A probability measure on a configuration space that assigns weight proportional to exp(-β H(ω)) relative to a reference measure, where H is an energy (Hamiltonian) and β is an inverse temperature parameter, provided the normalization (partition function) is finite.

Principle

Reweight configurations by their energy so that lower-energy states are exponentially more probable at positive β, encoding equilibrium statistical mechanics for systems with many degrees of freedom.

Demonstration

For a finite lattice Ising model with Hamiltonian H(σ) summing nearest-neighbour couplings, the Gibbs measure on spin configurations σ is μ(σ) ∝ exp(-β H(σ)), and expectation under μ gives thermodynamic averages.

Misapplication

Applying the finite-volume Gibbs formula without controlling the partition function in infinite volume or using it for non-equilibrium dynamics where detailed balance fails.

Consequence

Gives a framework for computing thermodynamic expectations, phase coexistence, and correlation decay; in the infinite-volume limit Gibbs measures characterize equilibrium phases and their phases transitions.

Reversal

A microcanonical measure conditions on exact energy and assigns equal weight to configurations at that energy; reversing the Gibbs idea replaces exponential energy reweighting by a sharp energy constraint.

Boundary

Requires an energy functional and reference measure such that exp(-β H) is integrable; excludes driven steady states without a Gibbsian density and settings where interactions produce non-normalizable weights.

Semantic Tension

Tension with stationary non-equilibrium measures: both are probability distributions over configurations, but Gibbs measures arise from energy reweighting and detailed balance while non-equilibrium steady states need not have such structure.

Synthesis

A Gibbs measure is the equilibrium probability distribution over system configurations obtained by exponential reweighting by energy (with inverse temperature β) and normalized by the partition function when that normalization exists.