Which Phases Are Thermodynamically Realizable? A Local Entropy Criterion

This paper establishes that for continuous actions of locally compact amenable groups on compact metrizable spaces with finite topological entropy, an ergodic measure is a thermodynamically realizable equilibrium state if and only if the entropy map is upper semicontinuous at that measure, thereby characterizing unrealizable phases as those hidden behind the convex envelope of the free energy and correcting previous results regarding equilibrium face realization.

The Gist

Imagine you are a master chef running a very complex kitchen. In this kitchen, the "dishes" you serve are different thermodynamic phases (like ice, water, or steam). Your goal is to figure out which dishes you can actually cook up and serve to your customers (the "equilibrium states") using a specific set of ingredients (the "potentials" or interactions).

For a long time, scientists knew the rules for cooking in a perfectly sealed, finite kitchen (compact systems). But they struggled with the question: Which specific dishes are actually possible to make, and which ones are just illusions that look good on paper but can't be cooked?

This paper by C. Evans Hedges acts as a universal recipe book that finally answers this question, even for kitchens that are huge, infinite, or have open doors.

Here is the breakdown of the paper's discoveries using simple analogies:

1. The Core Problem: The "Impossible Dish"

In thermodynamics, every state of matter has an Entropy (a measure of disorder or "messiness") and an Energy. Nature loves to find the perfect balance between being messy and having low energy. This balance is called the Free Energy.

• The Analogy: Imagine a landscape of hills and valleys. The "Free Energy" is the height of the land. Nature always wants to roll down to the lowest valley.
• The Problem: Sometimes, the landscape has a weird "dip" that looks like a valley, but it's actually hidden behind a hill. If you try to roll a ball there, it will roll past it to a lower spot. In physics, this means some theoretical phases (dishes) are unrealizable. You can't actually cook them, even if the math says they exist.

2. The Big Discovery: The "Smoothness" Test

The author proves a simple rule to tell if a phase is real or fake. It depends on how "smooth" the Entropy landscape is right at that specific spot.

• The Rule: A phase is realizable (you can cook the dish) if and only if the Entropy map is "upper semicontinuous" at that point.
• The Metaphor: Imagine you are walking on a hill.
  • Realizable Phase: The ground is smooth or has a gentle slope. You can stand there comfortably.
  • Unrealizable Phase: The ground has a sudden, sharp drop-off or a "cliff" right where you are standing. If you try to stand there, you fall.
  • The "Hidden" Phases: These are the spots hidden behind the "convex envelope" (the straight line connecting two peaks). In physics, this is called the Maxwell Construction. It's like trying to build a house on a cliff edge; nature just won't let it stay there. The paper says: If the ground is smooth enough at your feet, you can build a house (a phase) there. If there's a cliff, you can't.

3. The "Group" of Chefs (Amenable Groups)

The paper doesn't just look at one chef; it looks at teams of chefs working together in different ways (mathematically called "groups").

• The Analogy: Whether you have a small team or a massive, infinite crew, as long as they work together nicely (amenable), the "Smoothness Test" still works.
• The Result: If the entropy is smooth at a specific point, there is always a specific "recipe" (a continuous potential) that will make that exact phase the unique winner.

4. Fixing a Mistake in the Old Recipe Book

The author points out a famous previous rule (by Jenkinson) that said: "If the whole kitchen is smooth, then any group of chefs can cook together."

• The Flaw: The author found a counter-example. Just because the whole kitchen is smooth doesn't mean a specific group of chefs can cook together if their specific corner of the kitchen is bumpy.
• The Fix: The new rule says: To cook a group of phases together, two things must happen:
  1. The entropy must be smooth at every single point in the group.
  2. The entropy must be continuous across the whole group (no sudden jumps between the chefs).
  • Analogy: You can't have a team of chefs where one is a master and the next is a novice with a sudden, jarring skill gap if you want them to work in perfect harmony. They need to flow into each other smoothly.

5. The Infinite Kitchen (Non-Compact Systems)

Most of the old math only worked in "finite" kitchens (compact spaces). But what about infinite kitchens, like a system with an infinite number of possible states (like a Markov shift with infinite symbols)?

• The Trick: The author uses a technique called One-Point Compactification.
• The Metaphor: Imagine an infinite, sprawling city. It's too big to manage. So, the author adds a "magic door" at the very edge of the city that leads to a single point called "Infinity." Suddenly, the infinite city becomes a finite, manageable sphere.
• The Result: Once you turn the infinite city into a finite sphere, you can apply the "Smoothness Test." If the phase works in this new sphere, it works in the original infinite city, provided you use a special type of ingredient called a $C_0$ potential (which basically means the ingredients fade away to zero as you get further out in the infinite city).

Summary: What Does This Mean for You?

This paper solves a fundamental mystery in physics and math: Which states of matter are actually possible?

• Before: We had a list of "maybe" phases and a list of "impossible" phases, but the boundary was fuzzy.
• Now: We have a clear, local test. If the "entropy landscape" is smooth at a specific point, that phase is real. If it's jagged or hidden behind a convex curve, it's a ghost.
• The Impact: This helps scientists understand complex systems, from infinite grids of atoms to chaotic weather patterns, by telling them exactly which configurations nature will actually allow to exist.

In short: Nature is picky. It only serves dishes where the "messiness" (entropy) changes smoothly. If the recipe requires a sudden jump or a hidden cliff, nature says, "No thanks, we can't cook that."

Technical Summary

1. Problem Statement

The paper addresses a fundamental question in the variational approach to statistical mechanics and thermodynamic formalism: Which invariant measures (thermodynamic phases) can arise as equilibrium states for some continuous potential?

In the variational principle, an equilibrium state $\mu$ for a potential $\phi$ maximizes the pressure functional $P(\phi) = \sup_{\nu} (h(\nu) + \int \phi d\nu)$, where $h$ is the Kolmogorov-Sinai entropy. The set of all such equilibrium states corresponds to the set of "realizable" phases. The central problem is to characterize the subset of ergodic measures $\mu$ that are equilibrium states for some continuous potential $\phi \in C(X)$.

Previous work (e.g., Jenkinson, Israel, Phelps) often relied on global assumptions, such as the entropy map $h$ being upper semicontinuous (USC) everywhere on the space of invariant measures. This paper seeks to determine if a local condition is sufficient and to correct omissions in existing theorems regarding the realization of sets of phases (equilibrium faces).

2. Methodology

The author employs a synthesis of convex analysis, Choquet theory, and ergodic optimization.

• Choquet Simplex Framework: The space of invariant probability measures $M_T(X)$ is treated as a compact metrizable Choquet simplex, where the ergodic measures $M_T^{erg}(X)$ form the extreme boundary.
• Dual Formulation: The problem is recast using the Legendre-Fenchel conjugate. The entropy map $h$ is viewed as a function on the simplex. The pressure $P$ is its conjugate. The realizability of a phase $\mu$ is linked to whether $\mu$ lies on the "convex envelope" (biconjugate) of the free energy landscape.
• Local Regularity Analysis: Instead of requiring global USC of $h$, the author constructs specific continuous affine functions (potentials) that dominate the entropy map locally around a target measure $\mu$. This involves:
  • Using the facial topology on the set of ergodic measures.
  • Constructing "staircase" functions using distance functions and upper semicontinuity neighborhoods to approximate the entropy from above.
  • Applying Lau's Extension Theorem to extend continuous functions from the extreme boundary to the whole simplex.
• Embedding Techniques: For non-compact systems, the author utilizes one-point compactification and invariant subsystem embeddings to map the problem back to a compact setting where the variational principle holds.

3. Key Contributions

A. The Local Entropy Criterion (Main Theorem)

The paper establishes that the global regularity of the entropy map is unnecessary. The realizability of a single ergodic phase depends entirely on the local behavior of entropy at that point.

• Theorem 1 & Corollary 2: For a continuous action of a locally compact amenable group on a compact metrizable space with finite topological entropy, an ergodic measure $\mu$ is an equilibrium state for some continuous potential $\phi$ if and only if the entropy map $h$ is upper semicontinuous at $\mu$.
  • Implication: Phases "hidden" behind the Maxwell construction (where the biconjugate of the free energy is strictly less than the free energy) are exactly those where $h$ fails to be USC.

B. Correction of Equilibrium Face Realization

The paper corrects a result by Jenkinson (2006) regarding the realization of sets of phases (equilibrium faces).

• The Flaw: Jenkinson claimed that any weak-$\ast$ closed set of ergodic measures $E$ generates an equilibrium face if entropy is globally USC.
• The Counterexample: The author provides a counterexample (Example 3.4) on the full shift where $E$ is weak-$\ast$ closed and $h$ is globally USC, but $h$ is discontinuous on $E$. In this case, $E$ cannot be an equilibrium face.
• Theorem 3 (Corrected Statement): A weak-$\ast$ closed set of ergodic measures $E$ determines an equilibrium face (i.e., $Eq(\phi) = \text{co}(E)$) if and only if:
  1. $h$ is continuous when restricted to $E$ ($h|_E$ is continuous).
  2. $h$ is upper semicontinuous at every point $\mu \in E$.

C. Extension to Non-Compact Systems

The results are extended to systems that are not compact, specifically locally compact $\sigma$-compact systems and countable-state Markov shifts.

• Theorem 4 & 5: By embedding the system into a compact space (via one-point compactification) and ensuring the entropy is bounded, the local USC criterion applies.
• Corollary 6: For countable-state Markov shifts with finite entropy, every ergodic measure is the unique equilibrium state for some $C_0$ potential (continuous functions vanishing at infinity), provided the entropy is USC at that measure. (Note: For finite entropy shifts, $h$ is automatically USC everywhere, implying all ergodic measures are realizable).

4. Key Results Summary

1. Characterization of Realizable Phases: An ergodic measure $\mu$ is realizable $\iff$ $h$ is USC at $\mu$.
2. Geometric Interpretation: Realizable phases are those lying on the convex envelope of the free energy. Non-realizable phases correspond to the "flat" regions created by the Maxwell construction where the system prefers a mixture of phases over the pure phase.
3. Necessity of Continuity on Faces: For a set of phases to coexist at equilibrium, the entropy must be continuous across that set, in addition to being USC at each point.
4. Application to Markov Shifts: The theory resolves realizability questions for countable-state systems, showing that under finite entropy conditions, all ergodic measures are realizable via $C_0$ potentials.

5. Significance

• Refinement of Thermodynamic Formalism: The paper shifts the paradigm from global assumptions (global USC) to local conditions, providing a precise, necessary, and sufficient criterion for phase realizability.
• Resolution of Open Problems: It settles the question of which phases are "thermodynamically accessible" and corrects a significant gap in the literature regarding the realization of equilibrium faces (Jenkinson's Theorem).
• Bridging Compact and Non-Compact: By utilizing one-point compactification, the results unify the thermodynamic formalism for compact systems with the increasingly important field of countable-state Markov shifts and non-compact dynamical systems.
• Physical Insight: The results clarify the physical meaning of the Maxwell construction in dynamical systems: phases that are "unrealizable" are those that are thermodynamically unstable and will spontaneously decompose into a mixture of other phases to lower the free energy.

In summary, Hedges provides a rigorous, localized geometric characterization of thermodynamic phases, demonstrating that the "smoothness" (upper semicontinuity) of the entropy landscape at a specific point is the sole determinant of whether that phase can be stabilized by a physical interaction.