Questions: Liouville's Theorem

Question 1

Liouville's theorem says the ensemble cloud evolves 'like an incompressible fluid.' A student concludes this means the cloud cannot change shape. What is wrong with this interpretation?

• A. Nothing — the theorem guarantees both volume and shape are preserved under Hamiltonian evolution
• B. The cloud can change shape arbitrarily — stretching, twisting, and distorting — as long as its total phase-space volume is conserved
• C. The cloud actually shrinks over time as the system dissipates energy to the environment
• D. The incompressible fluid analogy is only approximate; the theorem strictly applies only at equilibrium

Answer: (B)

Liouville's theorem conserves phase-space volume, not shape. The ensemble cloud can deform drastically — stretching in some directions and contracting in others — as long as its total volume remains constant. This is exactly like an incompressible fluid: a blob of water can be stirred, stretched, and folded without changing its total volume. Shape preservation is a much stronger condition and is generally false. In practice, the cloud often becomes highly filamentary and convoluted, spreading across phase space while maintaining constant volume — a phenomenon connected to the approach to apparent equilibrium.

Question 2

Why does Liouville's theorem require that the equilibrium phase-space density ρ_eq must be a function of the Hamiltonian H alone?

• A. Because H is the only conserved quantity in Hamiltonian mechanics, so all equilibrium distributions must depend on it
• B. Because for a stationary ensemble ∂ρ/∂t = 0, which requires {ρ, H} = 0, and this Poisson bracket vanishes exactly when ρ depends only on H
• C. Because the Boltzmann distribution exp(−H/kT) is the unique solution to the Liouville equation in all cases
• D. Because kinetic and potential energy must be equally distributed at equilibrium by the equipartition theorem

Answer: (B)

For a stationary (equilibrium) ensemble, ∂ρ/∂t = 0. Liouville's equation then requires {ρ, H} = 0. The Poisson bracket {ρ, H} = Σᵢ (∂ρ/∂qᵢ ∂H/∂pᵢ − ∂ρ/∂pᵢ ∂H/∂qᵢ) vanishes if ρ depends on (q,p) only through H — because then ∂ρ/∂qᵢ = (dρ/dH)(∂H/∂qᵢ) and ∂ρ/∂pᵢ = (dρ/dH)(∂H/∂pᵢ), and the cross terms cancel. This is why the canonical ensemble ρ ∝ exp(−H/kT) automatically satisfies the Liouville condition — it is a function of H — and why equilibrium distributions in general take this form.

Question 3

Liouville's theorem implies that Hamiltonian evolution preserves information: no two phase-space trajectories can ever merge.

• T. True
• F. False

Answer: True

If two distinct trajectories merged at some point in phase space, the region of phase space between them would have been compressed to zero volume — a direct violation of volume conservation. Since Liouville's theorem guarantees volume is preserved, trajectories cannot merge, and the mapping from initial conditions to later states is one-to-one (invertible). This means no information about the initial microstate is ever lost under Hamiltonian evolution — a profound consequence that underlies debates about the arrow of time and quantum information.

Question 4

Liouville's theorem states that the phase-space density ρ at a fixed location in phase space remains constant over time.

• T. True
• F. False

Answer: False

This confuses the Lagrangian and Eulerian descriptions. Liouville's theorem states that ρ is constant along trajectories — that is, dρ/dt = ∂ρ/∂t + {ρ,H} = 0, which says the density 'following a moving point' doesn't change. But the density at a fixed point in phase space (the Eulerian view, ∂ρ/∂t) generally does change as the ensemble cloud flows through that location. Only in equilibrium, where {ρ,H} = 0, does ∂ρ/∂t = 0 at every fixed point.

Question 5

Explain how Liouville's theorem connects to the ergodic hypothesis and why this connection is foundational for statistical mechanics.

Model answer: Liouville's theorem establishes that Hamiltonian evolution preserves phase-space volume, so the ensemble cloud cannot compress into a smaller region. The ergodic hypothesis strengthens this: if a system is ergodic, the trajectory of a single system eventually samples every microstate on the energy surface with equal frequency. Because Liouville ensures volume is preserved under the flow, the time average of a quantity equals the ensemble average over the energy surface — allowing statistical mechanics to replace time averages (what we actually measure) with ensemble averages (what we can calculate).

Without Liouville's theorem, phase-space volumes could shrink over time, the system might preferentially revisit certain regions, and the identification of time averages with ensemble averages would fail. The theorem is what licenses the core move of statistical mechanics: trading a single, unknowable trajectory for a statistical ensemble. It also establishes that equilibrium ensembles must have ρ as a function of H alone (since {ρ,H}=0 at equilibrium), which justifies the form of the canonical, microcanonical, and grand canonical ensembles from first principles.