RESEARCH / LOCAL ONLY
Penalty & Shot Sweeps
Classical sampling illustrations on the same six-bit toy instance used by the Resilience Lab. No external API, no storage, no advantage claim.
C — Penalty Strength Sweep
Classical Boltzmann sampler, seed 42, 512 samples. Penalty P varies. Quantum/QAOA unavailable for these sweeps.
| P | Feasible % | Best coverage | Optimum hit? |
|---|
D — Shot Budget vs Quality
Classical Boltzmann sampler, P=36, seed 42. Sample count varies; these are not quantum shots.
| Shots | Feasible % | Best coverage | Optimum hit? |
|---|
Results use classical energy-weighted random sampling, not a quantum circuit or noiseless quantum simulator. They illustrate sampling behavior only; no quantum metrics or advantage are claimed. The Resilience Lab is a separate executable circuit experiment.
01 — Classical Distribution Mixing
Classical probability mixture with a uniform distribution. This is not a per-qubit noise channel or hardware measurement.
| Uniform-mixture weight | Feasible % | Best coverage |
|---|
02 — Uniform vs Energy-Biased Sampling
Compares two classical sampling distributions, not QAOA initialization or optimizer evaluation counts.
| Start | Best coverage | Optimum hit? |
|---|
03 — Energy Selection vs Post-Selection
Filtering samples after drawing them is not a constraint-preserving quantum mixer.
| Mode | Best coverage | Note |
|---|
04 — Circuit Angle Sensitivity (planned)
05 — Classical Sampling Variance (20 seeds)
Uses a fixed classical energy-weighted distribution and deterministic pseudo-random draws. These are not quantum-circuit runs.
| Min | Median | Max | Hit reference / all seeds | No feasible sample |
|---|
Read the method before the numbers
What the classical sweep shows · Sweep reading guide
Five planned research questions · How to prepare those experiments