Repository navigation
Qutrit conformance: pinned css-decoupling SUM/SWAP and toric fixtures - #624
Conversation
Reviewer's GuideAdds an offline, finite GF(3) qutrit conformance suite backed by a reproducibly pinned css-decoupling donor: a guarded generator produces canonical SUM/SWAP and toric fixtures, independent tests replay and validate their algebra and QEC behavior, and documentation records provenance, licensing, reproduction steps, and strict claim boundaries without altering production decoder code or dependencies. Sequence diagram for pinned qutrit fixture regenerationsequenceDiagram
participant Generator as generate_qutrit_css_fixtures.py
participant Donor as Pinned css_decoupling checkout
participant SymPy as SymPy and mpmath
participant Fixture as JSON fixture
Generator->>Donor: check_checkout()
Donor-->>Generator: pinned clean commit
Generator->>Donor: import Matrix, Ring, decouple, verify_result
Generator->>SymPy: verify pinned versions
Generator->>Donor: decouple(hx, hz, algorithm="general", output="circuit")
Donor-->>Generator: circuit_ready toric result
Generator->>Donor: verify_result(result, expected_HX=hx, expected_HZ=hz)
Donor-->>Generator: valid
Generator->>Fixture: canonical_bytes(build_fixture())
alt --check
Generator->>Fixture: compare bytes
else --output
Generator->>Fixture: open("xb") and write new artifact
end
Flow diagram for offline GF(3) qutrit conformanceflowchart LR
Fixture[Saved SUM/SWAP and toric fixtures] --> Replay[Independent finite-torus gate replay]
Replay --> Actions[Compare X/Z and inverse actions]
Replay --> Pairing["Preserve GF(3) symplectic pairings"]
Fixture --> Toric[Expand toric input on finite tori]
Toric --> Ranks[Compare ranks and logical dimensions]
Toric --> Syndrome[Check syndrome covariance]
Toric --> Decoder[ExactDecoder corrects 144 single-site errors]
Actions --> Result[Finite software conformance]
Pairing --> Result
Ranks --> Result
Syndrome --> Result
Decoder --> Result
File-Level Changes
Tips and commandsInteracting with Sourcery
Customizing Your ExperienceAccess your dashboard to:
Getting Help
|
|
Sorry @EmergentMonk, you've used your own review budget of 250,000 diff characters for the last 7 days. You can request another review in 8 hours and 17 minutes by commenting |
Codex Review SummaryThis comment shows the latest Codex review activity on this pull request.
ℹ️ About Codex in GitHubYour team has set up Codex to review pull requests in this repo. Reviews are triggered when you
Codex reacts with 👀 while any review is running, comments if it has suggestions, and reacts with 👍 once all reviews finish with no findings. |
QEC's exact qutrit tests now have an external prime-field donor from He et al., Toric codes all the way down (arXiv:2610.40592v1). Pin css-decoupling at
a48048c1097f4d1b37a55eca9a1796a823f0c124and add six small circuit fixtures, including the general algorithm's verified 17-gate toric output with four input ancillas.The 27 new offline tests independently replay SUM/SWAP actions and inverses on finite tori, check GF(3) symplectic pairings, detect four wrong gate conventions, compare toric ranks and syndrome covariance, and correct all 144 single-site errors on the 3×3 torus using QEC's unchanged decoder. The optional generator verifies the donor result against its input, checks source/import and dependency pins, compares every fixture byte, and refuses output overwrites. README, a linked research/roadmap supplement, source notes and MIT attribution are included. The migration-bound roadmap and preservation manifest remain unchanged.
Claim boundaries: the separate Lean formalization does not verify Python; the default SymPy backend does not inherit Proposition 1's polynomial runtime guarantee. These are finite software comparisons, not hardware, threshold, speedup, acoustic-protection or decoder-promotion evidence. Offline tests do not independently verify the full upstream decoupling certificate. No production decoder source, historical fixture, release number or core dependency changes.
Validation at
8851e4ad5792d3434cc3885ad08da48ef307ad31git diff --check: passed; pushed tree matches local files.Skipped paths include unavailable cryptography, Stim/PyMatching, CuPy and other optional modules, an unbuilt Rust TUI, a Windows-only case and a sandbox socket restriction. No GPU execution or Lean build is claimed.
GitHub Crossbar validation and Proof HTML checks have succeeded. The GitHub full regression/container pipeline is still in progress; local results above are complete. PR remains unmerged.
Summary by Sourcery
Add bounded, independently replayed GF(3) qutrit conformance coverage using pinned SUM/SWAP and toric-code fixtures without changing production decoders or core dependencies.
New Features:
Enhancements:
Documentation:
Tests: