The plane prime and its local coefficient ring
Fix a finite-generator Coxeter system. Let \(V_0\) be its original real root space, \(V\) its extended realization, and \(i_0:V_0\hookrightarrow V\) the embedding. Choose a finite basis of \(V\), giving \[ \Phi:\operatorname{Sym}_{\mathbb R}(V)\xrightarrow{\sim}A, \qquad A=\mathbb R[X_1,\ldots,X_N], \] and the linear-label map \[ \lambda:V_0\to A,\qquad \lambda(v)=\Phi(\iota(i_0v)). \]
The prime generated by a subspace
For any real subspace \(P\subseteq V_0\), define \[ \mathfrak p_P=(\lambda(v):v\in P)\subseteq A. \] This is planePrime. In the source it is the ideal map under \(\Phi\) of the symmetric-algebra ideal generated by \(i_0(P)\).
The linear-quotient theorem gives \[ A/\mathfrak p_P\simeq \operatorname{Sym}_{\mathbb R}(V/i_0(P)). \] The right side is a polynomial algebra over \(\mathbb R\), hence a domain. Therefore \(\mathfrak p_P\) is prime.
For positive roots \(a,b\) and \(P=\operatorname{span}_{\mathbb R}\{a,b\}\), \[ \mathfrak p_P=(\lambda(a),\lambda(b)). \] Linearity expresses the label of every vector in \(P\) as a real linear combination of these two labels. The plane-prime membership theorem records this specialization and the exact linear membership criterion.
The local coefficient ring
Define \[ R_P=A_{\mathfrak p_P} =(A\setminus\mathfrak p_P)^{-1}A, \qquad \eta_P:A\to R_P,\quad f\mapsto f/1. \] This is PlaneRing. Its elements are fractions \(f/s\), with \(s\notin\mathfrak p_P\). Since \(A\) is a domain, equality of fractions is checked by cross multiplication.
The nonunits are exactly the fractions whose numerator lies in \(\mathfrak p_P\). If \(f\notin\mathfrak p_P\), then \(s/f\) is the inverse of \(f/s\). If \(f\in\mathfrak p_P\) and \((f/s)(g/t)=1\), cross multiplication would give \(fg=st\). The left side lies in the prime ideal and the right side does not, a contradiction.
Fractions with numerator in \(\mathfrak p_P\) form a proper ideal: addition has numerator \(ft+gs\) when the two fractions are \(f/s,g/t\), and multiplication preserves membership in the prime ideal. Every element outside this ideal is a unit, so a proper ideal can contain only elements of it. It is therefore the unique maximal ideal \[ \mathfrak m_P=\mathfrak p_PR_P =\{f/s:f\in\mathfrak p_P,\ s\notin\mathfrak p_P\}. \] Thus \(R_P\) is a local ring. The geometric applications choose a two-dimensional root plane for \(P\).
Localized linear labels
The localized linear-label map is \[ \lambda_P:V_0\to R_P,\qquad \lambda_P(v)=\eta_P(\lambda(v)). \] This is localizedLinear and is real-linear. The plane-prime membership theorem gives \[ \lambda_P(v)\in\mathfrak m_P\Longleftrightarrow v\in P, \qquad \lambda_P(v)\text{ a unit}\Longleftrightarrow v\notin P. \]
In the illustrative ring \(A=\mathbb R[\alpha,\beta,\gamma]\) with \(\mathfrak p_P=(\alpha,\beta)\), both \(\gamma\) and \(1+\alpha\) become units. The elements \(\alpha,\beta\) lie in the maximal ideal.
Setting one localized root label equal to zero produces the localized edge ring, defined separately together with its reduction map.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.SymmetricPlaneIdeal.ideal: PlaneLocalization.lean:16.
OAI.KLInvariance.TitsSpace.embedLinear: PlaneLocalization.lean:346.
OAI.KLInvariance.TitsSpace.planePrime: PlaneLocalization.lean:355.
OAI.KLInvariance.TitsSpace.planePrime_isPrime: PlaneLocalization.lean:359.
OAI.KLInvariance.TitsSpace.PlaneRing: PlaneLocalization.lean:381.
OAI.KLInvariance.TitsSpace.localizedLinear: PlaneLocalization.lean:1305.