Theorem Perfect graded pairings on reduced-word sections
✓ theorem proved

Perfect graded pairings on reduced-word sections

Fix a finite-generator Coxeter system and a reduced word \(\mathbf t\) of length \(m\), with product \(b\). Choose a finite real basis of the extended realization, giving the symmetric coefficient algebra \(A\) its degree-one grading. Let \(S\) be the graded word sheaf on \([1,b]\).

The section isomorphism \(\Psi:M_{\mathbf t}\to\Gamma(S)\) transports the word pairing to a perfect symmetric pairing \[ \langle s,z\rangle_\Gamma =\langle\Psi^{-1}s,\Psi^{-1}z\rangle_{M_{\mathbf t}}. \] It has degree shift \(m\): \[ \langle\Gamma(S)_i,\Gamma(S)_j\rangle_\Gamma\subseteq A_{i+j-m}. \] If one section is supported at a vertex \(x\) and the other has zero value at \(x\), their pairing is zero, in either argument order.

After choosing polynomial coordinates for \(A\), these properties give the source's bottPolynomialPerfectPairing.

Proof

Step 1. Transport perfection and symmetry through the section isomorphism

Let \(\lambda:\Gamma(S)\to A\) be linear. The composite \(\lambda\Psi\) is a functional on the word module. Its perfect pairing supplies a unique vector \(v\) with \[ \langle v,w\rangle_{M_{\mathbf t}}=\lambda(\Psi w) \] for every \(w\). Then \(\Psi v\) represents \(\lambda\) for the displayed section pairing.

This also proves uniqueness, since \(\Psi\) is bijective. Symmetry follows from symmetry of the word pairing by substituting the two inverse images. The evaluation formula is exactly \[ \langle\Psi v,\Psi w\rangle_\Gamma =\langle v,w\rangle_{M_{\mathbf t}}. \]

Step 2. Detect the word degree from all its stalks

A homogeneous word vector of degree \(n\) projects to degree \(n\) in every stalk. Conversely, suppose all stalk projections of \(v\) have degree \(n\). Component transport gives \[ \pi_x(\operatorname{pr}_n v) =\operatorname{pr}_n(\pi_xv)=\pi_xv \] for every interval vertex \(x\).

The full vertex projection is injective for a reduced word, so \(\operatorname{pr}_n v=v\). Thus \(v\) has degree \(n\). This proves the source's bottPiece_iff_stalks criterion and shows that \(\Psi^{-1}\) preserves degree.

Step 3. Compute the pairing degree

For \(s\in\Gamma(S)_i\), \(z\in\Gamma(S)_j\), Step 2 puts their inverse images in word degrees \(i,j\). The word-pairing degree theorem gives \[ \langle\Psi^{-1}s,\Psi^{-1}z\rangle\in A_{i+j-m}. \] This is the required homogeneity.

Step 4. Prove the two support-orthogonality equations

Suppose \(s\) is supported at \(x\), and \(z_x=0\). Write \(s=\Psi v\), \(z=\Psi w\).

All subexpression weights of the reduced word lie in \([1,b]\). Thus the support of \(s\) says precisely that \(v\) has zero evaluations at every weight different from \(x\); and \(z_x=0\) says \(\pi_x(w)=0\).

Word support duality identifies vectors supported at \(x\) with the annihilator of \(\ker\pi_x\). Hence \(\langle v,w\rangle=0\). Step 1 gives \(\langle s,z\rangle_\Gamma=0\).

If \(s_x=0\) and \(z\) is supported at \(x\), symmetry exchanges the arguments and gives the other equation.

Step 5. Package and transport the pairing

Perfection, the degree identity and the two orthogonality equations are the fields of bottWordPerfectPairing. A graded coefficient isomorphism \(A\to P\) transports the section modules and applies that isomorphism to pairing values.

The pairing-transport theorem preserves perfection, the same degree shift, symmetry and support orthogonality. This gives bottPolynomialPerfectPairing on the polynomial word sheaf, with shift \(m\).

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.TitsSpace.bottSectionDuality: BalancedPairings.lean:346.

OAI.KLInvariance.TitsSpace.bottSectionDuality_apply: BalancedPairings.lean:354.

OAI.KLInvariance.TitsSpace.bottSection_symmetric: BalancedPairings.lean:397.

OAI.KLInvariance.TitsSpace.bottPiece_iff_stalks: BalancedPairings.lean:422.

OAI.KLInvariance.TitsSpace.bottSectionDuality_homogeneous: BalancedPairings.lean:443.

OAI.KLInvariance.TitsSpace.bottSection_left_orthogonal: CharacterReciprocity.lean:866.

OAI.KLInvariance.TitsSpace.bottSection_right_orthogonal: CharacterReciprocity.lean:876.

OAI.KLInvariance.TitsSpace.bottWordPerfectPairing: CharacterReciprocity.lean:884.

OAI.KLInvariance.TitsSpace.bottPolynomialPerfectPairing: CharacterReciprocity.lean:892.

Used by