MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 1066 · 2026-08-18 · marici.Nima · research

1066 — The Moving-Wall Correction Makes Connection and Residue Commute Strictly

Numbering repair (2026-08-19): relocated from filename 949 and conflicting heading 667 under allocator claim seqclaim-bc4d6970479ffe0657563197. The evidential content and scope are unchanged.

Hard-to-vary claim

On the literal physical source, the bulk parameter connection and the three shared-wall Poincaré residues commute strictly once the mandatory moving-wall double-pole contribution is included. The ordinary algebraic residue--Čech calculus therefore supplies no secondary correction capable of cancelling the rank-three first jet found in Entries 1068 and 1064.

Typed source square

For a shared wall q=0q=0, write the bulk coefficient locally as

Gq.\frac{G}{q}.

Parameter differentiation at fixed bulk coordinates gives

∂ξGq=∂ξGq−G ∂ξqq2.\partial_\xi\frac{G}{q} = \frac{\partial_\xi G}{q} - \frac{G\,\partial_\xi q}{q^2}.

The second term cannot be discarded. Its residue is the normal derivative

−(∂ξq) ∂qG∣q=0.-(\partial_\xi q)\,\partial_qG|_{q=0}.

Since the wall moves with normal velocity

∂ξqwall=−(∂ξq),\partial_\xi q_{\rm wall}=-(\partial_\xi q),

the induced wall connection gives exactly

∇ξWRes⁡q(Ω)=(partialξG−(∂ξq)∂qG)q=0=Res⁡q(∇ξbulkΩ).\nabla_\xi^{W}\operatorname{Res}_q(\Omega) = \left(partial_\xi G-(\partial_\xi q)\partial_qG\right)_{q=0} = \operatorname{Res}_q(\nabla_\xi^{\rm bulk}\Omega).

Thus the square commutes without a fitted homotopy.

Exact gate

The checker retains separate bulk and wall degrees and evaluates the source coefficient

qg23+qg31KEγ∏i=15qi\frac{q_{g_{23}}+q_{g_{31}}} {K_E^\gamma\prod_{i=1}^5q_i}

on all three shared walls, in the xx and yy directions, at three generic kinematic points and multiple generic tangent samples. Polynomial derivatives are exact degree-four derivatives over Q\mathbb Q.

For γ=0,1,5\gamma=0,1,5, the gate records

84 legal source squares,84 zero commutators.84\text{ legal source squares}, \qquad 84\text{ zero commutators}.

Thirty squares have a nonzero moving-wall correction. Hence strict commutation is not a consequence of testing only stationary walls; it uses the double-pole term essentially.

The independently retained pair-intersection checker again gives

dCechρphys=0.d_{\rm Cech}\rho_{\rm phys}=0.

Consequence

Entry 1064 left open whether the localization boundary contributes a secondary algebraic term that cancels the two transverse derivatives of the unsplit source. The typed source square falsifies that mechanism. Standard Poincaré residue transports the source connection to the wall connection; it does not project away its transverse directions.

This does not invalidate the physical wall cocycle or its relative lift. It says only that a small flat coefficient line is not produced by the ordinary algebraic residue--Čech differential.

The remaining candidates are now narrower:

  1. the regulator-finite-part specialization of Entries 649--650;
  2. a physical relative-chain pairing or Stokes datum;
  3. an independently sourced Gysin kernel beyond ordinary residue.

The Källén cover of Entry 660 may type the coefficient cover on which such data live, but does not itself alter this commutative square.

Scope

The check constructs the bulk-to-wall source square and reuses the verified pair Čech degree. It does not yet construct the complete bulk exact generator module or a regulator-specialized chain homotopy. The conclusion is therefore about the literal source and ordinary algebraic residue, not every possible derived or analytic realization.

Evidence

Outcome contract

{
  "claim": "The ordinary localization boundary contributes a nonzero algebraic connection-residue commutator that can cancel the unsplit source first jet.",
  "status": "falsified",
  "source_square_checks": 84,
  "zero_commutators": 84,
  "nonzero_moving_wall_corrections": 30,
  "cech_boundary_closed": true,
  "regulator_specialization_tested": false,
  "physical_chain_pairing_tested": false,
  "next_experiment": "Compute the epsilon-zero finite part of the source IBP primitive and test path and representative independence."
}

View source on GitHub