Flat base change of selected kernels and edge images
Let \(\varphi:R\to S\) be a flat homomorphism of commutative rings, and let \(B\) be an \(R\)-linear sheaf on an ordered graph with finitely many edges. For a vertex \(x\) and a set \(F\) of outgoing edges at \(x\), \[ (K_x^B(F))_S=K_x^{B_S}(F) \quad\text{inside }S\otimes_R B^x. \] Here \(P_S\) denotes the image of \(S\otimes_R P\) in the ambient tensor module.
For an edge \(e:x\to y\), let \(F\) be a set of outgoing conditions at \(x\), and \(H\) one at \(y\). Then \[ \bigl(\rho^{B,-}_e(K_x^B(F))\bigr)_S =\rho^{B_S,-}_e(K_x^{B_S}(F)), \] \[ \bigl(\rho^{B,+}_e(K_y^B(H))\bigr)_S =\rho^{B_S,+}_e(K_y^{B_S}(H)). \] Both image equalities hold inside the same module \(S\otimes_R B^e\).
If \(B\) also satisfies the upper-quotient rule for labels \(\alpha_e\), an edge with invertible \(\varphi(\alpha_e)\) has zero module after base change, and its outgoing vanishing condition is automatic.
Relation to the OpenAI notes
Explicit finite-flat construction.
OpenAI notes: The localization discussion uses flat extension of kernels and images and then decomposes the localized sheaf over the local ring on a dihedral coset. loc:models, loc:decomposition.
Lean (primary): The scalar-extension lemmas first treat general commutative rings and graph sheaves. Finite product maps identify tensor constraints; flat exactness gives section and selected-kernel identities. Image commutation is proved separately by tensor generators. MinimalExtensions:1157, MinimalExtensions:1162, TerminalKernels:279, TerminalKernels:360.
Role in the proof: These algebraic results precede the root-plane and coset specialization. The later pullback argument controls the surviving edges, while both localized edge images retain their common ambient module.
Proof
Step 1. Present the selected kernel by one constraint map
The outgoing constraint \[ \kappa_{x,F}:B^x\to\prod_{e\in F}B^e,\qquad v\mapsto(\rho^-_e(v))_e \] has kernel \(K_x^B(F)\).
The edge set is finite, so \(F\) is finite. Tensoring its product gives the coordinate isomorphism \[ \Psi_F:S\otimes_R\prod_{e\in F}B^e \xrightarrow{\sim}\prod_{e\in F}(S\otimes_R B^e). \] It sends \(s\otimes(v_e)_e\) to \((s\otimes v_e)_e\); the inverse sums the tensor coordinate inclusions.
Step 2. Compare the old and new constraints
The constraint of \(B_S\) satisfies \[ \kappa^{B_S}_{x,F}=\Psi_F(1\otimes\kappa^B_{x,F}). \] For a pure tensor \(s\otimes v\), both sides have \(e\)-coordinate \(s\otimes\rho^-_e(v)\). Additivity proves the identity for all tensors.
Since \(\Psi_F\) is injective, an element is killed by the new constraint exactly when it is killed by \(1\otimes\kappa^B_{x,F}\).
Step 3. Apply the embedded-kernel identity
Flatness gives \[ (K_x^B(F))_S =(\ker\kappa^B_{x,F})_S =\ker(1\otimes\kappa^B_{x,F}). \] Step 2 identifies the last kernel with \(\ker\kappa^{B_S}_{x,F}\), which is \(K_x^{B_S}(F)\). Every equality takes place in \(S\otimes_R B^x\).
Step 4. Commute each edge image with scalar extension
The embedded-image theorem gives \[ \bigl(\rho^{B,-}_e(K_x^B(F))\bigr)_S =(1\otimes\rho^{B,-}_e)\bigl((K_x^B(F))_S\bigr). \] Insert Step 3 and the definition \(\rho^{B_S,-}_e=1\otimes\rho^{B,-}_e\). This proves the lower-image formula.
Apply the same calculation to the upper restriction and \(K_y^B(H)\) to obtain the upper-image formula. The resulting submodules share the ambient edge tensor module.
Step 5. Evaluate a unit-label edge
Under the upper-quotient hypothesis, the base-change property theorem gives a surjective upper restriction with kernel \(\varphi(\alpha_e)B_S^y\).
If \(\varphi(\alpha_e)\) is a unit, every \(v\in B_S^y\) is \(\varphi(\alpha_e)(\varphi(\alpha_e)^{-1}v)\), so the whole stalk lies in the kernel. The upper map is zero and surjective, forcing \(B_S^e=0\).
Every restriction to this zero module vanishes. Deleting that equation therefore leaves any selected outgoing kernel unchanged. Taking \(S\) to be the root-plane localization ring gives this algebraic transport on the full Bruhat graph. The next step restricts the resulting sheaf to a surviving coset.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.MomentGraph.Sheaf.kernelConstraint_tensor: TerminalKernels.lean:330.
OAI.KLInvariance.MomentGraph.Sheaf.baseChange_upwardKernel: TerminalKernels.lean:347.
OAI.KLInvariance.MomentGraph.Sheaf.baseChange_lowerImage: TerminalKernels.lean:360.
OAI.KLInvariance.MomentGraph.Sheaf.baseChange_upperImage: TerminalKernels.lean:367.
OAI.KLInvariance.MomentGraph.Sheaf.edge_subsingleton_of_unit: PlaneLocalization.lean:534.