. You do not need it to design a chip, but it is the quickest way to see what a run did and what is still off target.

It is a React app (sources in reviewer/) built into src/quantcore_circuits/review/static/ and served by a small Python backend (src/quantcore_circuits/review/server.py). You run it; you do not need to build it.

Start it

  1. Open a terminal in the repository and activate the environment: conda activate qcd.

  2. Start the server:

    qcc dashboard              REM follows the run with the latest activity
    qcc dashboard t04          REM pins the dashboard to the run build/t04
    scripts\dashboard.cmd      REM the same, without activating the env
    
  3. Open http://127.0.0.1:8765/ (it opens by itself unless you pass --no-open-browser; --port changes the port).

qcc review DESIGN serves one design file instead of a tracked run. Without a run, the dashboard reads build/_live.json to find the run that changed last.

Feed it from your code

The dashboard shows a run only when something records it.

  • In a notebook or script, call track.start("t04", dashboard=True) or set the environment variable QCC_TRACK=1.

  • After that, build, DRC, fill, simulate, loop and optimise steps report by themselves (autotrack.py), and a watcher follows the FEM job folders.

  • The state is one file, build/<run>/state.json (its contract is reviewer/contract/state.ts). The screenshots of the layout go to build/<run>/shots/.

The backend has three routes: GET /api/state (the run), GET /api/events (server-sent events, so the page updates live) and GET /files/<rel> (images and job files).

Tabs

Choose a tab in the top bar, or add ?tab= to the address: parameters (the default), layout, circuit, hamiltonian, plots, simulation, history, log. ?el=Q1 selects an element and ?sample shows a fixture with no backend. The screenshots are of kept runs.

Parameters

The first tab. A chip overview with the qubit table, the couplers, and one card per value. Each card shows the value, its source tag (design.json, lom, FEM) and the goal with its tolerance. A value outside the tolerance gets a FAIL mark.

../_images/t04_parameters.png

Run t04 (two Xmons). This is the last revision the tutorial tracked; it misses its targets (0 of 4 met).

../_images/t01_parameters.png

Run t01 (12-qubit chip): the same view for a larger chip, with the element filter.

../_images/five_qubit_parameters.png

Run five_qubit (the IBM-style five-qubit chip).

Hamiltonian

The model (a transmon in the charge basis plus a lambda/4 resonator), its parameters, the dressed spectrum and the estimator consensus table. Estimators that disagree by more than the tolerance are flagged. pyepr is the chosen value where an eigenmode result exists.

../_images/t00_hamiltonian.png

Run t00.

Simulation

The FEM jobs of the run with their state, then the estimate against the FEM value per branch with the difference in percent, and the Maxwell C-matrix of the newest Q3D job.

../_images/t04_simulation.png

Run t04: nine kept FEM jobs (all DONE, results kept, finished 13-16 Nov).

History

The targets with their current value, the residual norm per feedback iteration (log scale, tolerance dashed) and the iteration table with the knobs and residuals.

../_images/t04_history.png

Run t04. The residual does not settle below tolerance in this tracked run.

Plots

Estimator values per parameter, and the deviation from the chosen value.

../_images/t01_plots.png

Run t01: all 12 qubits.

Circuit

The schematic of the lumped model (island, junction, coupling capacitor, claw, lambda/4, hanger kappa, 50 ohm port) next to the branch table with value, source (fem, estimate) and node list.

../_images/t00_circuit.png

Run t00.

../_images/t02_circuit.png

Run t02 (five qubits and two buses): the same view of a bigger chip.

Layout

KLayout captures (headless renders) from each revision. The dashboard does not draw the layout itself.

../_images/t04_layout.png

Run t04.

Log

The steps of the run in order, with their time and any error.

Limits

  • Fab is not a dashboard step. Per-instance values are shown here only, not on hover in KLayout.

  • The render and optimiser marks with a real AEDT run were unit-tested, and the chain check used them live once (2026-11-13).

  • q1_r1_first_run and q1_r1_loop start AEDT whenever it is installed, whatever RUN_FEM says.

  • Verified on every tutorial and the q1_r1 notebook (2026-11-13); t04 was served again on 2026-11-18. Tests: test_dashboard_tracking.py (27) and test_codesign_track.py (7).

See the flow for where the dashboard sits in the loop and the status page for what is verified.