Theorem KL polynomials in a dihedral Coxeter system
✓ theorem proved

KL polynomials in a dihedral Coxeter system

For a finite or infinite Coxeter system with two simple generators, \[ P_{x,y}(q)=1\qquad(x\le y). \]

Relation to the OpenAI notes

Same conclusion, different final identification.

OpenAI notes: After the lower-interval Hecke sum is shown bar invariant, the proof identifies it by the triangular characterization of the KL basis. loc:dihedral-kl.

Lean (primary): The same bar calculation is used to derive the sum identity for R-polynomials. Lean then verifies that the constant-on-comparable-pairs P-family satisfies NormalizedKL and applies the proved uniqueness theorem. Dihedral:557, Dihedral:605, Dihedral:615, Dihedral:656.

Role in the proof: The final steps use the R-sum identity and normalized-family uniqueness to obtain P=1.

Proof

Step 1. Define the candidate P-family

Set \[ Q_{x,y}= \begin{cases} 1,&x\le y,\\ 0,&x\not\le y. \end{cases} \] Pair it with the constructed ordinary R-family.

The ordinary R construction already supplies the first three normalization fields: diagonal one, zero outside Bruhat order, and the recurrence for every left descent.

Step 2. Check the diagonal, support and degree fields for Q

Reflexivity gives \(Q_{x,x}=1\). Its value is zero when \(x\not\le y\), by definition.

If \(x<y\), then \(Q_{x,y}=1\), whose natural degree is zero. Strict Bruhat length increase gives \(d(x,y)>0\), so \[ 2\,\operatorname{natDegree}(Q_{x,y})=0<d(x,y). \] Thus all three P value/degree fields hold.

Step 3. Verify reciprocity by the dihedral R-sum

For \(x\le y\), reflection gives \[ \operatorname{reflect}_{d(x,y)}Q_{x,y} =\operatorname{reflect}_{d(x,y)}1 =q^{d(x,y)}. \] Every \(z\in[x,y]\) satisfies \(z\le y\), so \(Q_{z,y}=1\). Therefore the required interval sum is \[ \sum_{z\in[x,y]}R_{x,z}Q_{z,y} =\sum_{z\in[x,y]}R_{x,z} =q^{d(x,y)}, \] by the dihedral R-sum theorem.

The finite interval sum is the same sum used by the normalization interface. This verifies its seventh field.

Step 4. Identify the candidate by uniqueness

Steps 1–3 prove that \((R,Q)\) is a normalized pair. The canonical-family theorem proves that the selected pair is normalized as well.

The normalized-pair uniqueness theorem identifies the two pairs. Their P-components agree, so for \(x\le y\), \[ P_{x,y}=Q_{x,y}=1. \]

The preceding R-sum proof obtains its identity from the lower-interval Hecke sums and their bar eigenvalue. The OpenAI notes finish the bar calculation through the triangular KL-basis characterization; the Lean proof finishes through the seven polynomial normalization fields and uniqueness.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.Dihedral.constantP: Dihedral.lean:612.

OAI.KLInvariance.Dihedral.constant_normalizedKL: Dihedral.lean:615.

OAI.KLInvariance.Dihedral.klPolynomial_eq_one: Dihedral.lean:656.