Image, symmetry, and grading of the local intersection map
Fix a finite-generator Coxeter system, its polynomial coefficient algebra \(A\), and the actual normalized boundary sheaf \(B(b)\) on \([1,b]\). At \(x\in[1,b]\), let \(M=B(b)^x\), let \(K_x\) be the full outgoing kernel, and let \[ I_x:M^*\to M,\qquad I_x=\iota_x\kappa_x^{-1} \] be the local intersection map from the constructed symmetric section duality. Put \(d=\ell(b)-\ell(x)\).
The map is injective and \(\operatorname{im}I_x=K_x\). It satisfies \[ \lambda(I_x\mu)=\mu(I_x\lambda). \] If \(\lambda(M_j)\subseteq A_{j+r}\) for every \(j\), then \[ I_x\lambda\in M_{r+d}. \] At every proper vertex \(x<b\), \[ \operatorname{im}I_x\subseteq A_+M. \]
Relation to the OpenAI notes
Different intermediate degree statement.
OpenAI notes: The normalization proposition identifies stalk graded ranks with canonical KL polynomials using the cited character theorem and a conversion from degree-two to degree-one labels. bg:normalization, eq:bg-ranks.
Lean (primary): The actual local intersection map is constructed from integral stalk–costalk duality. Boundary minimality puts its image in the positive ideal, and the strict degree theorem assumes that the specified diagonal coefficient is nonzero. PositiveTransfer:132, PositiveTransfer:160, PositiveTransfer:186.
Role in the proof: The local calculation gives \(2d_i<\ell(b)-\ell(x)\) for the basis generator named in that nonzero-diagonal hypothesis. Actual-character reciprocity supplies the endpoint identity in the comparison route.
Proof
Step 1. Prove injectivity and identify the image
The inverse duality \(\kappa_x^{-1}:M^*\to K_x\) is injective, as is the submodule inclusion \(\iota_x\). Their composite is therefore injective.
Every value \(I_x\lambda\) lies in \(K_x\) by construction, proving \(\operatorname{im}I_x\subseteq K_x\).
Conversely, take \(v\in K_x\). Take \(\lambda=\kappa_x(v)\). Then \[ I_x\lambda=\iota_x\bigl(\kappa_x^{-1}(\kappa_x(v))\bigr)=v. \] This proves the other containment and the image equality.
Step 2. Apply boundary-cover minimality
For \(x<b\), the boundary sheaf's minimality field states that an assignment \(f\) whose outgoing restrictions vanish at \(x\) satisfies \(f_x\in A_+M\).
Given \(v\in K_x\), define \(f_x=v\) and \(f_y=0\) for every \(y\ne x\). Each outgoing restriction at \(x\) vanishes because \(v\in K_x\). The minimality field therefore gives \(v\in A_+M\).
Conversely, a containment \(K_x\subseteq A_+M\) immediately applies to the \(x\)-coordinate of any such assignment. This explains the source's kernelMinimal_iff conversion.
Step 1 now gives \[ \operatorname{im}I_x=K_x\subseteq A_+M. \]
Step 3. Transport the homogeneous functional
The actual section pairing has shift \(L=\ell(b)\). The inverse stalk–costalk degree theorem sends a functional of degree \(r\) to kernel degree \[ r+L-n_x, \] where \(n_x\) counts incoming edges.
On the lower Bruhat interval \([1,b]\), incoming edges correspond to the reflection inversions of \(x\), so \(n_x=\ell(x)\). Hence \[ I_x\lambda\in M_{r+\ell(b)-\ell(x)}=M_{r+d}. \] Inclusion of the kernel preserves the induced grading.
Step 4. Apply inverse-duality symmetry
The chosen actual section pairing is symmetric. The stalk–costalk symmetry theorem therefore gives \[ \lambda\bigl(\kappa_x^{-1}(\mu)\bigr) =\mu\bigl(\kappa_x^{-1}(\lambda)\bigr). \] Evaluation uses the underlying stalk vector, so inserting the inclusions gives the claimed formula for \(I_x\).
Step 5. Read the matrix entries
For a homogeneous basis \((m_i)\) with degrees \(d_i\), the dual functional \(\lambda_j\) has degree \(-d_j\). Step 3 gives \(I_x\lambda_j\in M_{d-d_j}\). Applying \(\lambda_i\), whose degree is \(-d_i\), gives \[ a_{ij}=\lambda_i(I_x\lambda_j)\in A_{d-d_j-d_i}. \] Step 4 gives \(a_{ij}=a_{ji}\).
At \(x<b\), write \(I_x\lambda_j=\sum_s c_sv_s\) with \(c_s\in A_+\), using Step 2. Then \[ a_{ij}=\sum_s c_s\lambda_i(v_s)\in A_+. \] Thus every coordinate has both the specified homogeneous degree and positive-ideal membership.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.BruhatGraph.localIntersection_injective: PositiveTransfer.lean:140.
OAI.KLInvariance.BruhatGraph.localIntersection_range: PositiveTransfer.lean:146.
OAI.KLInvariance.BruhatGraph.localIntersection_minimal: PositiveTransfer.lean:160.
OAI.KLInvariance.BruhatGraph.localIntersection_homogeneous: PositiveTransfer.lean:167.
OAI.KLInvariance.BruhatGraph.localIntersection_symmetric: PositiveTransfer.lean:180.