Jacobi's formula for the logarithmic derivative of
ProvedFiniteFields.LinearAlgebra.logDerivative_fredholmDeterminantLet be a commutative ring, a free -module of finite rank , and . The Fredholm determinant of is the polynomial
defined without a basis as the reverse of the characteristic polynomial . Its constant term is , so is invertible in , and its negative logarithmic derivative is the power series
Statement. Let be a basis of indexed by a finite type , and let be the matrix of in . Then
where and is the classical adjugate matrix.
This is the integral, finite-dimensional form of Jacobi's formula: the numerator of the logarithmic derivative of is the trace of the adjugate resolvent. It is valid over any commutative ring, with no invertibility assumption on as a polynomial matrix, and underlies the identification of traces of powers of an operator with the coefficients of the logarithmic derivative of its Fredholm determinant.
Formalization Note. is PowerSeries.invOfUnit D_u 1, the power-series inverse of a series with constant coefficient , which exists over any commutative ring (over a field it is the ordinary inverse). The entries of are embedded in as constant polynomials, and the polynomials are viewed in by the canonical coercion. is fredholmDeterminant u and is fredholmLogDerivative u.
import Mathlib
import Definitions.Def_FiniteFields_Dwork_TraceFormula
open Module Polynomial PowerSeries
open FiniteFields.LinearAlgebra
universe u v w
variable {R : Type u} {M : Type v} [CommRing R]
variable [AddCommGroup M] [Module R M] [Module.Free R M] [Module.Finite R M]theorem FiniteFields.LinearAlgebra.logDerivative_fredholmDeterminant (u : Module.End R M)
{ι : Type w} [Fintype ι] [DecidableEq ι] (b : Basis ι R M) :
fredholmLogDerivative u =
((Matrix.trace
((1 - (X : R[X]) • (LinearMap.toMatrix b b u).map Polynomial.C).adjugate *
(LinearMap.toMatrix b b u).map Polynomial.C) : R[X]) : R⟦X⟧) *
PowerSeries.invOfUnit (fredholmDeterminant u : R⟦X⟧) 1 := by sorry