Theorem Zero, top and empty-word character values
✓ theorem proved

Zero, top and empty-word character values

Fix a finite-generator Coxeter system and a finite basis of its extended realization. Let \(N\) be the size of this basis.

If a word stalk \(M_{\mathbf t}^x\) is zero, then \(C_x^{\mathbf t}(q)=0\). If \(\mathbf t\) is reduced with product \(b\), then \[ C_b^{\mathbf t}(q)=1. \] In particular, the empty-word values are \[ C_x^\varnothing(q)= \begin{cases} 1,&x=1,\\ 0,&x\ne1. \end{cases} \] For a reduced word with product \(b\), weights \(x\not\le b\) have zero character.

Proof

Step 1. Compute the character of a zero stalk

Every homogeneous piece of a zero stalk has dimension zero, so its Hilbert sum is zero. The word-stalk Hilbert formula expresses that same sum as \(C_x^{\mathbf t}(t)(1-t)^{-N}\) for every \(0<t<1\).

The zero polynomial also represents it with this common factor. Since \((1-t)^{-N}\ne0\), uniqueness of the polynomial Hilbert numerator gives \(C_x^{\mathbf t}=0\).

Step 2. Identify the top stalk with the polynomial coefficient module

For a reduced word, the top-stalk theorem gives a degree-preserving real-linear isomorphism \[ M_{\mathbf t}^{b}\xrightarrow{\sim}A. \] Composing with the coefficient isomorphism \(A\to P=\mathbb R[X_i]\) gives bottTopPolynomialEquiv.

It induces isomorphisms of the homogeneous pieces, so \[ \dim_{\mathbb R}(M_{\mathbf t}^{b})_n=\dim_{\mathbb R}P_n \] for every integer \(n\).

Step 3. Recover the top numerator

The coefficient Hilbert series has sum \((1-t)^{-N}\). Thus the top stalk's dimension sequence has numerator \(1\).

The word-stalk Hilbert formula gives numerator \(C_b^{\mathbf t}\) for the same sequence and factor. Polynomial numerator uniqueness therefore gives \(C_b^{\mathbf t}=1\).

Step 4. Specialize to the empty word and to absent weights

The empty word is reduced, with product \(1\), so Step 3 gives its character \(1\) at weight \(1\). Every other empty-word stalk is zero, giving the other values by Step 1.

More generally, reduced-word support gives a zero stalk at each \(x\not\le b\), so Step 1 proves the final assertion.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.TitsSpace.bottStalkCharacter_zero: IntegralPairings.lean:237.

OAI.KLInvariance.TitsSpace.bottTopPolynomialEquiv: IntegralPairings.lean:248.

OAI.KLInvariance.TitsSpace.bottTopPolynomialEquiv_homogeneous: IntegralPairings.lean:254.

OAI.KLInvariance.TitsSpace.bottStalkCharacter_top: IntegralPairings.lean:260.

OAI.KLInvariance.TitsSpace.bottStalkCharacter_nil: IntegralPairings.lean:279.

Used by