Every positive root $\alpha \in \Phi^+$ has unit length for the Coxeter form: $B(\alpha,\alpha) = 1$. This follows from the fact that simple roots have unit length and the form is $W$-invariant.
Hypotheses isolating the two key properties of an affine, crystallographic Coxeter system used in the proofs: (1) coercivity of the form $B$ on every proper parabolic subspace, and (2) integrality of the coordinates of every positive root in the simple-root basis.
- roots_integer_coords (α : B → ℝ) : α ∈ standardΦpos M → ∀ (b : B), ∃ (n : ℤ), α b = ↑n
Instances For
Under the affine Coxeter hypothesis, the set of positive roots supported on any proper subset $I \subsetneq B$ of simple reflections is finite. This is the "spherical-on-parabolics" finiteness statement and is what makes proper parabolic subgroups finite.
Shifting by a radical vector does not change the Coxeter form: $B(w - a v_0, w - a v_0) = B(w,w)$ when $v_0 \in \mathrm{rad}(B)$.
The Coxeter form $B$ is coercive on the hyperplane $v_0^{\perp}$: there exists $c > 0$ with $c \|w\|^2 \le B(w,w)$ for all $w$ with $\langle v_0, w \rangle = 0$. This is the key spectral estimate driving finiteness of nonpositive-roots sets.
Pythagoras decomposition: $\|c v_0 + w\|^2 = c^2 \|v_0\|^2 + \|w\|^2$ when $w \perp v_0$.
Uniform coordinate bound: every root $\alpha \in \Phi^+$ with $\langle \alpha, x\rangle \le 0$ has all coordinates in $[0, C]$ for a constant $C$ depending only on $v_0$, $x$, and the coercivity constant.
Combining the coordinate bound with integrality: for $x$ on the affine hyperplane, the set of positive roots with $\langle \alpha, x\rangle \le 0$ is finite.
Packaging: from AffineCoxeterHyp we obtain the two finiteness axioms needed by the rest of
the development (subspan finiteness and hyperplane finiteness of nonposRoots).