The native dihedral later-count difference
Fix a plane basis \(B\) in a finite-generator real Coxeter system and let \(D\) be its induced dihedral group. For an intrinsic edge \(e:x\to_Dy\), write \(\alpha_e\) for its actual plane edge root. Given an upper bound \(y\le_Dc\), the native later-edge counts satisfy \[ h_D(x,\alpha_e;c) = h_D(y,\alpha_e;c) +\frac{\ell_D(y)-\ell_D(x)-1}{2}. \] All counts and the correction are natural numbers. The length gap is a positive odd integer, so the quotient is an integer.
The same counts may be taken over a relative interval: for any \(a\le_Dx\), any \(c\in D\), and any positive ambient threshold root \(\alpha\), \[ \sum_{z\in[a,c]_D}\varepsilon_D(x,\alpha,z) =h_D(x,\alpha;c). \] Consequently the first identity holds when both sums are restricted to \([a,c]_D\), provided \(a\le_Dx\).
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. Choose reduced admissible representatives
The period-and-representative theorem supplies admissible signed indices \(i,j\) with \[ w_i=x,\quad w_j=y,\qquad |i|=\ell_D(x),\quad |j|=\ell_D(y). \] Choose an admissible index \(k\) for \(c\) as follows.
If \(c=y\), take \(k=j\). Otherwise \(y\le_Dc\) and \(y\ne c\) imply \(\ell_D(y)<\ell_D(c)\). Choose any admissible reduced representative \(w_k=c\); it then satisfies \(|j|<|k|\).
Thus the exact top condition is \[ k=j\quad\text{or}\quad |j|<|k|. \] The signed edge criterion also gives \(|i|<|j|\) and odd \(i+j\).
Step 2. Expand each lower-interval sum by ranks
Set \(C=|k|\). The lower-interval signed-sum theorem says that for a function \(f\) with \(f(1)=0\), \[ \sum_{z\le_Dw_k}f(z) =f(w_k)+\sum_{n=0}^{C-1}\bigl(f(w_n)+f(w_{-n})\bigr). \] The top appears once. Positive ranks below \(C\) have two vertices. The two appearances of \(n=0\) contribute zero.
Apply this first to \(f(z)=\varepsilon_D(x,\alpha_e,z)\), and then to \(f(z)=\varepsilon_D(y,\alpha_e,z)\). The zero-at-identity assertion of the angular-indicator theorem verifies the required hypothesis in each case.
Step 3. Check every index used in the indicator translation
The chosen \(j,k\) are admissible and reduced, so the angular-indicator theorem applies to the top terms.
For an internal rank \(0\le n<C=\ell_D(c)\), the signed-length theorem, using \(c\) as a witness, gives \[ \ell_D(w_n)=n,\qquad \ell_D(w_{-n})=n. \] Both \(n\) and \(-n\) are admissible. Indeed, for any maximum length \(m\), one has \(n<C\le m\), hence \(-m<-n\le0\le n<m\). When lengths are unbounded the admissibility condition has no maximum-length element to test and holds for all indices.
All hypotheses of the same-source and next-source indicator formulas therefore hold at every term.
Step 4. Identify the two numerical counts
Put \(q=i+j\). The actual angular-indicator theorem converts Step 2 into \[ \begin{aligned} h_D(x,\alpha_e;c) &=\operatorname{aft}(i,q,k) +\sum_{n=0}^{C-1}\bigl(\operatorname{aft}(i,q,n)+\operatorname{aft}(i,q,-n)\bigr) =\operatorname{cnt}(i,q,k),\\ h_D(y,\alpha_e;c) &=\operatorname{aft}(j,q,k) +\sum_{n=0}^{C-1}\bigl(\operatorname{aft}(j,q,n)+\operatorname{aft}(j,q,-n)\bigr) =\operatorname{cnt}(j,q,k). \end{aligned} \] The final equalities are the definition of the finite signed count, including its singleton top.
These translations also establish the separate source lemmas laterCount_same and laterCount_next. The first requires admissibility of \(j,k\); the second requires only that of \(k\), in addition to the current edge and the three reduced-length identities.
Step 5. Substitute the signed-count identity
The signed-count difference theorem requires a positive odd gap and the top condition from Step 1. Positivity follows from \(|i|<|j|\). Since \(|r|\equiv r\pmod2\) for every integer \(r\), odd \(i+j\) makes \(|j|-|i|\) odd.
That theorem gives \[ \operatorname{cnt}(i,i+j,k) =\operatorname{cnt}(j,i+j,k)+\frac{|j|-|i|-1}{2}. \] Substitute Step 4 and the lengths from Step 1 to obtain the stated native count difference.
Step 6. Restrict a count to a relative interval
Let \(a\le_Dx\) and let \(\alpha\) be any positive ambient root. The set \([a,c]_D\) consists of those vertices of the lower interval of \(c\) which also satisfy \(a\le_Dz\).
For a removed vertex \(z\), suppose the indicator were nonzero. Its definition would give an upward edge \(x\to_Dz\), hence \(x\le_Dz\). Transitivity with \(a\le_Dx\) would give \(a\le_Dz\), contrary to the removal condition. Thus every removed term is zero.
The finite sums over the lower interval and over \([a,c]_D\) are therefore equal. For the count based at \(y\), use \(a\le_Dx\le_Dy\). Applying this restriction to both sides of the already proved difference completes the relative-interval form.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.TitsSpace.PlaneBasis.laterCount_same: DihedralIntervals.lean:1923.
OAI.KLInvariance.TitsSpace.PlaneBasis.laterCount_next: DihedralIntervals.lean:1948.
OAI.KLInvariance.TitsSpace.PlaneBasis.laterCount_difference: DihedralIntervals.lean:1974.
OAI.KLInvariance.TitsSpace.PlaneBasis.afterIndicator_eq_zero: DihedralIntervals.lean:2022.
OAI.KLInvariance.TitsSpace.PlaneBasis.interval_laterCount: DihedralIntervals.lean:2030.