Definition Lower and upper images in the same edge module
✓ definition formalized

Lower and upper images in the same edge module

Let \(R\) be a commutative ring, \(G\) an ordered graph, and \(B\) a sheaf of \(R\)-modules on \(G\). Fix an edge \(e:x\to y\), and choose outgoing-edge sets \[ F\subseteq\operatorname{Out}(x),\qquad H\subseteq\operatorname{Out}(y). \] The selected kernels are \[ K_x(F)=\{v\in B^x:\rho^-_f(v)=0\text{ for every }f\in F\}, \] and the corresponding kernel \(K_y(H)\) at \(y\).

The lower edge image and upper edge image are \[ L_e(F)=\rho^-_e(K_x(F)),\qquad U_e(H)=\rho^+_e(K_y(H)). \] Both are submodules of \(B^e\). Membership records an actual stalk preimage: \[ z\in L_e(F)\ \Longleftrightarrow \exists v\in K_x(F),\ \rho^-_e(v)=z, \] \[ z\in U_e(H)\ \Longleftrightarrow \exists w\in K_y(H),\ \rho^+_e(w)=z. \]

The named graded constructions

The source's lowerImage and upperImage implement these formulas for a graded sheaf over the polynomial coefficient algebra. Its upwardSubmodule is exactly the selected kernel of the underlying ungraded sheaf, typed as a submodule of the graded stalk.

Thus forgetting grading preserves the same sets, restrictions and embedded images. A later ungraded sheaf decomposition acts on these very submodules.

Homogeneous components

For a graded sheaf whose restrictions have degree zero, every selected kernel is closed under homogeneous components. If \(v\in K_x(F)\), then for any selected edge \(f\), \[ \rho^-_f(v_n)=(\rho^-_f(v))_n=0. \] Hence \(v_n\in K_x(F)\).

If \(z=\rho^-_e(v)\), its degree-\(n\) component is \[ z_n=\rho^-_e(v_n)\in L_e(F). \] The upper map gives the same argument for \(U_e(H)\). This is the component calculation behind upperImage_homogeneous and the graded image constructions used in Hilbert comparison.

Two outgoing conditions in polynomial coordinates

Take \(R=k[a,b]\) over a field \(k\), and a graph with two edges \[ e:x\to y\text{ labelled }a,\qquad f:x\to z\text{ labelled }b. \] Put \(R\) at each vertex, \(R/(a)\) on \(e\), \(R/(b)\) on \(f\), and use quotient maps at both endpoints.

Choose \(F=\{f\}\) and \(H=\varnothing\) at \(y\). Then \[ K_x(F)=bR,\qquad K_y(H)=R. \] Inside the current edge module \(E=R/(a)\), \[ L_e(F)=\overline b\,E,\qquad U_e(H)=E. \] The extra condition along \(f\) has become a divisibility condition in the quotient belonging to \(e\). These are the two embedded images compared by a scalar equality.

Relation to the OpenAI notes

The same embedded images through one edge component.

OpenAI notes: The model decomposition carries both localized kernels and their endpoint images into the model summands. Its single edge component carries both comparison submodules, and linearity carries the correction scalar through that same map. edge:bound.

Lean (primary): Selected-kernel membership and both image equalities are proved for arbitrary sheaf isomorphisms over a commutative ring. ScaledImages is an elementwise witness equivalence, identified with submodule equality. Coordinatewise witness transport proves the product and model-sum statements with one common scalar. PositiveTransfer:362, PositiveTransfer:366, CharacterReciprocity:443, GradedSheaves:631, GradedSheaves:639, TerminalKernels:19, CharacterReciprocity:548.

Role in the proof: The Blueprint defines the actual submodules and witnesses before transporting them. The generic ungraded transport is the one applied to the local model decomposition.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.MomentGraph.GradedSheaf.upwardSubmodule: MinimalExtensions.lean:31.

OAI.KLInvariance.MomentGraph.GradedSheaf.lowerImage: PositiveTransfer.lean:362.

OAI.KLInvariance.MomentGraph.GradedSheaf.upperImage: PositiveTransfer.lean:366.

OAI.KLInvariance.MomentGraph.GradedSheaf.upperImage_homogeneous: PositiveTransfer.lean:371.

Used by