After installing (see INSTALL.md):
python scripts/qec_demo.pyThis runs the full deterministic adaptive pipeline on 16 fixed input patterns:
metrics -> attractor -> strategy -> evaluation -> adaptation -> memory
Each step processes a deterministic signal pattern (constant, ramp, oscillation, step change, etc.), classifies the regime, selects a strategy, evaluates improvement, adapts bias, and records to memory — including transition learning and multi-step lookahead.
The output shows one block per input pattern, including:
- regime — classified state (stable, unstable, oscillatory, transitional, mixed)
- attractor — basin ID and basin score
- strategy — selected strategy and its score
- adaptation — global bias and trajectory score
- transition bias — learned bias from prior regime transitions
- multi-step factor — two-step lookahead influence
- evaluation — improvement score and outcome classification
At the end, a summary shows regime distribution and memory statistics.
Running the demo twice produces identical output. The system uses no randomness — all scoring, selection, and adaptation are fully deterministic.
For a more detailed report including per-metric breakdowns and strategy topology analysis:
python -c "from qec.experiments.metrics_probe import run_experiments, print_experiment_report; print_experiment_report(run_experiments())"This exercises the same pipeline with additional topology and calibration summary output.