Entropy additivity from exponential decay of correlations: a coarse-grained operator approach

This paper provides a constructive derivation of thermodynamic extensivity by demonstrating that coarse-grained entropy becomes additive in the thermodynamic limit for systems with short-range interactions, provided the pair potential satisfies stability, temperedness, and exponential decay of correlations, while quantifying non-additivity and surface corrections for systems with long-range forces.

The Gist

The Big Question: Why Does "More" Equal "More"?

Imagine you have a cup of coffee. If you have two cups of the exact same coffee, you expect the total amount of "coffee-ness" (volume, heat, etc.) to be exactly double. In physics, this idea is called extensivity. It's the rule that says if you double the size of a system, you double its properties like energy and entropy.

Usually, physicists just assume this rule is true. They say, "It's a postulate; it just works."

Bob Osano's paper asks: Why does it work? Can we prove it starting from the tiny, microscopic rules that govern how individual atoms talk to each other?

The answer is: Yes, but only if the atoms stop caring about each other quickly enough.

---

The Main Idea: The "Blurry Camera" Approach

To prove this, the author uses a clever trick called Coarse-Graining.

Imagine you are looking at a high-resolution photo of a crowded stadium. It's too detailed to understand the big picture. So, you take a blurry camera and zoom out. You divide the stadium into big blocks (cells). Instead of counting every single person, you just count how many people are in each block.

In this paper:

1. The System: A gas of $N$ particles (like the crowd).
2. The Cells: The author divides the space into small boxes (cells).
3. The Operator: A mathematical tool (the "Combined Coarse-Graining Operator") that takes the detailed, messy data of every particle and turns it into a simple list of probabilities: "What is the chance a particle is in Box A?"

The Three Rules for "Normal" Behavior

The paper proves that for the "More equals More" rule (extensivity) to hold, the interactions between particles must follow three specific rules:

1. Stability: The particles can't attract each other so strongly that they collapse into a black hole. They need to stay somewhat spread out.
2. Temperedness (The "Short-Range" Rule): This is the most important one. It means particles only really "feel" their neighbors. If you move a particle far away, the force it feels drops to zero very quickly.
   • Analogy: Think of a party. If you are talking to your friend, you don't care what the person 50 feet away is saying. Your conversation is "short-range."
3. Exponential Decay: If you move two groups of particles far apart, the statistical link (correlation) between them disappears very fast—like a light fading out exponentially.

The Big Discovery: Entropy is Additive (Mostly)

The author calculates the Entropy (a measure of disorder or information) of the whole system by adding up the entropy of each little box.

• The Result: If the particles follow the "Short-Range" rule, the total entropy is almost exactly the sum of the parts.
• The Catch: There is a tiny, tiny error. The paper shows this error is proportional to $e^{-\ell/\xi}$.
  • Translation: If your boxes are much bigger than the distance over which particles interact ($\ell \gg \xi$), the error is so small it's basically zero.
  • Metaphor: If you are measuring the temperature of a room, and you ignore the tiny draft from a window 100 miles away, your calculation is perfect. The "error" from that distant window is exponentially small.

What Happens When the Rules Break? (Long-Range Forces)

What if the particles don't stop caring about each other? What if they have Long-Range Interactions?

• Analogy: Imagine a party where everyone is shouting at everyone else, no matter how far apart they are. Or, think of gravity: the Earth feels the pull of the Sun even though they are millions of miles apart.
• The Consequence: In these cases (like gravity or unscreened electricity), the "Short-Range" rule fails. The particles stay connected over huge distances.
• The Result: The "More equals More" rule breaks. You cannot simply add up the entropy of the parts to get the whole. The paper quantifies this failure using Mutual Information (a measure of how much two boxes "know" about each other). If the boxes are still "talking" to each other across the room, the system is non-additive.

The "Averaging" Problem (The Cosmological Connection)

The paper also points out a subtle mathematical trap.

Imagine you have a bumpy road.

1. Method A: Measure the height of every bump, calculate the "roughness" (entropy) for each bump, and then average those roughness numbers.
2. Method B: First, smooth out the road (average the height), and then calculate the roughness of the smooth road.

The paper proves these two methods give different results.

• Why? Because "roughness" is a non-linear concept. You can't just average the inputs and expect the output to be the average.
• The Connection: The author notes this is the same problem cosmologists face when trying to average the universe. If you average the universe first, then calculate its expansion, you get a different answer than if you calculate the expansion of every tiny patch and then average them. This paper shows this isn't just a gravity problem; it's a fundamental thermodynamic problem.

The "Surface" Correction

Finally, the paper clarifies a confusion in older textbooks.

• Textbooks often say the error in thermodynamic calculations comes from the "surface" (the edges of the container).
• This paper says: There are actually two types of errors.
  1. Bulk Error: Caused by particles in the middle of the room still talking to each other (the exponential error discussed above). This vanishes if the room is big enough.
  2. Surface Error: Caused by the walls of the room. This is a different kind of error that exists even if particles don't talk to each other at all.

Summary

1. Extensivity isn't magic; it's a result of particles only caring about their immediate neighbors.
2. If particles are "local" (short-range forces), the whole is exactly the sum of its parts (plus a tiny, invisible error).
3. If particles are "global" (long-range forces like gravity), the whole is not the sum of its parts. The system behaves differently.
4. Averaging is tricky: You can't just average a system and then calculate its properties; the order of operations matters, and this creates "backreaction" errors.

The paper provides a mathematical "blueprint" showing exactly how microscopic rules build up to the macroscopic laws we use every day, and exactly where those laws stop working.

Technical Summary

Technical Summary: Entropy Additivity from Exponential Decay of Correlations

Problem Statement
Thermodynamic extensivity—the property that thermodynamic potentials are homogeneous of degree one in their extensive variables (entropy $S$, volume $V$, particle number $N$)—is traditionally introduced as a postulate. While foundational works by Ruelle, Fisher, and Lebowitz–Penrose established the existence of the thermodynamic limit for stable and tempered potentials, the explicit mechanism linking microscopic interaction decay to the additivity of entropy remains often implicit. Furthermore, standard treatments frequently conflate two distinct sources of non-extensivity: bulk correlation effects and finite-size surface corrections. Additionally, the non-commutativity of spatial averaging with nonlinear thermodynamic functionals (such as entropy density) is a structural issue with parallels in relativistic cosmology (the Buchert–Ostermann problem) that lacks a clear thermodynamic formulation in classical statistical mechanics.

Methodology
The paper constructs a derivation of extensivity based on microscopic conditions rather than postulates, utilizing a combined coarse-graining operator $\mathcal{C}$ acting on the single-particle phase space $M = \Lambda \times \mathbb{R}^d$. The methodology proceeds as follows:

1. Framework: The system is defined as an $N$-body classical canonical Gibbs state. The analysis focuses on reduced $s$-particle densities $f^{(s)}$ rather than the full $N$-body phase space density.
2. Coarse-Graining: The single-particle phase space is partitioned into spatial cells $V_i$ of diameter $\ell$ and momentum cells $\Pi_\alpha$. A projection operator $\mathcal{C}$ maps the continuous reduced density $f^{(1)}$ to a piecewise-constant function over these cells, generating dimensionless mesoscopic probabilities $\pi_{i,\alpha}$.
3. Cluster Expansion: The Ursell cluster decomposition is employed to factorize the reduced densities. The $s$-particle density is expressed as a sum of products of lower-order densities plus connected $s$-body Ursell functions $u^{(s)}$, which encode genuine correlations.
4. Conditions: The derivation relies on three microscopic conditions on the pair potential $\phi$:
   • Stability: Boundedness from below.
   • Temperedness: Decay faster than $|r|^{-(d+\epsilon)}$ at large distances.
   • Exponential Cluster Decomposition: The integrated Ursell functions decay exponentially with a correlation length $\xi$.
5. Scale Separation: The analysis assumes a regime where the cell diameter $\ell$ satisfies $\xi \ll \ell \ll L$ (system size), ensuring cells contain many particles while remaining macroscopically small.

Key Contributions and Results

1. Constructive Derivation of Extensivity:
   The paper proves that entropy additivity is an emergent property resulting from the statistical factorization of the reduced density across spatial cells. Under the condition of exponential cluster decomposition, the coarse-grained entropy $S_{CG}$ satisfies:
   $$S_{CG} = \sum_i S_i + O\left( \frac{|\Lambda|}{\ell^d} e^{-\ell/\xi} \right)$$
   where $S_i$ is the marginal entropy of cell $i$. The correction term is exponentially suppressed per cell, vanishing in the limit $\ell/\xi \to \infty$. This recovers the thermodynamic limit of Ruelle and Fisher using explicit operator language.

2. Quantification of Non-Additivity in Long-Range Systems:
   For systems violating temperedness (e.g., gravitational or unscreened Coulomb systems where $|\phi(r)| \sim r^{-s_0}$ with $s_0 \leq d$), correlations decay algebraically rather than exponentially. The paper demonstrates that the mutual information between cells $I(i,j)$ does not vanish as separation increases, leading to a divergent series in the multi-information expansion. Consequently, $S_{CG} \neq \sum S_i$, and the departure from extensivity is quantified exactly by the inter-cell mutual information.

3. Non-Commutativity of Averaging and Nonlinear Functionals:
   The paper establishes that spatial averaging does not commute with nonlinear thermodynamic functionals (e.g., entropy density $\eta(\rho) = -k_B \rho \ln \rho$). By applying Jensen's inequality, it is shown that the entropy of a spatially averaged density exceeds the spatial average of the local entropy density. This "Jensen correction" is identified as the thermodynamic analogue of the backreaction problem in relativistic cosmology (Buchert–Ostermann commutation problem).

4. Generalized Euler Relation:
   The authors derive a generalized Euler relation for finite systems:
   $$U = TS - PV + \mu N + E_\partial$$
   where $E_\partial = O(|\partial \Lambda|)$ represents a surface correction arising from the breaking of translation invariance at the boundary. This distinguishes surface effects from the bulk correlation corrections derived in the main theorem.

5. Numerical Illustrations:
   The theoretical results are validated through numerical simulations of a 1D classical gas in three regimes: ideal (non-interacting), short-range interacting (Yukawa potential), and long-range interacting (algebraic decay). The simulations confirm:

   • Exponential restoration of additivity for short-range interactions as $\ell/\xi$ increases.
   • Power-law decay of correlations and non-additivity for long-range interactions.
   • Convergence of the coarse-grained entropy to the Gibbs fine-grained entropy.
   • The non-commutativity of the entropy functional under spatial averaging.

Significance and Claims
The paper claims to provide a constructive, operator-based derivation of thermodynamic extensivity, moving beyond the standard postulate of homogeneity. Its primary significance lies in:

• Explicit Mechanism: Making the factorization mechanism of the partition function explicit and quantitative via a joint coarse-graining operator on single-particle phase space.
• Higher-Order Bounds: Proving entropy additivity to all orders of the cluster expansion (pairs, triples, and higher cumulants), yielding a clean bound on the total deviation from extensivity.
• Structural Insight: Identifying the non-commutativity of nonlinear functionals with spatial averaging as the structural source of non-extensivity, thereby connecting classical statistical mechanics to the averaging problems in relativistic cosmology.
• Distinction of Corrections: Clearly separating bulk correlation corrections (exponentially small for short-range forces) from surface corrections (algebraic in system size), a distinction often blurred in standard textbook treatments.

The work reformulates established thermodynamic-limit mechanisms within an explicit coarse-graining and information-theoretic framework, quantifying all correction terms and linking thermodynamic extensivity to the decay of correlations. The authors note that extensions to quantum systems, non-equilibrium states, and higher-order perturbation theory in cosmology are reserved for future work.