Quantum State Studio is a visual showcase for luma.gl rendering and browser GPU compute. It runs a classical state-vector simulation: it is not a quantum computer, a production quantum SDK, or a claim of quantum speedup.
Amplitudes are complex f32 pairs in computational-basis order, with qubit zero as the
least-significant index bit. Every gate writes a new slice in one GPU-resident history buffer. The
editable circuit can therefore be scrubbed gate by gate without replaying earlier gates or mapping
amplitudes back into JavaScript. One generic kernel applies a 2x2 complex matrix to a target qubit,
optionally conditioned on one control bit; common and controlled gates are small descriptors over
that primitive.
The simulation pipeline orders gate applications, derives probability and phase, and reduces probability normalization for every snapshot. A second graph derives the selected qubit's reduced Bloch vector and connected Z-correlation matrix. The luma.gl renderer binds those buffers directly to one cinematic linked-view model.
The probability landscape uses height for probability and a cyclic hue for complex phase. The history view exposes constructive and destructive interference across gate boundaries. The Bloch view shrinks when the selected qubit is entangled with the rest of the pure state, while the correlation matrix makes Bell- and GHZ-like structure legible even when individual probabilities do not explain it.
Presets include Bell and GHZ entanglement, an interference echo, a complete two-qubit Grover iteration, and a QFT decomposition with controlled phase rotations and bit reversal. Common gates can be appended directly in the showcase.
State-vector cost is exponential. The example targets fluid 8–16 qubit exploration and guards at
16 qubits; available circuit depth is also constrained by the full history retained for scrubbing.
All math is f32, so small normalization drift accumulates with depth. The visualization measures
that normalization but does not silently alter the simulated state.
Natural extensions include small density matrices, noise and decoherence channels, tensor-network backends, entanglement entropy, Wigner or other complex phase-space views, and import adapters for OpenQASM, QIR, or external circuit tools.