Definition Heights, normalized positive roots, and geometric slopes
✓ definition formalized

Heights, normalized positive roots, and geometric slopes

In the original real root space \(V_0=\mathbb R^I\), define the height functional by \[ h(v)=\sum_{i\in I}v_i. \] A positive root \(a\) is nonzero with nonnegative coordinates, so at least one coordinate is positive and \(h(a)>0\).

Its normalized root is \[ \widehat a=\frac{a}{h(a)}. \] It lies on the affine hyperplane \(h=1\). For a linear functional \(f:V_0\to\mathbb R\), define the slope \[ s_f(a)=f(\widehat a)=\frac{f(a)}{h(a)}. \]

Injectivity of normalization

If \(\widehat a=\widehat b\), then \(a=(h(a)/h(b))b\). The scalar is positive. Both roots have squared Tits norm \(2\), so its square is one and it equals one. Hence \(a=b\).

Thus normalization keeps the positive roots distinct, even when the Tits form is degenerate.

Generic slope functions on a root plane

A functional is generic on the positive roots when its slope function is injective. Comparing its slopes then gives a total order.

For a root plane \(P\), normalized roots lie on the affine line \(P\cap\{h=1\}\). A generic slope function distinguishes their normalized points. The generic-functional theorem proves existence, and the geometric-reflection-order entry describes the fixed choice used in Lean.

An A2 example

Let \(a,b\) be the two simple roots, so \(h(a)=h(b)=1\) and \(h(a+b)=2\). Choose \(f(a)=0,f(b)=1\). Then \[ s(a)=0,\qquad s(a+b)=\frac12,\qquad s(b)=1. \] The normalized middle root is \((a+b)/2\), the midpoint of the two normalized simple roots.

References

The local source declarations below give the definitions and their properties.

Lean formalization

OAI.KLInvariance.TitsSpace.height: ReflectionSchedules.lean:86.

OAI.KLInvariance.TitsSpace.PositiveRoot.height_pos: ReflectionSchedules.lean:93.

OAI.KLInvariance.TitsSpace.normalizedRoot: ReflectionSchedules.lean:103.

OAI.KLInvariance.TitsSpace.normalizedRoot_injective: ReflectionSchedules.lean:105.

Used by