Prime localization and scalar-denominator membership
Let \(R\) be a commutative ring and \(\mathfrak p\) a prime ideal. Its complement \(S=R\setminus\mathfrak p\) is closed under multiplication. The localization \(R_{\mathfrak p}=S^{-1}R\) consists of fractions with denominators in \(S\).
For an \(R\)-module \(E\), its localized module \(E_{\mathfrak p}\) has fractions \(x/s\). Equality means \[ x/s=y/t \quad\Longleftrightarrow\quad \text{some }u\notin\mathfrak p\text{ satisfies }u(tx-sy)=0. \] For a submodule \(M\subseteq E\), the localized submodule \(M_{\mathfrak p}\) consists of fractions represented by a numerator in \(M\).
Clearing a denominator for membership
The condition \(x/1\in M_{\mathfrak p}\) is equivalent to \[ \exists a\notin\mathfrak p,\qquad ax\in M. \] Indeed, write \(x/1=m/s\) with \(m\in M\), \(s\notin\mathfrak p\). The fraction equality gives \(u(sx-m)=0\) for some \(u\notin\mathfrak p\). Hence \((us)x=um\in M\), and \(us\notin\mathfrak p\).
Conversely, \(ax\in M\) and \(a\notin\mathfrak p\) give the representation \(x/1=(ax)/a\).
For a principal prime ideal \((p)\), being outside the ideal is \(p\nmid a\). The source uses this scalar-denominator formulation in the prime-factor removal proof.
Generic span and torsion-free cancellation
Over a domain, \(E\) is torsion-free when multiplication by every nonzero scalar is injective. A vector \(x\) belongs to the generic span of \(M\) when some nonzero \(c\) satisfies \(cx\in M\). The whole target has full generic span if this holds for every \(x\in E\).
These are the two scalar conditions used in membership proofs: a nonzero multiplier starts the argument, and torsion-free cancellation removes each prime factor from it.
Height-one primes
The height of a prime ideal measures the lengths of chains of prime ideals ending there. In a domain, a height-one prime lies immediately above the zero prime.
In a unique factorization domain, the ideal generated by a prime element has height one. To see the absence of an intermediate nonzero prime \(\mathfrak q\subsetneq(p)\), choose a nonzero element of \(\mathfrak q\) and factor it into primes. Primality of \(\mathfrak q\) puts one factor \(r\) in \(\mathfrak q\). Then \(r\in(p)\), so \(p\mid r\); primality makes \(p,r\) associates, forcing \((p)\subseteq\mathfrak q\), a contradiction.
Units and the local maximal ideal
An element \(a/1\) is a unit in \(R_{\mathfrak p}\) exactly when \(a\notin\mathfrak p\). In that case its inverse is \(1/a\).
If \(a\in\mathfrak p\) and \((a/1)(b/s)=1\), fraction equality gives some \(u\notin\mathfrak p\) with \(u(ab-s)=0\). Thus \(us=uab\in\mathfrak p\), contradicting primality because \(u,s\notin\mathfrak p\).
Fractions whose numerators lie in \(\mathfrak p\) form the unique maximal ideal. Every other fraction is a unit, so every proper ideal is contained in this one.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.EdgeLattice.exists_denominator_of_localized_mem: ModuleAlgebra.lean:89.
OAI.KLInvariance.EdgeLattice.mem_iff_localized_at_height_one: ModuleAlgebra.lean:106.