Download RELEASE_NOTES.md from Snapkitty/topological-quantum-computer: direct link, hf CLI and curl.
- Browser
- Download file 3.29 kB
-
https://huggingface.co/Snapkitty/topological-quantum-computer/resolve/main/RELEASE_NOTES.md
- Command line
-
hf download hf://Snapkitty/topological-quantum-computer/RELEASE_NOTES.md
-
curl -L -o RELEASE_NOTES.md https://huggingface.co/Snapkitty/topological-quantum-computer/resolve/main/RELEASE_NOTES.md
Topological Quantum Computer v1.0.1
Release type: staged research package Release date: 2026-08-18
v1.0.1 Correction
- Removed the temporary non-Python policy backend that skewed repository language metrics.
- Added the Q-Lambda DSL implementation in
python/qlambda/compiler.py. - Added explicit SHA-520 arrays in
python/qlambda/arrays.py. - Updated SHA-520 to emit a 65-byte, 520-bit digest from the 9-word IV surface.
- Added the QIR-to-Fibonacci-braid resource backend in
python/topological/. - Added focused tests for arrays, DSL compilation, policy selection, and braid resource estimates.
Summary
This release packages the topological quantum-computing research repository for public review. It presents a Fibonacci-anyon model, reduced-round SHA-family experiments, a constraint/proof-search boundary, explicit falsification criteria, tri-license terms, and audit notes.
The core claim is deliberately bounded: generic SHA-style preimage search does not gain more than Grover-style square-root speedup, and the model does not demonstrate a practical full-round cryptanalytic attack.
Included
- Lean 4 formalization surfaces for Fibonacci anyons, logical qubits, braid compilation, and quantum gates.
- Python modules for reduced-round classical validation, toy permutations, Q-Lambda reversible-oracle synthesis, Grover-style search, tensor-network simulation, and Qiskit integration paths.
- Four experiment phases covering classical validation, quantum simulation, resource validation, and theoretical topological compilation.
- Documentation for architecture, falsification, resource analysis, threat model, experiment protocol, cryptanalysis notes, and setup.
- PAX-style tri-license file and array-backed Python license-policy backend.
- Package manifest and About metadata for GitHub release hygiene.
Validation Snapshot
Observed locally during the v1.0.1 correction pass:
| Check | Result |
|---|---|
| Python AST syntax scan | PASS |
pyproject.toml parse |
PASS |
| Module import smoke test | PASS |
| Phase 1 classical validation | PASS |
| Phase 2 quantum simulation | RESOURCE_ESTIMATE_NO_QISKIT when Qiskit is unavailable |
| Phase 3 resource validation | ESTIMATE_ONLY |
| Phase 4 topological compilation | PASS-THEORETICAL |
| Lean/Lake build | PASS-LOCAL for staged Lean modules |
Production Boundary
For this release, "production" means packaged, auditable, and documented as a research artifact. It does not mean physical topological quantum hardware, machine-checked closure of every theorem, full-round cryptanalysis, or commercial deployment.
License
This release follows LICENSE.tri:
- BSL-1.1 source-available path with commercial restrictions until
2028-08-08. - AGPL-3.0 network-copyleft path.
- MPL-2.0 file-level copyleft path.
- Commercial license path for copyleft bypass.
Use the policy engine:
PYTHONPATH=python python -m qlambda.license_policy select saas_wrapper
PYTHONPATH=python python -m qlambda.license_policy select enterprise_restricted
PYTHONPATH=python python -m qlambda.license_policy select file_level_mod
PYTHONPATH=python python -m qlambda.license_policy select copyleft_bypass