Actual root comparisons give the signed indicator tables
Fix a plane basis \(B\) in a finite-generator real Coxeter system, its induced dihedral group \(D\), and its actual geometric slope \(\sigma\). Let \(i,j,z\in\mathbb Z\) have their correct signed lengths: \[ \ell_D(w_i)=|i|,\qquad \ell_D(w_j)=|j|,\qquad \ell_D(w_z)=|z|. \] Suppose \(e:w_i\to_Dw_j\) is an upward reflection edge, with actual plane edge root \(\alpha_e\).
The two slope comparisons have the following conditions.
| Configuration | Admissibility assumptions | Equivalent integer comparison |
|---|---|---|
| Another edge \(w_i\to_Dw_z\) | \(j,z\) are admissible signed indices | \(\sigma(\alpha_e)<\sigma(\alpha_{w_i\to w_z})\ \Longleftrightarrow\ \operatorname{Before}(i+j,i+z)\) |
| A next edge \(w_j\to_Dw_z\) | \(z\) is an admissible signed index | \(\sigma(\alpha_e)<\sigma(\alpha_{w_j\to w_z})\ \Longleftrightarrow\ \operatorname{Before}(i+j,j+z)\) |
For arbitrary \(z\) with the displayed length and admissibility assumptions, these comparisons extend to indicators: \[ \begin{aligned} \varepsilon_D(w_i,\alpha_e,w_z)&=\operatorname{aft}(i,i+j,z) &&\text{if }j,z\text{ are admissible},\\ \varepsilon_D(w_j,\alpha_e,w_z)&=\operatorname{aft}(j,i+j,z) &&\text{if }z\text{ is admissible}. \end{aligned} \] Here \(\operatorname{aft}(p,a,z)\) is the numerical arrow indicator: it is one when \(|p|<|z|\), \(p+z\) is odd, and \(\operatorname{Before}(a,p+z)\); otherwise it is zero.
Relation to the OpenAI notes
From actual root slopes to nonnegative endpoint counts.
OpenAI notes: The later-edge lemma first transports a relative interval, then uses the midpoint-order table to calculate the difference of two outgoing counts. Its proof also treats reversal of the angular order by subtracting earlier and later edges from total outgoing counts. dih:midpoint-order, dih:count.
Lean (primary): The native count is first defined over the full lower ideal in the induced dihedral group. Actual root comparisons are translated into signed indicators with explicit reduced-length and GoodIndex hypotheses. The two nonnegative counts are evaluated separately. interval_laterCount then removes zero terms to restrict either count to a relative interval. DihedralIntervals:1852, DihedralIntervals:1814, DihedralIntervals:1833, DihedralIntervals:1919, DihedralIntervals:1923, DihedralIntervals:1948, DihedralIntervals:1974, DihedralIntervals:2030.
Role in the proof: The Blueprint gives the native count independently of interval transport, with the exact hypotheses of both indicator tables. Its slope orientation is the increasing orientation of the chosen PlaneBasis.
Proof
Step 1. Extract the edge parameters and their bounds
The signed edge criterion turns the current edge into \[ |i|<|j|,\qquad i+j\text{ odd}. \] Its left reflection is \(w_{i+j}\). A second edge from \(i\) gives \(|i|<|z|\), odd \(i+z\), and reflection \(w_{i+z}\). A next edge from \(j\) gives \(|j|<|z|\), odd \(j+z\), and reflection \(w_{j+z}\).
Step 2. Apply the angular theorem when lengths are unbounded
If lengths are unbounded, the angular-order theorem identifies slope comparison of any two actual edge roots with Before on their odd parameters. Substituting the labels from Step 1 proves both rows directly.
Step 3. Apply the constrained residue formulas at maximum length
If lengths are bounded, their nonempty set of natural values has a maximum, attained by an element of \(D\); denote it by \(m\). The angular-order theorem expresses each comparison using odd residues modulo \(2m\).
For the first row, admissibility supplies \[ -m<j\le m,\qquad -m<z\le m. \] Together with \(|i|<|j|\) and \(|i|<|z|\), these are exactly the hypotheses of the same-source finite-residue comparison theorem. It replaces \[ (i+j)\bmod2m<(i+z)\bmod2m \] by \(\operatorname{Before}(i+j,i+z)\).
For the second row, admissibility of \(z\) gives \(-m<z\le m\), while the two edges give \(|i|<|j|<|z|\). The next-source finite-residue comparison theorem replaces \[ (i+j)\bmod2m<(j+z)\bmod2m \] by \(\operatorname{Before}(i+j,j+z)\). These are precisely its required bounds. This proves both slope rows in the bounded case.
Step 4. Include vertices which are not edge targets
For the first indicator, split according to whether \(w_i\to_Dw_z\).
If it is an edge, the length and parity tests in \(\operatorname{aft}\) hold. Both sides are consequently one or zero according to the first slope row, so they agree.
If it is not an edge, the actual indicator is zero. The signed edge criterion says that the conjunction \(|i|<|z|\) and odd \(i+z\) is false. Thus the numerical indicator is also zero, independently of the Before test.
The identical split with source \(w_j\), using the second row, proves the next-source identity. Finally an edge to the identity would give \(\ell_D(x)<\ell_D(1)=0\), so every actual after-indicator at target \(1\) is zero, as used in the lower-interval sum.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.TitsSpace.PlaneBasis.edgeRoot_same_table: DihedralIntervals.lean:1814.
OAI.KLInvariance.TitsSpace.PlaneBasis.edgeRoot_next_table: DihedralIntervals.lean:1833.
OAI.KLInvariance.TitsSpace.PlaneBasis.afterIndicator_one: DihedralIntervals.lean:1858.
OAI.KLInvariance.TitsSpace.PlaneBasis.afterIndicator_same: DihedralIntervals.lean:1867.
OAI.KLInvariance.TitsSpace.PlaneBasis.afterIndicator_next: DihedralIntervals.lean:1885.