Skip to content

qubit_6_2_3, sslp (arXiv:2610.29955) - #450

Merged
valbert4 merged 4 commits into
mainfrom
paper/2610.29955
Nov 12, 2026
Merged

valbert4 merged 4 commits into
mainfrom
paper/2610.29955

Conversation

@valbert4

@valbert4 valbert4 commented Nov 7, 2026

Copy link
Copy Markdown
Member

Changes:

  • ~ sslp.description: now opens with the construction (logical states on disjoint subset-sum sectors of a finite-order diagonal transversal gate, then an LP screen), for any code dimension. Adds completeness (Thm. III.2/III.4, Cor. III.5) and the gate-order bound (Thm. III.9). BD examples kept only in transversal_gates, now cited.
  • ~ qubit_6_2_3.description: "smallest known" replaced by the proven result: minimum length among distance-≥3 qubit codes whose single-block transversal gates cannot all be made Clifford in one logical basis. Also: up to local unitaries and permutations, the only ((6,2,3)) code with an order-five transversal logical action (Cor. IV.7).
  • ~ qubit_6_2_3.features.transversal_gates: removed a duplicate item; five is the largest logical order of a transversal diagonal gate on any ((6,2,3)) code (Thm. IV.3).
  • ↔ qubit_6_2_3 → sslp (now sole parent): logical states lie in subset-sum sectors a·x≡0 and a·x≡4 mod 5 with a=(1,1,1,1,2,3) (checked numerically). The implied qubits_into_qubits and small_distance_quantum parents are dropped.
  • declined: the 66-candidate and 15-realizable class counts (they count symmetry data, not codes), and the weight-enumerator sets (two still open).

Open questions:

  • The existing sslp cousins need a call: the binary_dihedral_permutation_invariant detail copies transversal_gates without a citation, and the stab_15_1_3 detail only compares qubit counts. Rewrite each as a cited relationship, or delete?

https://arxiv.org/abs/2610.29955

🤖 Generated with Claude Code

https://claude.ai/code/session_01NyJvrSZfCdKrfziCXK643Q

Tamara and others added 2 commits November 7, 2026 13:58
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NyJvrSZfCdKrfziCXK643Q
Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NyJvrSZfCdKrfziCXK643Q
@valbert4

Copy link
Copy Markdown
Member Author

Check if there are relations between sslp and the PI codes. if none, delete both existing sslp cousins

Tamara and others added 2 commits November 11, 2026 20:54
…tab_15_1_3 (arXiv:2610.29955)

No source relates SS-LP codes to permutation-invariant codes: arXiv:2504.20847
only compares qubit counts with the ((11,2,3)) and ((19,2,3)) binary-dihedral
PI codes, and the stab_15_1_3 detail was a qubit-count comparison.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NyJvrSZfCdKrfziCXK643Q
@valbert4 valbert4 added the comments addressed Zookeeper's review comments addressed; ready for another look label Nov 12, 2026
@valbert4

Copy link
Copy Markdown
Member Author

Addressed in 36b9f47 (main merged in df7e290): found no relation between SS-LP and PI codes, so both sslp cousins (binary_dihedral_permutation_invariant and stab_15_1_3) are deleted.

  • 2504.20847 only says its 7-qubit T and √T codes use fewer qubits than the ((11,2,3)) and ((19,2,3)) binary-dihedral PI codes (2310.17652).
  • A numpy check (2610.29955.verify/bd_pi_vs_sslp.py) shows the SS-LP BD16, BD36 and BD64 codes (n=7, 7, 8) are not permutation-invariant. Their two-qubit reduced purities differ from pair to pair, so none is equivalent to a PI code under local unitaries and qubit permutations.
  • The PI codes do fit the SS-LP form with equal angles on every qubit, e.g. (T^3)^{⊗11} on the ((11,2,3)) code. But Cor. III.5 gives that for any code with a diagonal transversal gate, so no cousin was kept on that basis.
  • Side finding: the BD32 codewords printed in 2504.20847 fail the Knill–Laflamme conditions. The squared coefficients sum to 31, not 32, and the printed angles are off by a factor of 2. Setting the |0001101⟩ coefficient to √2 at θ=0 gives a valid code, which is also not equivalent to any PI code.

@valbert4
valbert4 merged commit 683ec85 into main Nov 12, 2026
1 check failed
valbert4 pushed a commit that referenced this pull request Nov 12, 2026
…he FLM compile on main

A lone % is rejected by FLM (flm.lang.percent-forbidden). It arrived with PR #444 and has failed
every PR check since (#449, #450, #455).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01NyJvrRprUF7ohuENJJ84qp
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

comments addressed Zookeeper's review comments addressed; ready for another look

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant