Open-source diagnostic tools for NISQ quantum hardware stability —
calculating the "real" vs "marketed" qubit count from public calibration data
git clone https://github.com/UniversalModel/Quantum-triadic-autopsy
cd Quantum-triadic-autopsy
pip install -r requirements.txt
python run_all.py # runs all tools, outputs go to results/No IBM account needed. Demo mode uses a public 127-qubit Brisbane snapshot (Feb 2026).
IBM Brisbane, 127 qubits — average SI_Q = 0.11
That means 99% of the chip is computationally useless by an objective metric.
IBM counts qubits. This repo counts useful computation.
A single hardware health score computed from three independent properties:
SI_Q = cbrt(F_Q x P_Q x A_Q) / (1 + delta)^2
F_Q = min(1, (T1 + T2) / 500 us) -- Form (coherence lifetime)
P_Q = min(1, n_neighbors / 6) -- Position (connectivity)
A_Q = max(0, 1 - gate_err*100 - ro_err*10) -- Action (gate fidelity)
delta = (max(F,P,A) - min(F,P,A)) / max(F,P,A) -- Triadic imbalance
| SI_Q | Zone | Meaning |
|---|---|---|
| >= 0.618 | Utility Zone | Error suppression exceeds encoding overhead |
| 0.40-0.617 | Warning Zone | Marginal -- overhead approaching break-even |
| < 0.40 | Dead Zone | Active error amplification -- computationally useless |
No current real hardware platform on > 40 qubits reaches even 0.50.
python IBM_Q_SI_Q_Autopsy.py --demo # no account needed
python IBM_Q_SI_Q_Autopsy.py --backend ibm_brisbane # live dataOutputs: chip heatmap (PNG), per-qubit F/P/A bar chart (PNG), full report (TXT).
Brisbane 127q result: avg SI_Q = 0.11 -- 99.2% qubits in Dead Zone
python sisyphus_diagram.py
python sisyphus_diagram.py --lang bg # Bulgarian labelsFour-panel diagram from IBM public data: physical qubits (exponential) vs useful logical qubits (flat zero) vs U-Core projection vs Dead Zone ratio.
"Sisyphus pushes the boulder up. The boulder stays at zero useful qubits."
python golden_ratio_challenge.pyGenerates a ranked leaderboard + challenge declaration.
Current standings (Feb 2026):
| Platform | Qubits | SI_Q | Status |
|---|---|---|---|
| Quantinuum H2-1 | 32 | 0.47 | Below 40q threshold |
| QuEra Aquila | 256 | 0.39 | Estimated |
| IBM Heron R2 | 133 | 0.36 | Estimated |
| Google Willow | 105 | 0.34 | Estimated |
| IBM Brisbane | 127 | 0.11 | Measured |
Gap to 0.618: >= 0.148
python u_core_simulator_v25_2.py # interactive
python u_core_simulator_v25_2.py --no-show # headless / CISimulation of a Gaussian Concentric QPU designed to reach SI_Q >= 0.618.
| Metric | IBM Brisbane | U-Core Simulation |
|---|---|---|
| Avg SI_Q | 0.11 | 0.71 |
| Dead Zone | 99.2% | 0% |
| Logical overhead | Surface Code ~10,000:1 | MELQ ~3:1 |
Note: U-Core results are simulation -- not yet validated on real hardware.
The threshold is derived from MELQ DFS encoding at 3:1 overhead with typical superconducting imbalance delta ~ 0.07:
Logical error rate: eps_L = 3 * eps_P^2 (DFS second-order suppression)
Break-even: eps_P < 1/3 -> A_Q > 0.667
With delta=0.07: SI_Q_threshold = 0.667 / (1.07)^2 = 0.618
This falls in the tight neighborhood of phi^-1 = 0.6180... under realistic parameters.
| Encoding overhead | Avg delta | Threshold |
|---|---|---|
| 2:1 (optimistic) | 0.03 | 0.707 |
| 3:1 (MELQ baseline) | 0.07 | 0.618 |
| 4:1 (conservative) | 0.12 | 0.553 |
We do not claim 0.618 is universally exact. We claim no platform reaches even 0.50.
Falsifiable. Public. Open to everyone.
Achieve avg SI_Q >= 0.618 on real quantum hardware with > 40 physical qubits.
- Run
python IBM_Q_SI_Q_Autopsy.py --backend YOUR_BACKEND - Open a Validation Submission issue
- Or email petar@u-model.org with subject
[SI_Q] <backend> <date>
Reward: credited as "Triadic Optimizer 2026/27" in U-Theory v26.0 and next arXiv preprint.
| ID | Prediction | How to falsify |
|---|---|---|
| QC-P1 | TQC achieves SI_Q > 0.65 on 43q U-Core | Show standard VQE on 86q beats TQC-43q |
| QC-P3 | Triadic VQE reduces gate depth 30%+ vs Qiskit Level-3 | Show Qiskit transpiler matches on LiH |
| QC-P17 | delta < 0.3 guarantees polynomial VQA gradient | Show random ansatz with delta<0.3 has exponential gradient |
git clone https://github.com/UniversalModel/Quantum-triadic-autopsy
cd Quantum-triadic-autopsy
pip install -r requirements.txt
# Optional -- for live IBM backend access:
pip install qiskit-ibm-runtimeLive IBM access — one-time token setup:
from qiskit_ibm_runtime import QiskitRuntimeService
QiskitRuntimeService.save_account(channel="ibm_quantum", token="YOUR_IBM_TOKEN", overwrite=True)
# Token: https://quantum.ibm.com/ -> Account -> API tokenThen run: python IBM_Q_SI_Q_Autopsy.py --backend ibm_brisbane
Requirements: numpy >= 1.24, matplotlib >= 3.7, networkx >= 3.0, Python 3.9+
See CONTRIBUTING.md -- the most valuable contribution is running the tools on hardware we do not have access to and reporting the results.
@software{nikolov2026quantum_triadic_autopsy,
author = {Nikolov, Petar},
title = {Quantum Triadic Autopsy -- NISQ hardware diagnostic tools},
year = {2026},
publisher = {GitHub},
url = {https://github.com/UniversalModel/Quantum-triadic-autopsy},
doi = {10.17605/OSF.IO/74XGR}
}| Full theory | U-Theory v25.2 -- Zenodo |
| OSF (DOI) | osf.io/74xgr |
| GitHub Release | TheoryofEverything |
| Author | Petar Nikolov, 2026 |
| Contact | petar@u-model.org |
| Website | U-Model.org |
Copyright (c) 2026 Petar Nikolov -- CC BY 4.0