Preparing the interactive workspace…Etkileşimli çalışma alanı hazırlanıyor…
00
Quantum Machine Learning
N/M Quantum AI
Train a variational quantum circuit directly in your browser. Declare trainable parameters, define a cost from a Pauli expectation, and watch a parameter-shift gradient optimizer converge - no setup, no hardware.
paramexpect <Z>parameter-shiftin-browser training
Preparing active experiment summary…
01
01 / Variational systems
A training run you can inspect.
Choose an N/M model, follow every optimization step, and carry the same source into the Playground. The browser simulator keeps the experiment local and reproducible.
Model
N/M modelLoad to editor
module rotation_fit;
fn main() {
let q: QReg<1> = qreg[1];
param w: Angle = 0.3;
Ry(q[0], w);
let cost = expect Z(q[0]);
metric z = expect Z(q[0]);
return cost;
}0.40
50
Model files and comparisonsExport, import, compare optimizers, or run the optional kernel tool.
Bloch sphere - state trajectory
Train a single-qubit model to animate its Bloch-sphere trajectory.
Cost vs. step
Pick a model and press Train to watch the cost converge.
Preparing benchmark launcher…
Classical and quantum model comparison
Compare four fixed model families on the same held-out data: a majority baseline, logistic regression, RBF kernel ridge and a variational quantum classifier. This experiment uses the selected dataset, not the circuit in the editor.
02
02 / Applied experiments
From decision boundaries to molecular energy.
Use the same measured-cost workflow across classification, VQE, bond scans, and trainability analysis. Each panel states its model, output, and practical limit.
Variational quantum classifier
A labeled toy dataset of two blobs. Each point is angle-encoded into one qubit; the model predicts a class from sign(<Z>). Training optimizes the weights so the learned decision boundary separates the classes. The shaded region is the model's prediction across feature space.
class +1 class -1
A single-qubit <Z> produces a linear boundary; multi-qubit ansatze (roadmap) unlock nonlinear decision boundaries.Train the classifier to see the learned decision boundary.
VQE - molecular ground state
The variational quantum eigensolver: minimize <H> of a two-qubit, H2-flavored Hamiltonian over a hardware-efficient ansatz. The optimizer drives the energy down to the exact ground-state energy - the canonical near-term quantum algorithm.
Run VQE to watch <H> converge to the ground-state energy.
Hamiltonian
I-0.4804
Z0+0.3435
Z1-0.4347
Z0 Z1+0.5716
Y0 Y1+0.0910
X0 X1+0.0910
Dissociation curve (bond-length scan)
Run VQE across a range of bond lengths to trace the potential-energy surface of a model H2 molecule. The orange points are VQE energies; the purple dashed line is the exact ground state. The well's bottom marks the equilibrium bond length.
Run the scan to trace the molecule's potential-energy curve.
Barren plateau explorer
Some circuit families and objectives develop very small gradients as they scale. This explorer samples random hardware-efficient ansatze with 1–5 qubits and measures Var(d<Z>/dtheta). This small experiment does not establish an exponential scaling law.
Run the analysis to compare sampled gradient variance across qubit counts.