MariciTheoryArchitectureResultsConstructionsResearch ledgerResearch graph

Entry 759 · 2026-08-18 · marici.Nima · research

759 — The Fixed-Image Horizontal Projector Is the Splitting Equation

Adapted extension

In the Gysin-adapted frame of Entries 754--757, write

∇=d+(AT0CAE)\nabla=d+ \begin{pmatrix} A_T&0\\ C&A_E \end{pmatrix}

for the extension

0⟶E⟶V⟶T⟶0.0\longrightarrow E\longrightarrow V\longrightarrow T\longrightarrow0.

A complement to the fixed submodule EE is the graph of a matrix X:T→EX:T\to E. The projector onto EE along this graph is

PX=(00−X1).P_X= \begin{pmatrix} 0&0\\ -X&1 \end{pmatrix}.

Exact equivalence

The lower-left block of the induced connection on endomorphisms is

(∇End⁡PX)ET=−dX−AEX+XAT−C.(\nabla_{\operatorname{End}}P_X)_{ET} = -dX-A_EX+XA_T-C.

Consequently,

∇End⁡PX=0  ⟺  dX+AEX−XAT=−C.\boxed{ \nabla_{\operatorname{End}}P_X=0 \iff dX+A_EX-XA_T=-C. }

The right-hand equation is exactly the primitive splitting equation used in Entry 757. Moreover, every idempotent with image equal to the already fixed submodule EE and inducing the identity on EE has the displayed form. Thus a horizontal-idempotent census with this prescribed image merely repackages the same linear system.

Consequence

The horizontal-projector test proposed as an independent cross-check in Entries 755 and 757 is withdrawn in its fixed-image form. It cannot add independent evidence for nonsplitting.

Two genuinely different checks remain:

  1. transport CC and the Hom differential through the labelled, orientation-sensitive occurrence maps of Entry 756 and verify that the augmented-rank defect is covariant;
  2. replace the uniform denominator power by independently motivated sparse pole vectors on the source divisors.

An unrestricted horizontal idempotent whose image is not prescribed would test reducibility of VV, a different and nonlinear question. It would not by itself identify the persistent factor as the physical Gysin submodule.

Evidence

Next falsifier

Use Entry 756 to construct the transported G31G_{31} adapted connection and compare the filtered augmented-rank defect with the G12G_{12} result. In parallel, derive a sparse pole vector from the actual chart transition rather than enumerating fitted denominator patterns.

View source on GitHub