The stalk-character identification condition
Fix a finite-generator Coxeter system, \(u\le b\), and a finite real basis of its extended realization. Use the resulting degree-one polynomial coefficient algebra \(A\), and let \(B_{u,b}\) be the chosen normalized boundary sheaf on \([u,b]\).
For \(x\in[u,b]\), its actual stalk character is \[ C_x^{[u,b]}(q)=\operatorname{ch}(B_{u,b}^x;q)\in\mathbb Z[q]. \] If a homogeneous free basis of the stalk has degrees \(d_i\), this character is \(\sum_iq^{d_i}\). Let \(P_{x,b}(q)\) denote the canonical KL polynomial defined by the R-recurrence and strict degree normalization.
The stalk-character identification condition is the proposition \[ \forall x\in[u,b],\qquad C_x^{[u,b]}(q)=P_{x,b}(q) \quad\text{in }\mathbb Z[q]. \] This is the complete content of the Lean definition StalkCharacterObligation.
Polynomial equality and basis degrees
The equality says, coefficient by coefficient, that the number of stalk basis generators in degree \(n\) is the coefficient of \(q^n\) in \(P_{x,b}\). Basis independence makes the count independent of the homogeneous basis used to compute it.
The equality is an identity of integer polynomials. Applying a ring homomorphism to the coefficients and evaluating at a parameter preserves it. In particular it gives equality of the scaled actual-stalk character and scaled KL polynomial at a nonzero square parameter.
At the top vertex, both sides equal \(1\): the boundary top is \(A\) with generator degree zero, and the canonical family has diagonal polynomial \(P_{b,b}=1\).
Independence of the chosen boundary model
Two normalized boundary sheaves on the same interval are graded-isomorphic. Their stalk isomorphisms preserve the generator degrees, so they have the same character. Consequently the condition can be used with any normalized boundary model on \([u,b]\).
The upper-interval kernel theorem also identifies \(C_x^{[u,b]}\) with the character of the original \(B(b)^x\) on \([1,b]\). This is the identification behind the shorter notation \(C_{x,b}\).
Use in endpoint formulas
The KL endpoint substitution theorem takes this equality at a vertex as an explicit hypothesis. It replaces \(C_{x,b}(t^2)\) by \(P_{x,b}(t^2)\) in the empty-condition state, and \(C_{x,b}(t^{-2})\) by \(P_{x,b}(t^{-2})\) in the full-kernel state.
The actual-character reciprocity theorem supplies the identity for \(C\) used in the simultaneous comparison proof. The canonical family \(P\) enters the later polynomial-rigidity argument through its own recurrence and degree bound.
Relation to the OpenAI notes
Different intermediate statement.
OpenAI notes: The grading-normalization proposition identifies the stalk graded rank with the canonical KL polynomial and uses the Elias–Williamson character theorem through the cited moment-graph bridges. bg:normalization.
Lean (primary): The development defines the equality of actual stalk characters and canonical KL polynomials as StalkCharacterObligation. The unconditional theorem used in the main route is R-reciprocity for the actual character. The local strict half-degree implication retains an explicit nonzero-diagonal hypothesis. SheafMorphisms:219, DyerPaths:62, PositiveTransfer:186, CharacterReciprocity:1807, TerminalKernels:1273, TerminalKernels:1284.
Role in the proof: The comparison uses the actual character C and its proved R-reciprocity. Canonical P is handled by the later polynomial-rigidity induction. The local degree result applies to a diagonal coefficient satisfying its stated nonvanishing hypothesis.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.BruhatGraph.StalkCharacterObligation: SheafMorphisms.lean:219.