GUI

The Streamlit dashboard exposes QQA as an interactive tool with four pages.

pip install "qqa[gui]"
qqa gui
# or equivalently:
streamlit run app/streamlit_app.py

Pages

Home

Pick a problem family, size, seed, and problem-specific parameters. A preview is rendered as a graph (for combinatorial problems), a coupling heatmap (for spin problems), or a pattern matrix (for the binary perceptron).

Solve

Tune QQA hyper-parameters (sol_size, learning_rate, temp, min_bg, max_bg, curve_rate, div_param, num_epochs) and hit Run QQA to launch the anneal. A StreamlitCallback streams the progress bar, current metrics, and a live loss/best plot.

Visualize

Four tabs — dynamics, best trajectory, schedule, solution heatmap — all backed by the :mod:qqa.visualization helpers in their Plotly flavour.

Compare

Run a small grid over min_bg × max_bg × div_param on the current problem and inspect the outcome with a parallel-coordinates plot plus an overlaid best-objective trajectory.

Programmatic access

StreamlitCallback lives inside the Solve page (it depends on Streamlit's runtime and session state). If you want to reuse it in your own Streamlit app, copy the class out of app/pages/1_Solve.py or import it from there:

import importlib.util
from pathlib import Path

spec = importlib.util.spec_from_file_location(
    "solve_page", Path("app/pages/1_Solve.py")
)
solve_page = importlib.util.module_from_spec(spec)
spec.loader.exec_module(solve_page)
cb = solve_page.StreamlitCallback(...)

For non-Streamlit use cases, just subclass :class:qqa.callbacks.Callback directly — the base class is framework-agnostic.