Definition Frobenius traces and trace duality
✓ definition formalized

Frobenius traces and trace duality

Let \(R,A\) be commutative rings and let \(\eta:R\to A\) give \(A\) its \(R\)-algebra structure. An \(R\)-linear trace is a map \[ t:A\to R,\qquad t(a+b)=t(a)+t(b),\qquad t(\eta(r)a)=r\,t(a). \] Its associated trace pairing is the \(R\)-bilinear map \[ A\times A\longrightarrow R,\qquad (a,b)\longmapsto t(ab). \] The word “trace” here names this chosen linear map. Its normalization is part of the data.

Trace-duality data consist of this trace and an \(R\)-linear isomorphism \[ D:A\xrightarrow{\sim}\operatorname{Hom}_R(A,R) \quad\text{with}\quad D(a)(b)=t(ab). \] Thus, for each \(R\)-linear functional \(\lambda:A\to R\), there is exactly one \(c\in A\) such that \[ \lambda(b)=t(cb)\qquad\text{for every }b\in A. \] This representative is \(D^{-1}(\lambda)\). Commutativity gives symmetry \(t(ab)=t(ba)\) and the identity \[ D(ca)(b)=D(a)(cb). \]

The finite free extension used in tensor induction

A finite free Frobenius extension in this development is an algebra map \(\eta:R\to A\) with:

  1. a finite \(R\)-basis of \(A\);
  2. an \(R\)-linear trace \(t\);
  3. the perfect trace pairing \(D\) above.

The source passes these as parameters: tr is \(t\), d is \(D\), and hd is the formula \(D(a)(b)=t(ab)\). The typeclass hypotheses Module.Free R A and Module.Finite R A supply the finite free module. The next construction combines this data with a finite free paired \(A\)-module.

Restriction of a perfect pairing through the trace uses the trace-duality data alone. Finite freeness supplies a finite basis when the construction subsequently extends scalars by \(A\otimes_R-\).

The reflection example

For \(A=k[X]\), where \(k\) is a characteristic-zero field, take \(R=k[X^2]\), with its inclusion into \(A\). Write \[ f=f_0(X^2)+Xf_1(X^2),\qquad t(f)=f_1(X^2). \] The \(R\)-basis is \(1,X\). If \(\lambda:A\to R\) is \(R\)-linear, its trace representative is \[ c=\lambda(X)+X\lambda(1). \] Indeed, \[ t\bigl(c(b_0+Xb_1)\bigr) =\lambda(X)b_1+\lambda(1)b_0 =\lambda(b_0+Xb_1). \] Testing against \(1\) and \(X\) proves uniqueness. This computation is the model for the general reflection Frobenius theorem.

References

The local source declarations below give the construction and proof.

Lean formalization

OAI.KLInvariance.FrobeniusDuality.restrictPairing: Frobenius.lean:180.

OAI.KLInvariance.FrobeniusDuality.inducedModule: Frobenius.lean:270.

Used by