Derive ensemble density ρ_g from Hamiltonian dynamics to match equilibrium thermodynamics

Develop a principled method to construct an ensemble phase-space probability density ρ_g(q,p) directly from the underlying many-particle Hamiltonian dynamics such that ensemble averages computed with ρ_g reproduce thermodynamic quantities measured in equilibrium. The construction should specify how ρ_g is obtained from the microscopic equations of motion and clarify under what conditions this derivation is valid.

Background

Within the discussion of Gibbs entropy as a functional of an ensemble density ρ_g(q,p), the paper emphasizes that ρ_g is not inherently a property of the system and that its choice often relies on educated guesses or phenomenological arguments. Despite successes of ensembles in statistical mechanics, the authors highlight the difficulty of deriving the appropriate ρ_g starting from Hamiltonian dynamics in a way that connects directly to measurable thermodynamic quantities.

The text further notes that even establishing ergodicity via chaotic dynamics may neither be practically sufficient nor necessary for ensemble averages to equal equilibrium thermodynamic measurements, underscoring the challenge of grounding ρ_g in microscopic dynamics. This motivates the explicit unresolved question of how to systematically obtain ρ_g from first-principles dynamics.