Validation
Cross-Validation
Two independent simulators confirm the validity of AGE-TGI-2S3/IDS + QEC results
Independent Verification
The scientific validity of any simulation result depends on its reproducibility. To ensure the robustness of our AGE-TGI-2S3/IDS + QEC results, we have developed two independent simulators with different code architectures and verified that they produce numerically consistent results.
The comparison script runs both simulators with identical inputs and compares outputs metric by metric. What it validates and what it does not: since both share the same logical-error model, the agreement (differences < 10⁻¹²) verifies the algebra implementation (2×2 matrices), not the physical model; model validation belongs to the published scripts and to the TM and QC simulation banks.
Simulator 1
Direct 2×2 matrix multiplication
Simulator 2
Euler angle parameterization
Verification Methodology
Cross-validation follows a rigorous protocol. Both simulators receive identical input parameters (operators A and B, time step h, code distance d, detection efficiency ηdet) and independently calculate:
- The approximate unitary U_approx via the 7-segment sequence
- Gate fidelity F = |Tr(U_target†U_approx)|²/4
- Physical error per segment 1 - Feff
- Logical error εlogical for the chosen QEC code
- The repetition-code statistics for each d, ηdet pair
Simulator Differences
Simulator 1
Direct 2×2 matrix multiplication
Simulator 2
Euler angle parameterization
Comparison Results
| Metric | Simulator 1 | Simulator 2 | |Δ| | Status |
|---|---|---|---|---|
| F (h=0.1) | 0.999999999847 | 0.999999999847 | < 10-12 | ✔ OK |
| 1 - Feff | 4.762×10-2 | 4.762×10-2 | < 10-12 | ✔ OK |
| εlog (d=5, η=0.95) | 7.45×10-3 | 7.45×10-3 | < 10-12 | ✔ OK |
| εlog (d=5, η=0.99) | 1.71×10-3 | 1.71×10-3 | < 10-12 | ✔ OK |
| εlog (d=13, η=0.95) | 7.24×10-5 | 7.24×10-5 | < 10-12 | ✔ OK |
| εlog (d=13, η=0.99) | 2.51×10-6 | 2.51×10-6 | < 10-12 | ✔ OK |
Verification Script
The repetition-code model (repetition_code_model.py) and the regenerated CSV are published in the open-data section of Results, so any researcher can replicate the figures without relying on an external repository. The script uses only the Python standard library.
To run the verification, download the script and CSV and run:
wget https://cuantic-control.tech/wp-content/themes/age-tgi-qec/assets/data/repetition_code_model.py
python3 repetition_code_model.py
The script accepts parameters for the platform, detection efficiency, and code distance, allowing verification of any configuration reported on this website.
repetition_code_model.py
- Language
- Python 3
- Dependencies
- Standard library only
- Output
- CSV
- License
- MIT