A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
The ergodic hypothesis asserts that a system in equilibrium, over sufficiently long times, explores all accessible microstates with equal probability. This postulate justifies the use of ensemble averages to describe measurable quantities, since time-averaged observables converge to ensemble averages for ergodic systems.
From your study of ensemble theory you know that statistical mechanics replaces tracking the exact microstate of a system with averages over a probability distribution on phase space. From Liouville's theorem you know that this probability distribution flows like an incompressible fluid — volumes in phase space are preserved. But there is a conceptual gap: when you measure the pressure of a gas with a gauge, you are getting a *time average* over nanoseconds of molecular collisions, not a simultaneous average over imaginary copies of the system. The ergodic hypothesis is what bridges these two averages, asserting they are equal.
The hypothesis can be stated precisely: a system is ergodic if a single long trajectory in phase space visits every region of the energy surface (the hypersurface where the total energy equals the measured value) with a frequency proportional to that region's phase-space volume. In other words, *time average = ensemble average*. Formally: (1/T)∫₀ᵀ A(q(t),p(t)) dt → ⟨A⟩_{ensemble} as T → ∞. If this holds, then the single trajectory your real physical system traces through phase space produces the same observable statistics as the microcanonical ensemble that assigns equal weight to all accessible microstates.
The hypothesis is not automatically true for all systems — it is a postulate that must be justified or verified case by case. Some systems are definitively non-ergodic: a harmonic oscillator on a 2D torus with an irrational frequency ratio has trajectories that fill the surface densely, while one with a rational ratio produces closed loops that only sample a 1D subset. More physically interesting violations occur in integrable systems (too many conserved quantities to explore phase space freely) and in glasses or spin glasses (where the system gets stuck in metastable regions for astronomically long times). The KAM theorem from classical mechanics shows that even weakly perturbed integrable systems retain islands of non-ergodic behavior.
For practical statistical mechanics — gases, simple liquids, weakly interacting solids — ergodicity is a reliable working assumption because the number of degrees of freedom is enormous (∼10²³) and interactions are strong enough to scramble trajectories rapidly. The deeper justification comes from the modern theory of mixing: if the dynamics act chaotically on phase space, nearby trajectories diverge exponentially, and the time average converges to the ensemble average on practical timescales. The ergodic hypothesis, then, is not a rigorous theorem but a physically motivated bridge that makes the ensemble formalism predictively useful for real materials.