Bloch disk (drag to rotate)
trajectory ŷ = +1 ŷ = −1
Decision boundary
Hover the heatmap →
Loss (orange) / Accuracy (teal) vs. training step

Quantum Data Re-uploading Classifier — 2D Bloch Disk

A single qubit, reset to |0⟩ and re-encoded with the same 2D input point across several layers, is enough to learn a nonlinear decision boundary — a result from the "data re-uploading" family of quantum machine-learning circuits. This 2D-native build runs the identical single-qubit statevector math and exact parameter-shift-rule gradient as the 3D version, but renders it entirely with flat canvas drawing: a draggable Bloch-disk projection, a raster classification heatmap, and a live-scrolling loss/accuracy chart, so you can watch a genuinely quantum-native optimizer sculpt a decision boundary — and converge — in real time.