Near thermodynamic equilibrium, flows of extensive quantities (heat, matter, charge) depend linearly on the thermodynamic forces (gradients of the conjugate intensive variables). Some of the coefficients in these linear laws describe cross effects: a temperature gradient driving an electric current (Seebeck effect), a voltage driving a heat flow (Peltier effect), a temperature gradient driving diffusion (Soret effect) and its converse (Dufour effect). In 1931 Lars Onsager showed that, as a consequence of the time-reversibility of microscopic dynamics, the matrix of these coefficients is symmetric (Onsager 1931). These reciprocal relations were confirmed experimentally across many classes of irreversible processes (Miller 1960), and Onsager received the 1968 Nobel Prize in Chemistry for them.
This mission formalizes the abstract, finite-dimensional version of Onsager's principle as presented by Landau and Lifshitz (Statistical Physics, Part 1, 1975) and reproduced in the Wikipedia article Onsager reciprocal relations, sections "Abstract formulation" and "Proof".
Let be the fluctuations of thermodynamic quantities from their equilibrium values. For small fluctuations the entropy is quadratic, and Boltzmann's formula gives the Gaussian fluctuation distribution
where is a positive definite symmetric real matrix and summation over repeated indices is understood. Write for the average of with respect to .
The thermodynamic conjugate quantities are . In the quasi-stationary regime the fluctuations relax according to for a real matrix , so that with the kinetic coefficients
The mean value at time of a fluctuation which takes the value at is , and . The time correlation is .
If fluctuations are symmetric under time reversal,
then the kinetic coefficients are symmetric:
The symmetry is the mathematical content of the reciprocal relations: it reduces the number of independent transport coefficients from to and yields the equality of measurable cross-coefficients such as the Peltier and Seebeck coefficients. Together with the positive semi-definiteness of the coefficient matrix (from non-negative entropy production), it is the basic structural input of linear irreversible thermodynamics.
The result is classical and its proof is standard in physics textbooks; this mission produces a machine-checked version of the finite-dimensional argument, including the Gaussian moment identity and the differentiation of the time-correlation function under the integral sign. No machine-checked formalization of this statement is known to the proposer.
The algebraic core is short, but the formal argument needs analytic input that the textbook treats as routine: Gaussian integrals over for a general positive definite (integrability of polynomial moments against and the value of the second moments), differentiation of a parameter-dependent integral involving a matrix exponential, and the passage from an identity valid for to an identity between derivatives at the boundary point , where only a one-sided derivative is available from the hypothesis.
Vectors are functions Fin n → ℝ with Lebesgue measure; matrices are Matrix (Fin n) (Fin n) ℝ, and β.PosDef (which includes symmetry) is assumed in every theorem. The average is the ratio of Bochner integrals, so the normalizing constant cancels. The mean relaxation is the matrix exponential . The time-reversal hypothesis is assumed only for . The entropy and Boltzmann's constant are not modelled; is the primitive datum, as in the source's statement of the principle. The hypothesis is non-vacuous (e.g. satisfies it), and the conclusion concerns , not .
All definitions are collected in Definitions.Def_OnsagerReciprocal_basic (namespace OnsagerReciprocal). Reusable Gaussian-moment lemmas for positive definite quadratic forms are welcome contributions.
namespace OnsagerReciprocal
theorem onsager_reciprocal_relations {n : ℕ} (β lam : Matrix (Fin n) (Fin n) ℝ)
(hβ : β.PosDef)
(hrev : ∀ t : ℝ, 0 ≤ t → ∀ i k : Fin n,
timeCorrelation β lam t i k = timeCorrelation β lam t k i) :
∀ i k : Fin n, kineticCoeff β lam i k = kineticCoeff β lam k i := by sorry
end OnsagerReciprocalOnsager's principle (principle of symmetry of kinetic coefficients).
Let be fluctuations of thermodynamic quantities from equilibrium, distributed according to the Gaussian law
where is a positive definite symmetric real matrix (in the source, ). In the quasi-stationary regime for a real matrix , so the mean value at time of the fluctuation that equals at is . The thermodynamic conjugate quantities are , and the kinetic coefficients are
Assume the symmetry of fluctuations under time reversal: for every and all ,
where is the average with respect to . Then the matrix of kinetic coefficients is symmetric:
This is the abstract form of the Onsager reciprocal relations, which express the equality of cross-coefficients between thermodynamic flows and forces (e.g. Peltier/Seebeck, thermodiffusion/Dufour) near equilibrium.
Formalization Note The Gaussian average is the ratio over Lebesgue measure on (so the normalizing constant cancels). The time-reversal hypothesis is imposed only for , where the quasi-stationary relaxation is meant to apply.
No open leaves. Every sub-goal is proved or awaiting decomposition.