Invariant Boltzmann–Gibbs Distribution

• Invariant Boltzmann–Gibbs Distribution is a fundamental equilibrium measure in statistical mechanics that emerges from equiprobability under additive constraints in high-dimensional phase spaces.
• Geometric derivations from both open and closed systems reveal that energy distributions converge to an exponential law in the thermodynamic limit, ensuring ensemble equivalence.
• Its robustness is demonstrated across diverse applications, from kinetic exchange models to diffusive stochastic systems, highlighting its universal significance in physical and economic contexts.

The invariant Boltzmann–Gibbs distribution is a fundamental object in statistical mechanics and probability theory, describing the stationary (equilibrium) state of many-particle and stochastic systems under broad conditions. Originating in the treatment of energy distributions in gas systems, it has become a universal concept for equilibrium distributions, appearing in physics, mathematics, and even economics. For a system with total conserved quantity (such as energy or money), when the number of degrees of freedom becomes large, the marginal distribution of the conserved quantity per constituent converges generically to the exponential Boltzmann–Gibbs law, independently of boundary conditions, provided equiprobability in the allowed phase-space region. Its geometric, probabilistic, and dynamical underpinnings ensure extreme robustness across ensemble treatments and application domains.

1. Geometric Derivation: Open and Closed Statistical Systems

The Boltzmann–Gibbs distribution emerges naturally from geometric considerations on high-dimensional phase spaces defined by additive constraints. Consider $N$ real, non-negative variables $E_i$ subject either to a simplex constraint $\sum_{i=1}^N E_i \leq E$ (open system) or a hyperplane constraint $\sum_{i=1}^N E_i = E$ (closed system). Equiprobability of all admissible states determines the volume or surface measure, leading to marginal distributions for a single $E_i$ as follows (Lopez-Ruiz et al., 2010):

• Open system (simplex):

$$p(E_i) = N E^{-1} \left(1-\frac{E_i}{E}\right)^{N-1}$$

• Closed system (hyperplane):

$p(E_i) = (N-1) \frac{(E-E_i)^{N-2}}{E^{N-1}}$

In the thermodynamic limit $N \to \infty$, setting the mean per site $\epsilon = E/N$, both marginals converge:

$p(E_i) \to \frac{1}{\epsilon} e^{-E_i/\epsilon}$

This is the canonical Boltzmann–Gibbs form, with $\epsilon = k T$ and $\beta = 1/(kT)$. Corrections to this law are $O(1/N)$ and vanish as system size grows, demonstrating that the exponential invariant measure is independent of the "open" (canonical) or "closed" (microcanonical) geometric construction (Lopez-Ruiz et al., 2010).

2. Maxwellian Distribution and High-dimensional Geometry

A parallel geometric argument applies in velocity or momentum space for ideal gases. The Maxwell–Boltzmann velocity law is derived considering accessible phase space inside or on the surface of a high-dimensional sphere ($\sum p_i^2 \leq R^2$ or $= R^2$). The marginal distribution for any component, for large $N$, becomes

$$f(p_i) \to \sqrt{\frac{1}{2\pi\epsilon}} \exp\left(-\frac{p_i^2}{2\epsilon}\right),$$

with $$\epsilon = \frac{1}{2} m \left\langle v^2 \right\rangle = \frac{1}{2} kT$$, leading to the classical Maxwellian distribution. The equivalence of microcanonical and canonical constructions persists here as well, underlining the high-dimensional geometric origin of such invariant measures (Lopez-Ruiz et al., 2010).

3. Dynamical and Probabilistic Foundations

Dynamical systems with local conserved quantities and simple unbiased exchange rules also exhibit invariant Boltzmann–Gibbs distributions. The unbiased money-exchange model is defined as a Markov process over $N$ agents with integer holdings $\{S_i\}$, total wealth fixed at $N m_0$, and exchange steps consisting of random donor–receiver pairs transferring unit wealth. In the mean-field large-$N$ limit, the process propagates to a deterministic ODE system, and the unique stationary law for individual holdings is

$$p^*_k = (1-\rho)\rho^k, \quad k \geq 0,\quad \rho = \frac{m_0}{1 + m_0}$$

which is geometric and approaches the exponential Boltzmann–Gibbs law for large $m_0$ (Cao et al., 2022). This stationary law is reached at an almost-exponential rate in relative entropy via entropy–dissipation inequalities.

A rigorous combinatorial treatment yields the same exponential limit for conservative dollar-exchange Markov chains on complete or connected graphs. The marginal law for each site is uniform over all integer compositions summing to $M$, which is exactly exponential in the limit of large $N,\,M$ (Lanchier, 2017).

System Type	Constraint	Invariant Law
Hamiltonian gas	$\sum E_i = E$ or $\leq E$	$\propto e^{-\beta E_i}$
Kinetic exchange	$\sum x_i = M$	$\propto e^{-x_i/T}$ (as $N \to \infty$)
Markovian agent	Stationary chain	Geometric $\to$ exponential

4. Gibbs Invariance in Diffusive Stochastic Systems

For continuous overdamped diffusion processes on $\mathbb{R}^d$ with drift $b(x) = -(\nabla U(x) + \ell(x))$ and additive noise, under appropriate growth and regularity conditions, the unique invariant measure is

$$\mu_\epsilon(dx) = \frac{1}{Z_\epsilon} e^{-U(x)/\epsilon} dx$$

This is the Boltzmann–Gibbs (or Gibbs–Boltzmann) distribution for arbitrary $U$ subject to Morse and confining conditions. It is invariant under the generator of the dynamics and the unique stationary distribution for the associated Fokker–Planck equation (Lee, 2024).

In metastable landscapes with multiple wells, the invariant Boltzmann–Gibbs measure remains unique, but convergence is governed by exponentially slow inter-well transitions. Mixing times scale as $\exp(\Delta U/\epsilon)$, depending on communication heights between minima, and the precise approach to equilibrium can be resolved by analyzing effective Markov chains on the clusters of wells (Lee, 2024).

5. Robustness and Universality Across Contexts

The essential mechanism underlying invariant Boltzmann–Gibbs measures is equiprobability under additive constraints in high dimensions. These distributions are insensitive to the choice of ensemble in the thermodynamic limit—differences between open (canonical) and closed (microcanonical) constructions vanish. The same law arises for microstates in physical systems, wealth in agent-based models, and more. It is also stable under a broad class of local exchange rules, provided they are conservative and reversible, and for non-global interaction topologies as long as connectivity and reversibility hold (Lopez-Ruiz et al., 2010, Lanchier, 2017).

6. Deviations and Generalizations: Nonstandard Invariant Measures

While the Boltzmann–Gibbs distribution is robust, certain kinetic settings break the classical invariance. For example, cold atoms in dissipative optical lattices with nonlinear momentum-dependent friction and confinement reach stationary phase-space measures of the form

$P_E(E) \propto (1+\sqrt{1+2E})^{-2/D}$

with $D$ characterizing friction nonlinearity. These solutions display power-law energy tails $(\propto E^{-1/D})$ and violate the canonical equipartition. Only in the deep-lattice ($D \to 0$) limit does the standard Boltzmann–Gibbs law return, demonstrating that nonlinearity and constrained kinetics can generate genuine nonequilibrium stationary states (Dechant et al., 2014).

7. Significance and Implications

The invariant Boltzmann–Gibbs law explains equilibrium statistics in a vast array of models, including Hamiltonian gases, diffusive stochastic dynamics, and economics-inspired agent systems. The geometric derivation clarifies the deep link between high-dimensional additive constraints and exponential stationary distributions, illuminating why canonical and microcanonical ensembles become equivalent as $N \to \infty$. Deviations signal novel physics, as in confined optical lattices, and motivate the study of generalized invariant measures. Understanding the conditions and limits of Boltzmann–Gibbs invariance remains central for both foundational and applied research in statistical mechanics and complex systems (Lopez-Ruiz et al., 2010, Cao et al., 2022, Lee, 2024, Lanchier, 2017, Dechant et al., 2014).