The scaled actual-stalk character vector
Fix a finite-generator Coxeter system, the real polynomial coefficient algebra of its extended realization, and \(b\in W\). The actual character \(C_{x,b}(q)\in\mathbb Z[q]\) counts homogeneous generators of the chosen boundary stalk on \([1,b]\).
For \(x\in[1,b]\) and a nonzero parameter \(t\) in a field \(F\), define \[ c_x^b(t) =t^{\ell(x)-\ell(b)}C_{x,b}(t^2) =t^{-d(x,b)}C_{x,b}(t^2), \qquad d(x,b)=\ell(b)-\ell(x). \] The integer coefficients are mapped into \(F\) by its canonical ring homomorphism from \(\mathbb Z\). The sheaf still supplies the same integer polynomial; \(F\) is the field in which that polynomial is evaluated.
This is the nonzero-parameter expression of the source's scaledStalk. As \(x\) varies, it gives a vector \(c^b(t)\) indexed by \([1,b]\).
Reciprocal evaluation
Replacing \(t\) by \(t^{-1}\) gives \[ c_x^b(t^{-1}) =t^{d(x,b)}C_{x,b}(t^{-2}). \] These are finite polynomial evaluations multiplied by integer powers of the nonzero field parameter.
The ordinary polynomial parameter is \(q=t^2\). The normalized R-family uses \(t-t^{-1}\) in the inverse-reciprocity form and \(t^{-1}-t\) after the parameter is inverted.
Real Hilbert normalization
In the comparison range \(0<t<1\), if the coefficient algebra has \(N\) degree-one variables, multiplication by \[ (1-t^2)^Nt^{-d(x,b)} \] removes the polynomial coefficient Hilbert factor from a stalk Hilbert sum at \(t^2\). The result is \(c_x^b(t)\).
For a full outgoing kernel, the reflected character gives \(c_x^b(t^{-1})\). The normalized-endpoint theorem proves both identities directly.
For the one-edge interval \([1,s]\), both stalk characters are \(1\), so \(c_1^s(t)=t^{-1}\) and \(c_s^s(t)=1\).
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.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.BruhatGraph.scaledStalk: EdgeLocalization.lean:377.