Skip to content

Determine sharp four-qubit Clifford threshold conditioning - #7772

Open
aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-084
Open

aliromman wants to merge 1 commit into
CrocSwap:mainfrom
aliromman:codex/consecutive-result-084

Conversation

@aliromman

@aliromman aliromman commented Nov 19, 2026 •

Copy link
Copy Markdown
Contributor

Enabling stability result: the normalized full four-qubit Clifford Gram matrix at its first independent tensor width, thirteen, has exact condition number 101244/1519. The normalized tensor map has condition number below 8.165 and coefficient-recovery norm below 3.021. The same bounds hold at every larger integer width and for every subalphabet. No new kappa or physical decoder follows; T(4)=13 is unchanged.

Read the proof and recovery contract, complete spectrum, checker, source pins and credits.

The columns are vec(U_a^tensor f)/16^(f/2), one representative per scalar coset of the full Gaussian-dyadic group. The proof binds all 81 source quotient multiplicities to actual projective Gram blocks. There are 52 distinct eigenvalues.

Endpoint Quotient row Exact eigenvalue Full eigenspace dimension
Minimum 1 470624175/4294967296 256
Maximum 2 7841980575/1073741824 313600

No other block attains either endpoint. The Gram condition number is their ratio; the tensor-map condition number is its square root. The inverse coefficient-recovery norm is the reciprocal square root of the minimum. Schur products and principal submatrices extend these bounds to later widths and subalphabets. The constants are sharp for the full width-thirteen alphabet; sharpness at a particular later width or proper subset is not claimed.

Python 3.12 and the standard library:

python3.12 -B scripts/research_package.py --package research/clifford-threshold-conditioning

Both modes print:

PASS_CLIFFORD_THRESHOLD_CONDITIONING
81 blocks, 52 distinct eigenvalues; complete rank 12128668876800
Sharp minimum 470624175/4294967296; sharp maximum 7841980575/1073741824
Gram condition 101244/1519; normalized tensor-map condition < 8.165; recovery norm < 3.021
Complete 24-element one-qubit Gram calibration, cubic annihilator and exact moments pass

All spectral values, complete regular dimensions and trace moments are exact. A separate Pauli-lift count gives the same second moment. The complete one-qubit Gaussian Clifford group supplies a literal 24-by-24 Gram matrix: all 576 group products, direct quotient characters, a cubic annihilator and exact spectral moments agree. Wrong group-order, representation-degree and column-normalization controls fail. No floating-point eigensolver is used.

Eighteen retained files are authenticated. Unchanged source replay, normal/optimized registered replay, syntax/indentation, committed bytes/source history/modes, export and sealed read-only Git-free serial replay pass. Hosted CI did not certify this package. The plan job failed in the shared test_current_ci_section_matches_selected_record: the base guide names frontier6030 while the selected record requires make frontier7000-verify. The log reports 760 tests, one failure and 24 skipped; package replay was skipped. This package changes no shared CI file. Local replay remains the scientific evidence.

Ali Romman (@aliromman) supplied PR #7766's threshold proof, 81 multiplicities and checker. Nick Petalas (@petalas) supplied PR #7747's Sp(8,2) projection and archive parser. alephnan (@alephnan) supplied the projective Gram normalization, odd-sector framework and Schur argument in PR #6714. The ordinary table credits the ATLAS and Thomas Breuer's CTblLib. All earlier notices remain intact.

The published table remains a mathematical input. This package consumes its authenticated projection and does not repeat the optional archive extraction. No full four-qubit group, Gram matrix or inverse was materialized. The full three-qubit and even-sector suites, selected cold replay, Lean and Docker were not run. Native preparation, inverse realization, precision allocation, routing, recovery, stock and complete caller costs remain unpriced. All five framework assumptions and the selected frontier remain unchanged. No independent human/formal-kernel acceptance or global literature priority is claimed.

Base: canonical main e54523517d0f03e58352e647c403a5af3e525a57, priority revision 8. Source pins do not assert ancestry. Prepared by Ali Romman with substantial OpenAI Codex assistance in research, implementation, verification and writing. Apache-2.0. Canonical claim.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant