Definition The localized edge ring and its reduction map
✓ definition formalized

The localized edge ring and its reduction map

Fix a finite-generator Coxeter system, its original root space \(V_0\), and a polynomial coefficient algebra \(A\) from a finite real basis of the extended realization. Let \[ \lambda:V_0\to A,\qquad \eta_P:A\to R_P=A_{\mathfrak p_P},\qquad \lambda_P=\eta_P\lambda \] be the linear label, localization, and localized label maps for a subspace \(P\subseteq V_0\).

For any \(a\in V_0\), define the localized edge ideal \[ J_{P,a}=(\lambda_P(a))\subseteq R_P \] and the localized edge ring \[ E_{P,a}=R_P/J_{P,a}. \] These are localizedEdgeIdeal and LocalizedEdgeRing. Write \[ q_{P,a}:R_P\to E_{P,a} \] for the quotient map.

Reduction of root-space vectors

The localized reduction map is the real-linear map \[ r_{P,a}:V_0\to E_{P,a},\qquad r_{P,a}(v)=q_{P,a}(\eta_P(\lambda(v))). \] This is localizedReduction. Its value at \(a\) is zero, because \(\lambda_P(a)\) generates the quotient ideal: \[ r_{P,a}(a)=0. \]

The construction specifies \(P\) and \(a\) independently. The later edge calculation takes \(a\) to be the current edge root and \(P\) to be a root plane containing it.

The map from the original edge quotient

Let \(E_a=A/(\lambda(a))\), with quotient map \(q_a\). Since \(q_{P,a}\eta_P\) kills \(\lambda(a)\), it factors through \[ \overline\eta_{P,a}:E_a\to E_{P,a}, \qquad [f]\longmapsto q_{P,a}(\eta_P(f)). \] Changing \(f\) by \(\lambda(a)g\) changes its image by \(q_{P,a}(\lambda_P(a)\eta_P(g))=0\), so this map is well defined. It satisfies \[ \overline\eta_{P,a}\,q_a=q_{P,a}\eta_P. \]

The tensor-quotient theorem identifies its scalar extension: \[ R_P\otimes_A E_a\xrightarrow{\sim}E_{P,a}, \qquad s\otimes[f]\longmapsto q_{P,a}(s\eta_P(f)). \] Thus a root label has the same image under either route through these coefficient maps.

A concrete reduction

In the illustrative ring \(A=\mathbb R[\alpha,\beta,\gamma]\), take \(\mathfrak p_P=(\alpha,\beta)\) and kill the root label \(\alpha\). Then \[ E_{P,\alpha} =A_{(\alpha,\beta)}/(\alpha/1). \] The class of \(\gamma/1\) is a unit, and the class of \((\alpha+2\beta)/1\) is twice the class of \(\beta/1\). The localized root-reduction theorem derives the corresponding scalar identity for actual positive roots.

Relation to the OpenAI notes

Explicit coefficient maps and scalar reduction.

OpenAI notes: The edge-image proof localizes at the plane prime, passes to the quotient by the current root label, and expresses the remaining plane labels as uniformizers times units in the resulting localized edge ring. edge:bound.

Lean (primary): The coefficient operations are explicit tensor and quotient maps. A real-linear reduction kills the current root; writing another root as sa+rb gives a nonzero field coefficient r and therefore a unit multiple after reduction. The root-label compatibility formulas identify these elements with the actual polynomial labels. MinimalExtensions:1239, PlaneLocalization:1315, PlaneLocalization:1268, PlaneLocalization:1330, PlaneLocalization:1555.

Role in the proof: The reduced-label comparison is an equality up to an explicitly obtained unit. Later product arguments use associated elements in the localized quotient ring to compare actual scalar-image submodules.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.TitsSpace.localizedEdgeIdeal: PlaneLocalization.lean:1309.

OAI.KLInvariance.TitsSpace.LocalizedEdgeRing: PlaneLocalization.lean:1312.

OAI.KLInvariance.TitsSpace.localizedReduction: PlaneLocalization.lean:1315.

OAI.KLInvariance.TitsSpace.localizedReduction_self: PlaneLocalization.lean:1320.

Used by