Benchmarks & evaluation

Q-Lock is being evaluated on both simulators and real hardware to measure stability, distribution preservation, and sensitivity to compilation changes.

Benchmark goals

The goal is not to claim “magic error correction,” but to rigorously quantify when identity-locked perturbations improve stability and how they affect distributions under realistic noise.

Distribution fidelity

Compare locked vs. unlocked circuits on noiseless simulators to confirm that ideal distributions remain effectively unchanged.

Stability under noise

Measure histogram variance across repeated runs with depolarizing and relaxation noise, with and without Q-Lock.

Compiler sensitivity

Vary layout and optimization settings to see whether Q-Lock circuits exhibit reduced sensitivity to compilation drift.

Planned hardware examples

Hardware benchmarks will be documented using blurred or anonymized screenshots and summary tables to avoid exposing private backend details while still conveying behavior.

GHZ-style chains

4–12 qubit GHZ circuits used as a baseline for locking behavior: measuring how well the |00...0⟩ and |11...1⟩ peaks are preserved under noise.

Random entangling ladders

Higher-depth random entangling networks designed to probe how identity-locked perturbations interact with more chaotic circuits.

As benchmarks are finalized, this page will link to concrete results in the /benchmarks/ directory and include before/after plots for representative experiments.