Selected kernels and edge images on the localized coset
Fix a finite-generator Coxeter system, a root plane \(P\) with a chosen plane basis and induced subgroup \(D_P\), a least-length representative \(a_0\) in \(D_Pa_0\), and \(b\in W\). Choose a finite real basis of the extended realization, giving the polynomial coefficient algebra \(A\). Let \[ j:K\to[1,b],\qquad j_V(d)=da_0,\qquad K=\{d\in D_P:1\le_Dd,\ da_0\le_Wb\} \] be the finite relative coset graph map. Let \(C\) be an \(A\)-linear sheaf on \([1,b]\) satisfying the polynomial upper-quotient rule, and put \[ R=R_P=A_{\mathfrak p_P},\qquad C_R=R\otimes_A C,\qquad \mathcal C=j^*(C_R). \]
For an arbitrary set \(T\) of ambient edges, \(K_y^C(T)\) denotes vanishing on the members of \(T\) starting at \(y\). At a relative vertex \(d\), let \(T^K=j_E^{-1}(T)\). Then \[ K_d^{\mathcal C}(T^K) =K_{j_V(d)}^{C_R}(T) =\bigl(K_{j_V(d)}^C(T)\bigr)_R \] inside \(\mathcal C^d=R\otimes_A C^{j_V(d)}\). The subscript \(R\) denotes the embedded image of the tensor of a submodule inclusion.
For a relative edge \(a:d\to c\), and arbitrary ambient edge sets \(T,H\), both selected edge images satisfy \[ \rho_a^{\mathcal C,-}\bigl(K_d^{\mathcal C}(j_E^{-1}(T))\bigr) =\bigl(\rho_{j_E(a)}^{C,-}(K_{j_V(d)}^C(T))\bigr)_R, \] \[ \rho_a^{\mathcal C,+}\bigl(K_c^{\mathcal C}(j_E^{-1}(H))\bigr) =\bigl(\rho_{j_E(a)}^{C,+}(K_{j_V(c)}^C(H))\bigr)_R. \] These are equalities of submodules of the same edge module \(\mathcal C^a=R\otimes_A C^{j_E(a)}\).
Relation to the OpenAI notes
Relative extension through an explicit graph map.
OpenAI notes: The coset intersection is a finite relative lower ideal. A relative upper subset is closed under the surviving plane-labelled edges. Descending ambient extension gives relative flabbiness after localization. loc:relative-flabbiness.
Lean (primary): The same descending extension is factored into extend_closedUnder and a graph-pullback theorem. The coset map supplies vertex injectivity and explicit outgoing lifts; the root/unit criterion makes all outside edge modules zero. A separate theorem records selected-kernel equality under that pullback. Sheaves:106, MinimalExtensions:1520, PlaneLocalization:497, PlaneLocalization:591, TerminalKernels:393.
Role in the proof: Relative order determines the upper subsets on the coset, while the ambient graph supplies the extension. The localized stalks and edge maps remain the actual tensor-extended ambient modules, so selected kernels and their edge images retain their embeddings.
Proof
Step 1. Specify the surviving edges and their lifts
Set \(L=\{e:\tau_e\in D_P\}\), where \(\tau_e\) is the ambient edge’s left reflection label. The localized coset properties theorem shows that every edge module of \(C_R\) outside \(L\) is zero.
At the fixed vertex \(d\), the coset outgoing-lift theorem supplies a relative outgoing lift for each edge in \(L\) starting at \(j_V(d)\).
Step 2. Apply the selected-kernel pullback theorem
Apply the selected-kernel pullback theorem to \(C_R,j,d,T,L\). Step 1 verifies its two hypotheses, and it gives \[ K_d^{\mathcal C}(j_E^{-1}(T)) =K_{j_V(d)}^{C_R}(T) \] in the identical stalk module.
In detail, every selected relative equation is an ambient selected equation. Conversely, a selected ambient equation either has a relative lift or takes values in a zero edge module. These are precisely the two containment arguments in that theorem.
Step 3. Identify the ambient localized kernel
The ambient Bruhat graph has finitely many edges, and \(R=A_{\mathfrak p_P}\) is flat over \(A\). The flat kernel-and-image theorem gives \[ K_{j_V(d)}^{C_R}(T) =\bigl(K_{j_V(d)}^C(T)\bigr)_R. \] Combining with Step 2 proves both equalities of selected kernels.
Step 4. Transport the lower edge image
The pullback lower restriction at \(a\) is the lower restriction of \(C_R\) at \(j_E(a)\), under the endpoint identification. Step 2 therefore gives \[ \rho_a^{\mathcal C,-}(K_d^{\mathcal C}(j_E^{-1}(T))) =\rho_{j_E(a)}^{C_R,-}(K_{j_V(d)}^{C_R}(T)). \] The lower-image part of the flat kernel-and-image theorem identifies the right side with \(\bigl(\rho_{j_E(a)}^{C,-}(K_{j_V(d)}^C(T))\bigr)_R\).
Step 5. Transport the upper edge image
Apply Steps 2–3 at \(c\) with the independently chosen set \(H\). The pullback upper restriction is the upper restriction of \(C_R\) at the same ambient edge \(j_E(a)\). The upper-image part of the flat kernel-and-image theorem gives the second displayed image equality.
Both calculations use that single ambient edge module. They supply the submodules to which the later local model isomorphism is applied.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.TitsSpace.PlaneBasis.localizedCoset_kernel: TerminalKernels.lean:393.
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.