Far-from-equilibrium complex landscapes

This paper generalizes the concept of complex landscapes to far-from-equilibrium systems by demonstrating how a stochastic spin model with non-reciprocal and heterogeneous interactions exhibits hidden spontaneous oscillations that can be unveiled through entropy production density and characterized via a configurational entropy counting nonequilibrium collective states.

The Gist

Imagine a massive, chaotic dance floor filled with thousands of dancers. In a calm, "equilibrium" world, this dance floor eventually settles down. Everyone finds a comfortable spot, stops moving, and the whole room becomes still. This is like a frozen lake or a glass of water that has stopped flowing.

But what happens if the music changes, the dancers start pushing each other in weird, non-repeating ways, and the room is full of obstacles? That is the world of far-from-equilibrium systems that Laura Guislain and Eric Bertin are exploring.

Here is a simple breakdown of their discovery using everyday analogies:

1. The "Rugged Landscape" of Possibilities

Scientists often describe complex systems (like evolving species, brain networks, or even crowds) as a landscape.

• The Old View: Imagine a mountain range with many valleys. A ball rolling down will eventually get stuck in one valley. That valley represents a "state" where the system settles down.
• The New View: The authors show that when systems are pushed hard (far from equilibrium) and the interactions are messy (disordered), the landscape isn't just about still valleys. It's full of spinning merry-go-rounds.

In this new landscape, the system doesn't just sit still; it gets stuck in loops, spinning endlessly. These are spontaneous oscillations.

2. The Magic Trick: Why You Can't See the Dance

The researchers built a mathematical model (a "spin model") to test this. They found something tricky:

• The Illusion: If you look at the "average" of the whole dance floor (like looking at the total magnetization of the room), everything looks boring and still. The disorder (the messy obstacles) hides the movement. It's like watching a stadium from far away; you might just see a blur of color, not realizing that specific groups of people are doing synchronized dances.
• The Reveal: To see the truth, you have to look at specific "generalized" angles. When the researchers tuned their "lens" to look at specific groups, they saw that different groups were indeed spinning in different loops.

3. The "Entropy Production" Meter

How do you know if the system is truly spinning and not just sitting still?

• The Metaphor: Think of entropy production as a "friction meter" or a "waste heat gauge."
• Stillness: If the system is just sitting in a valley (equilibrium), it produces no waste heat. The meter reads zero.
• Spinning: If the system is stuck in a loop (oscillating), it is constantly fighting against itself. It generates "friction." The meter reads positive.
• The Discovery: The authors found that even when the system looks still to the naked eye, this "friction meter" is ticking. This proves the system is alive, active, and far from equilibrium.

4. Counting the Merry-Go-Rounds (Configurational Entropy)

The most exciting part is how they counted these spinning states.

• The Problem: In a huge system, there are so many possible spinning states that counting them one by one is impossible.
• The Solution: They invented a way to count them using Configurational Entropy. Think of this as a "population census" for the different types of merry-go-rounds.
  • They asked: "How many different spinning loops exist that produce a specific amount of 'friction'?"
  • They found that in certain conditions, there isn't just one or two loops. There are exponentially many of them. The number of possible spinning states grows so fast it becomes a massive "forest" of possibilities.

5. The Battle: Stillness vs. Spinning

The paper describes a competition between two types of states:

1. The Sleepers: States where everything is still (fixed points).
2. The Dancers: States where everything is spinning (oscillating).

The authors found that which one wins depends on the "temperature" (how much energy is in the system):

• Too Hot: The system is too chaotic to hold any shape; it's just a paramagnetic blur.
• Just Right: The "Dancers" win. There are so many more spinning states than still ones that the system must be spinning. The whole system becomes a macroscopic, irreversible machine.
• Too Cold: The "Sleepers" win. The system freezes into a glassy, stuck state (a Spin Glass).

Summary

In simple terms, this paper shows that when you take a complex, messy system and push it out of balance, it doesn't just freeze or settle. It can get trapped in a vast, hidden universe of spinning loops.

Even though these loops might be invisible if you look at the system from a distance, they are real. They generate "friction" (entropy), and there are often so many of them that they dominate the system's behavior. This helps us understand how complex things like biological clocks, neural networks, or crowds can stay active and rhythmic without ever settling down.

Technical Summary

Technical Summary: Far-from-Equilibrium Complex Landscapes

Problem Statement
Complex systems, such as glasses, biological evolution models, and neural networks, are frequently described using the concept of a "rugged landscape" containing a vast number of collective states. In equilibrium systems (e.g., supercooled liquids), these states correspond to time-independent configurations (fixed points) characterized by a free energy landscape. However, systems driven far from equilibrium by external forces or local activity often exhibit non-reciprocal interactions, breaking detailed balance. This leads to time-dependent collective behaviors, such as spontaneous temporal oscillations. The challenge lies in generalizing the complex landscape concept to these far-from-equilibrium regimes, particularly when disorder obscures the observation of these oscillations in standard macroscopic observables.

Methodology
The authors investigate a minimal mean-field stochastic spin model designed to incorporate non-reciprocal and heterogeneous interactions. The model consists of $2N$ microscopic variables: $N$ spins ($s_i = \pm 1$) and $N$ fields ($h_i = \pm 1$).

• Dynamics: The system evolves via stochastic Markov dynamics where single spins or fields flip with transition rates $W$ dependent on an energy function $E_k$.
• Non-Reciprocity: A parameter $\mu$ controls the breaking of detailed balance. For $\mu \neq 0$, interactions between spins and fields become non-reciprocal, driving the system out of equilibrium.
• Disorder: The model includes separable disorder ($K_{ij} = \epsilon_i \epsilon_j$) and quenched random ferromagnetic coupling constants $J(\epsilon)$ drawn from a Gaussian distribution.
• Glassy Behavior: Inspired by the Random Energy Model (REM), the disorder configuration $\epsilon$ evolves on a timescale much slower than the spin dynamics, allowing the system to explore different "states" (labeled by $\epsilon$) associated with different energy levels and coupling constants.
• Observables: To detect hidden oscillations, the authors analyze:
  1. Generalized Magnetization ($m_\alpha$): Defined for specific disorder configurations $\alpha$, revealing oscillations that are averaged out in the global magnetization $m$.
  2. Entropy Production Density ($\sigma$): Defined as $\sigma = \frac{1}{N} \sum_{C' \neq C} W(C'|C)P(C) \ln \frac{W(C'|C)}{W(C|C')}$, serving as an order parameter for irreversibility.
  3. Configurational Entropy: A statistical measure counting the number of collective states with a given entropy production density.

Key Contributions and Results

1. Generalization of the Landscape Concept: The paper demonstrates that the complex landscape framework can be extended to far-from-equilibrium systems. The landscape includes not only time-independent fixed points but also spontaneously oscillating states (limit cycles).
2. Identification of Hidden Oscillations: The study reveals that in disordered systems, spontaneous oscillations in different collective states correspond to distinct cycles in phase space. These are invisible to standard macroscopic observables (like global magnetization) because the phases of oscillations in different states are uncorrelated. However, they are explicitly revealed by state-dependent generalized magnetizations and, crucially, by a non-zero entropy production density.
3. Phase Diagram Characterization: The authors map the phase diagram in terms of temperature ($T$) and non-reciprocity ($\mu$):
   • Paramagnetic Phase: High $T$ for all $\mu$.
   • Spin-Glass Phase: Low $T$ and low $\mu$ (characterized by a non-zero Edwards-Anderson parameter).
   • Complex Oscillating Phase: Low $T$ and high $\mu$. In this regime, the system exhibits multiple glassy oscillating states.
4. Configurational Entropy of Oscillating States: The authors define a configurational entropy $S_c(\sigma)$ that counts the number of oscillating states with a specific entropy production density $\sigma$.
   • For $T > T_g$ (glass transition temperature), the number of oscillating states grows exponentially with system size $N$.
   • A competition exists between oscillating states (with $\sigma > 0$) and time-independent states (fixed points, $\sigma = 0$).
   • The macroscopic behavior is determined by which set of states has the higher total configurational entropy ($S^*_c$ vs. $S^*_{fp}$).
   • In the regime $T_g < T < T_0$, $S^*_c > S^*_{fp}$, meaning oscillating states dominate, leading to macroscopic irreversibility ($\sigma > 0$). Conversely, for $T > T_0$, fixed points dominate, and the system appears time-independent on average.

Significance
The paper claims that this framework provides a statistical description of far-from-equilibrium complex landscapes where collective states are defined by their irreversibility (entropy production) rather than just energy or fitness. The work suggests that the notion of a complex landscape with time-dependent states is relevant for systems exhibiting disordered and non-reciprocal interactions. The authors explicitly identify potential applications of this picture in diverse contexts, including neural network dynamics, coupled biological clocks, biological evolution in changing environments, population dynamics with interacting species, glasses under cyclic shear, and the laminar-turbulent transition. The study establishes that entropy production density serves as a fundamental observable to characterize and count these nonequilibrium collective states, offering a tool to distinguish between equilibrium-like glassy behavior and genuine far-from-equilibrium dynamics.