Clearing a denominator in an embedded tensor localization
Let \(R\) be a commutative ring, \(P\subseteq R\) a multiplicative submonoid, \(S=P^{-1}R\) a localization, and \(M\) an \(R\)-module. For a submodule \(L\subseteq M\), let \[ L_S=\operatorname{im}(S\otimes_RL\longrightarrow S\otimes_RM) \] be its embedded scalar extension.
The submodule \(L_S\) is the localized submodule represented by numerators in \(L\). In particular, if \(x\in M\) satisfies \[ 1\otimes x\in L_S, \] there is \(d\in P\) such that \(dx\in L\).
Relation to the OpenAI notes
Two denominator factors convert localized membership into a scalar witness.
OpenAI notes: The notes pass from localized image equality to local membership in the free target, using flatness to commute localization with the finite kernel and image constructions. edge:bound, edge:lattice.
Lean (primary): The embedded tensor image is identified with a localized submodule. A numerator denominator and a second factor clearing equality in the localized module give an original-module scalar witness. The sheaf application verifies flat lower/upper image identities and compatibility of the scalar map on pure tensors before applying this lemma. TerminalKernels:909, TerminalKernels:914, TerminalKernels:946.
Role in the proof: The Blueprint displays both denominator factors and explains the exact transition from an equality of localized edge images to a multiplier outside the plane prime.
Proof
Step 1. Identify embedded tensors with localized submodule elements
The scalar extension \(L_S\) is the \(S\)-span of the images \(1\otimes m\), for \(m\in L\). The localized submodule is generated by those same images: a fraction \(m/s\), \(m\in L\), equals \[ (s/1)^{-1}(1\otimes m) \] under the usual tensor description of the localized module. Conversely, for a finite sum of scalar multiples, write each scalar as \(r_i/s_i\), with \(s_i\in P\). Under the tensor-to-fraction description it becomes \[ \sum_i(r_i/s_i)\otimes m_i = \frac{\sum_i r_i(\prod_{j\ne i}s_j)m_i}{\prod_i s_i}. \] The numerator belongs to \(L\) and the denominator belongs to \(P\). Thus every element of the span has a numerator in \(L\).
Thus the two constructions describe the same submodule of \(S\otimes_RM\). This is the source's baseChange_eq_localized identity.
Step 2. Choose a numerator in the original submodule
By Step 1, write the given localized element as \[ x/1=m/s,\qquad m\in L,\quad s\in P. \] Equivalently, the images of \(m\) and \(sx\) in the localized module agree.
The definition of equality of module fractions supplies another element \(t\in P\) such that \[ t(m-sx)=0\quad\text{in }M. \] Therefore \(tm=(ts)x\) in the original module.
Step 3. Multiply the two denominator factors
Set \(d=ts\). The submonoid is closed under multiplication, so \(d\in P\). Because \(m\in L\) and \(L\) is an \(R\)-submodule, \(tm\in L\). Hence \[ dx=(ts)x=tm\in L, \] which is the required original-module membership.
For a prime localization \(S=R_{\mathfrak p}\), the conclusion reads \(d\notin\mathfrak p\) and \(dx\in L\). The two factors \(s,t\) respectively represent the fraction and clear its equality in the module.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.TensorLocalMembership.baseChange_eq_localized: TerminalKernels.lean:909.
OAI.KLInvariance.TensorLocalMembership.denominator: TerminalKernels.lean:914.