Coefficient transport of dual modules and perfect pairings
Let \(k\) be a field, \(e:A\xrightarrow{\sim}B\) an isomorphism of commutative \(k\)-algebras, and \(M\) an \(A\)-module with compatible \(k\)-scalar action. Write \({}^eM\) for coefficient transport, with \(b\star m=e^{-1}(b)m\).
The dual module is \(\operatorname{Hom}_A(M,A)\). There is a \(B\)-linear isomorphism \[ {}^e\operatorname{Hom}_A(M,A) \xrightarrow{\sim}\operatorname{Hom}_B({}^eM,B), \] sending \(\lambda\) to the functional \[ \lambda^e(m)=e(\lambda(m)). \]
A perfect pairing represented by an \(A\)-linear isomorphism \(D:M\to\operatorname{Hom}_A(M,A)\) transports to a perfect pairing with value \[ D^e(m,n)=e(D(m,n)). \]
Proof
Step 1. Check the forward functional
For \(b\in B\), \[ \lambda^e(b\star m) =e(\lambda(e^{-1}(b)m)) =e(e^{-1}(b)\lambda(m)) =b\,e(\lambda(m)). \] Additivity follows from additivity of \(\lambda\) and \(e\). Thus \(\lambda^e\) is \(B\)-linear.
Step 2. Construct the inverse functional
For a \(B\)-linear functional \(\mu:{}^eM\to B\), set \[ \lambda(m)=e^{-1}(\mu(m)). \] Old multiplication by \(a\in A\) is new multiplication by \(e(a)\). Hence \[ \lambda(am) =e^{-1}(e(a)\mu(m)) =a\,e^{-1}(\mu(m)), \] so \(\lambda\) is \(A\)-linear.
The composites are identities because \(e^{-1}e=1_A\) and \(ee^{-1}=1_B\).
Step 3. Verify linearity on the dual modules
The transported scalar action on the old dual is \(b\star\lambda=e^{-1}(b)\lambda\). Evaluating its forward image at \(m\) gives \[ e(e^{-1}(b)\lambda(m))=b\,e(\lambda(m)). \] Thus the forward functional correspondence is itself \(B\)-linear. Together with Step 2 it is the stated linear isomorphism.
Step 4. Transport the pairing isomorphism
Coefficient transport sends the given isomorphism \(D\) to an isomorphism \[ {}^eM\longrightarrow{}^e\operatorname{Hom}_A(M,A). \] On underlying vectors this is the original \(D\), and its inverse is the transported original inverse.
Compose it with the dual isomorphism from Steps 1–3. The result is a \(B\)-linear isomorphism from \({}^eM\) to its \(B\)-dual, hence a perfect pairing. Evaluating that composite gives \(D^e(m,n)=e(D(m,n))\).
Step 5. Preserve the relevant degree and symmetry identities
If \(e(A_j)=B_j\) and the module grading is transported, a homogeneous functional of degree \(r\) still sends degree \(n\) into degree \(n+r\).
Likewise, if the old pairing sends degrees \(i,j\) into \(A_{i+j-L}\), the new value lies in \(B_{i+j-L}\), preserving its shift \(L\). Applying \(e\) to a symmetry equation preserves symmetry.
This is the coefficient-level functional calculation used when symmetric coefficient rings are replaced by polynomial coordinates.
References
The local source declarations below give the construction and proof.
Lean formalization
OAI.KLInvariance.CoefficientChange.mapEquiv: Support.lean:1771.
OAI.KLInvariance.CoefficientChange.dualTo: Support.lean:1778.
OAI.KLInvariance.CoefficientChange.dualFrom: Support.lean:1787.
OAI.KLInvariance.CoefficientChange.dualEquiv: Support.lean:1802.
OAI.KLInvariance.CoefficientChange.pairingEquiv: Support.lean:1824.
OAI.KLInvariance.CoefficientChange.pairingEquiv_apply: Support.lean:1829.