Theorem Scaled reciprocity and its restriction to an upper interval
✓ theorem proved

Scaled reciprocity and its restriction to an upper interval

Fix the actual boundary character on \([1,b]\) in a finite-generator Coxeter system. Write \(c_x^b(t)=t^{\ell(x)-\ell(b)}C_{x,b}(t^2)\) for its scaled actual-stalk character, and let \(\widetilde R_{x,z}\) be the normalized R-polynomial family. Integer coefficients are evaluated in the indicated field.

For \(x\in[1,b]\) and a nonzero parameter \(v\) in any field, \[ c_x^b(v^{-1}) =\sum_{z\in[1,b]}\widetilde R_{x,z}(v-v^{-1})\,c_z^b(v). \] Equivalently, for \(t\ne0\), \[ c_x^b(t) =\sum_{z\in[1,b]}\widetilde R_{x,z}(t^{-1}-t)\,c_z^b(t^{-1}). \] For \(x\in[a,b]\), the second sum can be restricted to \(z\in[a,b]\), using the same characters from the boundary sheaf on \([1,b]\).

Relation to the OpenAI notes

Equivalent parameter conventions.

OpenAI notes: The comparison uses \(0<q<1\), scale \(q^{-d/2}(1-q)^N\), and transfer coefficient \(q^{-1/2}-q^{1/2}\). eq:state, eq:step.

Lean (primary): The comparison uses \(0<t<1\) and substitutes \(q=t^2\). Its scale is \(t^{-d}(1-t^2)^N\) and its transfer coefficient is \(t^{-1}-t\). Dihedral:730, PositiveTransfer:591, PositiveTransfer:621, EdgeLocalization:377, EdgeLocalization:423, PositiveTransfer:535, PositiveTransfer:603.

Role in the proof: The substitution \(q=t^2\) identifies the two normalizations. Linear root labels have degree one throughout the calculation.

Proof

Step 1. Evaluate actual-character reciprocity at the square parameter

Put \(d_x=\ell(b)-\ell(x)\). Evaluate the integer polynomial reciprocity identity at \(q=v^2\). The degree bound permits reciprocal evaluation of reflection, giving \[ (v^2)^{d_x}C_{x,b}(v^{-2}) =\sum_{z\in[1,b]}R_{x,z}(v^2)C_{z,b}(v^2). \]

Step 2. Normalize the left side

Multiply by \(v^{-d_x}\). Since \(v\ne0\), integer powers obey the exponent laws: \[ v^{-d_x}(v^2)^{d_x}=v^{d_x}=(v^{-1})^{-d_x}. \] The left side is therefore \(c_x^b(v^{-1})\).

Step 3. Normalize each R-polynomial term

The square-parameter normalization gives \[ R_{x,z}(v^2) =v^{\ell(z)-\ell(x)} \widetilde R_{x,z}(v-v^{-1}). \] For \(d_z=\ell(b)-\ell(z)\), the scalar exponents satisfy \[ -d_x+\ell(z)-\ell(x) =\ell(z)-\ell(b)=-d_z. \] Thus each right-hand term becomes \[ \widetilde R_{x,z}(v-v^{-1}) \,v^{-d_z}C_{z,b}(v^2) =\widetilde R_{x,z}(v-v^{-1})c_z^b(v). \] This proves the first identity.

Step 4. Invert the parameter

Apply the first identity with \(v=t^{-1}\). Then \(v^{-1}=t\), and the normalized R argument becomes \(t^{-1}-t\). This gives the second identity.

Step 5. Restrict the sum to an upper interval

Let \(a\le x\le b\). If a vertex \(z\le b\) does not satisfy \(a\le z\), then \(x\le z\) is impossible: it would imply \(a\le x\le z\).

The support of \(\widetilde R\) therefore makes its coefficient at this \(z\) zero. Remove all such terms. The remaining vertices are exactly \([a,b]\), with the same underlying group elements and the same actual-stalk values \(c_z^b(t^{-1})\).

This is the source's upper-interval formula. The character vector is the restriction of the original \([1,b]\) vector, which is the vector used in the later schedule comparison.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.BruhatGraph.scaledStalk_reciprocity_inverse: EdgeLocalization.lean:381.

OAI.KLInvariance.BruhatGraph.scaledStalk_reciprocity: EdgeLocalization.lean:423.

OAI.KLInvariance.BruhatGraph.scaledStalk_reciprocity_upper: EdgeLocalization.lean:430.

Used by