Extending a sheaf morphism from an upper set
Let \(k\) be a field and \(A\) a connected nonnegatively graded Noetherian domain over \(k\). Let the ordered graph have finite vertex set \(V\) and edge set \(E\), with labels \(\alpha_e\).
Suppose \(B,C\) are graded sheaves satisfying the upper-quotient rule for these labels. Assume both are flabby and globally generated, and every stalk of \(B\) is free.
For an upper set \(U\), every degree-zero partial morphism \(j\) on \(U\) extends to a degree-zero sheaf morphism \(f:B\to C\) with \[ f_x=j_x\qquad(x\in U). \] The upper-quotient rule also makes its edge maps agree with \(j_e\) whenever \(t(e)\in U\).
Relation to the OpenAI notes
Graded extension preceding local decomposition.
OpenAI notes: The local-ring decomposition descends through the coset vertices, lifting stalk maps over the common boundary image and retaining both endpoint restrictions. loc:decomposition.
Lean (primary): The graded morphism-extension theorem works over a connected nonnegatively graded Noetherian domain. A homogeneous free-basis lift constructs the next stalk map, upper quotients construct the incoming edge maps, and finite-complement induction extends the partial morphism. SheafMorphisms:675, SheafMorphisms:723, SheafMorphisms:834.
Role in the proof: This graded extension is used in boundary recognition and splitting. The ungraded local-ring decomposition is a later construction with its own local-ring and model hypotheses.
Proof
Step 1. Set up induction on the missing vertices
Let \(T=V\setminus U\). Use strong induction on the finite set \(T\), proving extension for every partial morphism on its upper complement.
If \(T=\varnothing\), then \(U=V\). The partial morphism already supplies all vertex maps, all edge maps and both commuting squares. The source's ofUniv construction regards these as a full morphism.
Step 2. Choose a maximal missing vertex
Assume \(T\ne\varnothing\), and choose a maximal element \(x\in T\). If \(x<y\), then \(y\notin T\): otherwise maximality would force \(y\le x\), contradicting \(x<y\). Hence every vertex strictly above \(x\) lies in \(U\).
Apply the boundary-morphism lifting theorem. Its hypotheses hold because the edge set is finite, the source stalk \(B^x\) is free, and both sheaves are flabby and generated. It gives a degree-zero \(l_x:B^x\to C^x\) with \[ r_C l_x=j_{\partial x}r_B. \] Evaluating this tuple equation at an outgoing edge gives \(j_e\rho^{B,-}_e=\rho^{C,-}_e l_x\).
Step 3. Extend across the chosen vertex
The one-vertex morphism extension theorem applies to \(U,x,j,l_x\) and the upper-quotient rules. It gives a partial morphism \(j'\) on \(U'=U\cup\{x\}\) that agrees with all old vertex maps.
The new set is upper. If \(y\in U\) and \(y\le z\), then \(z\in U\). If \(y=x\), either \(z=x\), or \(x<z\) and the maximality argument places \(z\) in \(U\).
Its complement is \(T\setminus\{x\}\), a strict subset of \(T\).
Step 4. Apply the strictly smaller induction instance
The induction hypothesis extends \(j'\) to a full morphism \(f\) agreeing on \(U'\). For each \(y\in U\), \[ f_y=j'_y=j_y. \] The first equality is the smaller-set induction conclusion; the second is preservation in the one-vertex construction. This completes the induction and proves the vertex agreement claimed by exists_extension.
Step 5. Recover agreement on the old edge maps
Let \(e\) have target \(y\in U\). Both morphisms have the upper commuting square, and \(f_y=j_y\), so \[ f_e\rho^{B,+}_e =\rho^{C,+}_e f_y =\rho^{C,+}_e j_y =j_e\rho^{B,+}_e. \] The upper restriction \(\rho^{B,+}_e\) is surjective. Every edge vector has an upper-stalk preimage, and evaluating the displayed equality on that preimage gives \(f_e=j_e\).
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.MomentGraph.GradedSheaf.OnHom.ofUniv: SheafMorphisms.lean:607.
OAI.KLInvariance.MomentGraph.GradedSheaf.OnHom.exists_extension: SheafMorphisms.lean:834.