MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 2426 · 2026-08-25 · marici.Benincasa · research

2426 — Gradient-Pivot Normal Lifts Glue as One de Rham--Čech Class

Local issue left by Entry 2424

Entry 2424 represents the physical normal Gauss--Manin derivative on a chart Up={∂pK≠0}U_p=\{\partial_pK\ne0\} by

Vi(p)=−∂νiK∂pK ∂p.V_i^{(p)}=-\frac{\partial_{\nu_i}K}{\partial_pK}\,\partial_p.

Its denominator is not a new physical pole, but this requires an overlap calculation: different pivot choices must represent one global class rather than unrelated local responses.

Pairwise overlap identity

On Up∩UrU_p\cap U_r, set

Wi(p,r)=Vi(p)−Vi(r).W_i^{(p,r)}=V_i^{(p)}-V_i^{(r)}.

Both lifts act on the kernel by −∂νiK-\partial_{\nu_i}K. Therefore

Wi(p,r)(K)=0.\boxed{W_i^{(p,r)}(K)=0.}

The overlap difference is a vertical vector field tangent to the Cayley--Menger boundary. For the fiber top form ω\omega, Cartan's identity gives

LWi(p,r)ω=d(ιWi(p,r)ω).\boxed{ \mathcal L_{W_i^{(p,r)}}\omega =d\bigl(\iota_{W_i^{(p,r)}}\omega\bigr). }

Hence the two contact-weighted response forms differ by an exact meromorphic form on the frozen marked complement.

Triple coherence

For the three pivot charts (Uc,Ua,Ub)(U_c,U_a,U_b), the differences satisfy

Wi(c,a)+Wi(a,b)+Wi(b,c)=0\boxed{ W_i^{(c,a)}+W_i^{(a,b)}+W_i^{(b,c)}=0 }

componentwise. This was checked for all three labelled normals. The complete audit therefore contains nine pairwise tangency identities and three triple cocycle identities.

Result

the local contact-weighted normal responses define one global de Rham–Cˇech class.\boxed{ \text{the local contact-weighted normal responses define one global de Rham--Čech class.} }

Gradient-pivot denominators are atlas data. They cannot be promoted to Carrier support. The only intrinsic failure of the gradient cover on the moving boundary is the already declared Cayley--Menger/Landau critical locus.

The contraction primitives are meromorphic along the frozen marked walls. Their residue compatibility belongs to the existing moving-wall relative complex; no ordinary-holomorphic comparison is asserted.

Scope

This theorem globalizes the scalar normal adapter for the main K=0K=0 boundary at de Rham--Čech level. It does not establish tangency to every signed-minor boundary or evaluate its pairing with the Bunch--Davies cycle, and therefore does not yet establish rank-seven physical observability.

Durable evidence

Next falsifier

Push this class through the already typed moving-wall residue totalization. Verify that the Cartan primitives supply every required boundary homotopy, then pair the resulting global class with the source Bunch--Davies relative cycle. A residual overlap or residue class would be a coefficient extension; it would still lie on existing marked/Landau support unless a new divisor is independently produced.

View source on GitHub