A scalar difference annihilates a pair-projection quotient
Let \(R\) be a commutative ring, and let \(f:N\to X\) and \(g:N\to Y\) be surjective \(R\)-linear maps. Form the pair-projection quotient \[ Q=(X\oplus Y)/\{(f(n),g(n)):n\in N\}, \] with maps \(\ell(x)=[(x,0)]\) and \(u(y)=-[(0,y)]\).
Suppose \(T:N\to N\) is \(R\)-linear and \(a,b\in R\) satisfy \[ f(Tn)=a f(n),\qquad g(Tn)=b g(n)\qquad(n\in N). \] Then \((a-b)q=0\) for every \(q\in Q\).
Proof
Step 1. Represent the quotient class by one endpoint
The map \(\ell\) is surjective because \(g\) is. To see this directly, write a quotient element as \([(x,y)]\), choose \(n_0\) with \(g(n_0)=y\), and subtract its pair relation: \[ [(x,y)]=[(x-f(n_0),0)]. \] Thus write \(q=\ell(x_0)\).
Step 2. Lift the endpoint and cancel the second projection
Surjectivity of \(f\) supplies \(n\) with \(f(n)=x_0\). Put \[ h=Tn-bn. \] The two given action identities imply \[ f(h)=a f(n)-b f(n)=(a-b)x_0,\qquad g(h)=b g(n)-b g(n)=0. \] Therefore \(((a-b)x_0,0)\) belongs to the pair relation.
Step 3. Take its quotient class
The pair from Step 2 has zero quotient class. By linearity of \(\ell\), \[ (a-b)q=(a-b)\ell(x_0)=\ell((a-b)x_0)=0. \] This proves the claim. In the word-module application, \(T\) is right multiplication by one coefficient, while \(a,b\) are its coefficient characters at two weights.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.PairProjection.difference_annihilates: Decomposition.lean:213.