Repository navigation
Conversation
…51/mixed bit word on CrocSwap#451 (+0.670%) Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01TGoMCHAqctZe4wtb4qr7vv
|
Full verification PASS (uninterrupted, not --quick):
|
|
Correction: please do not admit this κ as it stands. The complex kernel here hosts each entering helper's residual as a width-8 interval. Its route is the exact lowering of T_FULL·T_E⁻¹ (monomial·C₈·monomial), but nothing places the helper's ninth, spectator coordinate. The same gap was identified for #449's hosting by #458, and #484 notes that complex entrances are isotropic rather than coordinate entrances. #471 closes the gap with full-invocation normalisation and same-owner spectator borrowing (credit to Dugongue). This package does neither. What remains valid here:
The hosted complex accounting, and therefore κ = 0.001274415506145115, depends on the unplaced spectator, so I am withdrawing that claim and returning the PR to draft. The bit-side packages #457 and #459 are unaffected: bit entrances are coordinate entrances with explicit residual isometries. The 56 proportional-pair entries found here are of the same type as #449's kernel79 entries; I have not checked whether they are a subset. |
Conditional κ = 254883101229023 / (2·10¹⁷) = 0.001274415506145115: +0.670% over #451 (0.001265931727546219) and +0.622% over our #459. Bit binds at c ≈ 1.276041713·10⁻³ (#459's word, unchanged). The complex endpoint rises to b ≈ 1.293526619·10⁻³ (#451: 1.268142297·10⁻³).
New: a shared-donor response kernel on the complex supplier
#433's split-centre complex word cancels each helper's dirt with frame-zero target additions (
compensation, 12,760 rational adds), just as the bit word does with its D gates. The #417 kernel therefore carries over with rational coefficients:The word has exactly 56 proportional helper pairs, and every pair shares an entrance vector. That gives 56 pivots, 56 donors and entrance rank 56, so 56 complex residuals shrink from width 9 to width 8. Width-8 residuals pack exactly into the rank-8k idle children, so the remainder children drop:
{1:174864, 2:78848, 3:8680, 4:19712, 5:38120, 7:2944, 8:8000, 30:8, 64:2160};Checks, all regenerated by
verify.pycode/ckernel.py).complex_phases_full.py(≈3.5 min).code/complex_kreplay.py):lower_childas monomial·F_r·monomial with r = 9 − e, re-verified on both tableau and phase.Status:
verify.py. The result will be posted.Review point: as in the bit kernel lineage, a helper may start at a nonzero frame. Its hosted residual is the rank-(9−e) child from its start frame to FULL, carried by the stored exact Clifford route.
PROOF.md("Complex response kernel") gives the scalar identity and the route argument.Bit side: #457's rank151 kernel plus #450's target-prefix mixing (DreamingOfClouds), as in #459. Inherited interfaces are unchanged, including #442 hosting. Cross-copy hosting is not used. No unconditional theorem is claimed.
Credits:
Complex kernel, composition and packaging by Rohan Arun with Claude Opus 5.5 (Anthropic).
🤖 Generated with Claude Code
https://claude.ai/code/session_01TGoMCHAqctZe4wTb4Qr7VV