Plane weights measure the change of the plane inversion count
Let \(a,b\) be distinct positive roots, let \(P=P(a,b)\), and suppose \(x^{-1}a\) is positive. Put \(y=r_a x\). Then \[ L_P(y)-L_P(x)=1+2k_P(x,a), \] where \(k_P(x,a)\) is the geometric plane weight.
Proof
Step 1. Express the count difference by inversion indicators
By definition, \[ L_P(y)-L_P(x) =\sum_{c\in\Phi^+\cap P}\bigl(n_y(c)-n_x(c)\bigr). \] Only finitely many summands are nonzero, because both inversion sets are finite.
The hypothesis gives \(n_x(a)=0\). Since \(y^{-1}a=x^{-1}r_a(a)=-x^{-1}a\), we have \(n_y(a)=1\). Thus the distinguished root contributes one.
Step 2. Identify the remaining terms with the plane defect
For every positive \(c\ne a\), the root-span criterion gives \[ c\in P\quad\Longleftrightarrow\quad P(a,c)=P. \] Therefore, after removing the distinguished root, the remaining sum is precisely \[ D_P(x,y;a) =\sum_{\substack{c\ne a\\ P(a,c)=P}}\bigl(n_y(c)-n_x(c)\bigr). \] Hence \(L_P(y)-L_P(x)-1=D_P(x,y;a)\).
Step 3. Substitute the partner-pair count
The plane-weight theorem applies to the ascent root \(a\). It groups the remaining inversion changes into partner pairs inside \(P\) and proves \[ D_P(x,y;a)=2k_P(x,a). \] Substitution gives the required equality.
References
The local source declarations below define the objects and prove the stated properties.
Lean formalization
OAI.KLInvariance.TitsSpace.rootPlane_grouped_erase: ReflectionSchedules.lean:708.
OAI.KLInvariance.TitsSpace.planeDefect_eq_relative_difference: ReflectionSchedules.lean:733.
OAI.KLInvariance.TitsSpace.planeWeight_relative_degree: ReflectionSchedules.lean:742.