🔬 condensed matter

Measurement-Induced Phase Transitions in Informational Active Matter

This paper proposes a theoretical framework for informational active matter where adaptive particles achieve collective behaviors like flocking through local measurements and decision-making protocols, leading to measurement-induced phase transitions and "informational activity" that compresses phase space without performing work.

Original authors: Bryan VanSaders, Michel Fruchart, Vincenzo Vitelli

Published 2026-04-16

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Bryan VanSaders, Michel Fruchart, Vincenzo Vitelli

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a crowded dance floor where everyone is bumping into each other randomly, spinning in chaotic directions. This is how most gases behave: a jumble of particles bouncing off one another due to thermal energy (heat). Usually, this chaos is impossible to organize without an external force, like a DJ shouting "Everyone face North!" or a giant hand pushing the crowd.

But what if the dancers themselves could decide, based on who they see around them, whether to shrink or grow? And what if, by doing so, they could somehow turn that random chaos into a synchronized dance, all without anyone pushing them?

That is the core idea of this paper: Informational Active Matter.

Here is the story of the "Demon Gas," explained through simple analogies.

1. The Smart Dancers (The Maxwell Demons)

In physics, there is a famous thought experiment called "Maxwell's Demon." Imagine a tiny, invisible creature that watches gas molecules. If a fast molecule approaches a door, the demon opens it; if a slow one approaches, it keeps it closed. Over time, one side gets hot and the other cold, creating energy out of thin air (which seems to break the laws of physics, but actually uses information).

This paper takes that idea and scales it up. Instead of one demon, imagine thousands of particles that are all "smart."

The Rule: Every few seconds, each particle looks around.
The Decision: If it sees its neighbors moving in the same direction, it shrinks to be small and slippery. If it sees neighbors moving in a different direction, it grows big and clunky.
The Result: By changing its size, the particle changes how often it bumps into others. A small particle slips through crowds easily; a big particle gets stuck and bounces off.

2. The "Information Engine"

Usually, to make things move, you need fuel (like gas in a car or food for a bird). This system doesn't use fuel. It uses information.

Think of it like a game of "Red Light, Green Light" played by the particles themselves.

They don't push each other.
They don't have engines.
They simply measure their environment and choose a size.

By shrinking when they are "in sync" and growing when they are "out of sync," they subtly bias the random collisions. It's like a crowd of people in a hallway. If everyone who is walking the right way shrinks to a tiny mouse and slips past, while everyone walking the wrong way grows into a giant and gets blocked, the hallway eventually fills with people walking the right way.

The paper calls this an "Informational Engine." The "fuel" is the data they gather about their neighbors.

3. The "Flocking" Phase Transition

The most exciting discovery is that this leads to a sudden change in behavior, called a Phase Transition.

Before: The particles are a chaotic mess, bouncing everywhere.
After: Suddenly, they all start moving in the same direction (or opposite directions, like a nematic liquid crystal).

The authors compare this to a quantum measurement-induced phase transition, but in a classical world. In quantum physics, looking at a system can change its state. Here, the act of the particles "measuring" their neighbors and making a decision forces the whole group to organize.

The paper shows that the more information the particles gather, the stronger the organized movement becomes. There is a direct limit: you can't get more order than the amount of information you measured.

4. Noise is Your Friend

Here is the counter-intuitive part: Chaos helps them.

In most engineering, noise (vibration, turbulence, static) is bad. It ruins precision. But for these "smart particles," noise is the engine.

The particles need the random bumps (noise) to move.
They use their measurements to "rectify" (straighten out) that noise.
The paper shows that if you shake the system more (add more noise), the particles actually form clearer, sharper patterns faster.

The Analogy: Imagine trying to sort a pile of mixed red and blue marbles.

Normal way: You pick them up one by one (slow, requires energy).
This way: You put the marbles in a shaking box. The red ones are smart; when they feel a bump, they shrink and roll to the left. The blue ones grow and get stuck on the right. The more you shake the box (the more noise), the faster they sort themselves out.

5. Why Does This Matter?

The authors suggest this could change how we build the future:

Microrobots: Imagine tiny robots swimming in your bloodstream or in a river. Instead of carrying heavy batteries, they could use the random turbulence of the water to move, guided by simple sensors that tell them when to "shrink" or "grow."
Self-Healing Materials: Materials that can sense stress and reorganize themselves without external power.
Understanding Nature: This might explain how cells, bacteria, or bird flocks organize themselves. They might not be following complex force laws, but rather simple "decision protocols" based on local information.

The Bottom Line

This paper reveals a new way to create order from chaos. It shows that information is a physical force. By simply measuring the world and making tiny, smart choices, a group of random particles can spontaneously organize into a coordinated, moving system, turning the "noise" of the universe into a useful, directed motion.

It's the ultimate proof that knowledge is power—even for a gas particle.

1. Problem Statement

Traditional active matter systems are typically modeled as collections of particles exerting forces (enthalpic activity) to drive themselves out of equilibrium. However, many biological and synthetic systems (e.g., cells, ant colonies, programmable colloids) operate as information engines. These systems rectify environmental noise not by applying external forces, but by performing local measurements and making control decisions (similar to Maxwell's Demon).

The central challenge addressed in this paper is the lack of a theoretical framework for many-body informational active matter. Standard coarse-graining approaches rely on force laws, but these systems are better described by decision-making protocols. The authors ask: How do microscopic decision-making protocols, based on measurements, generate macroscopic active states (like flocking) without performing mechanical work? Can these systems undergo phase transitions driven purely by information acquisition?

2. Methodology

The authors introduce a minimal theoretical model called the "Demon Gas" and analyze it using a multi-scale approach combining information theory, kinetic theory, and hydrodynamics.

The Model (Demon Gas):
- A system of hard disks at finite temperature undergoing random collisions.
- The "Demon" Mechanism: Each particle can change its diameter (and thus its scattering cross-section) based on local measurements of its velocity and the state of its neighbors.
- Constraints: The model operates in a limit where particles do not perform work on each other. Size changes occur only between collisions to avoid overlaps, ensuring conservation of mass, linear momentum, and total kinetic energy.
- Breaking Detailed Balance: Although energy and momentum are conserved, the coupling between particle size and velocity breaks time-reversal symmetry (detailed balance) because the scattering cross-section depends on the state, preventing the "undoing" of collisions.
Analytical Tools:
- Information Thermodynamics: Used to quantify the non-equilibrium free energy generated by measurements and the mutual information ( $I$ ) between the system state and measurement outcomes.
- Kinetic Theory: The authors derive a modified Boltzmann equation with a collision operator that includes a phase-space compression factor ( $\alpha \neq 1$ ) representing the bias introduced by the diameter choice function.
- Hydrodynamic Theory: A coarse-grained description is derived for the nematic order parameter, treating the system as a fluid with an active stress tensor.
- Numerical Simulations: Molecular Dynamics (MD) simulations using HOOMD-Blue were performed to validate theoretical predictions, including systems with non-thermal noise (vibration, turbulence) and reinforcement learning (RL) agents.

3. Key Contributions

A. Informational Flocking as a Phase Transition

The authors demonstrate that particles can achieve nematic flocking (alignment of velocities modulo reflection) solely through local decision-making.

Mechanism: Particles reduce their diameter when aligned with neighbors (increasing collision probability with misaligned particles) and increase it when misaligned. This selectively biases scattering events to promote alignment.
Result: The system spontaneously breaks rotational symmetry, transitioning from a disordered gas to an ordered flocking state.
Phase Transition: The onset of this order is identified as a Measurement-Induced Phase Transition (MIPT). The magnitude of the order parameter ( $Q$ ) is strictly bounded by the amount of information ( $I$ ) measured:
$\left(\frac{Q_0}{P_0}\right)^2 \leq \frac{32}{\pi^2} I$
This parallels quantum MIPTs where measurements counteract information scrambling, but here it occurs in a classical, many-body system driven by feedback control.

B. Thermodynamics of Information-Driven Fluids

The paper establishes the thermodynamic cost and gain of these systems:

Free Energy Injection: Measurements inject non-equilibrium free energy ( $\Delta F$ ) into the system. This energy is bounded by the mutual information ( $k_B T I$ ).
Entropy Production: The active compression of phase space drives entropy production. The rate of entropy production is a monotonic function of the excess free energy.
Zero-Work Activity: Crucially, the system achieves macroscopic order and can perform work on external loads (e.g., inflating a balloon asymmetrically) without the particles exerting direct mechanical forces on one another. The "fuel" is the information gained.

C. Exploitation of Non-Thermal Noise

The authors extend the model to non-thermal environments (e.g., turbulent flows, vibrated granular beds).

Noise-Independent Patterning: Unlike self-propelled particles whose patterns degrade with increased noise (effective temperature), informational agents improve their pattern resolution with increased noise.
Mechanism: The agents convert the magnitude of environmental fluctuations into directed motion. The effective potential depth scales linearly with temperature, perfectly compensating for the noise. This allows for robust pattern formation even in highly chaotic environments.

D. Reinforcement Learning Integration

The model was extended to agents using Reinforcement Learning (RL) to optimize diameter functions in complex environments. The RL agents successfully learned policies that maximized propulsion and pattern formation, confirming that the "demon" strategy is robust and learnable.

4. Key Results

Nematic Order Parameter Bound: The steady-state nematic order parameter is fundamentally limited by the information measured. As measurement frequency increases, the system moves further from equilibrium, but the order is capped by the information rate.
Phase Space Compression: The kinetic theory reveals that the feedback loop acts as a compressible flow in phase space ( $\alpha \neq 1$ ), which is the mathematical signature of the active state.
Temperature Independence: In the long-time limit, the resolution of patterns formed by demon gases is independent of the noise level (temperature), whereas self-propelled particles degrade. The time to reach a specific resolution scales as $1/\sqrt{T_{eff}}$ .
Work Extraction: The system can perform work (e.g., inflating a balloon into an ellipsoid) by rectifying thermal or non-thermal fluctuations. The work output scales with the magnitude of environmental noise, distinguishing it from conventional chemical fuel-driven engines.

5. Significance

Theoretical Unification: The paper bridges the gap between information theory, statistical mechanics, and active matter. It provides a rigorous framework for systems where "decision-making" replaces "force generation" as the primary driver of non-equilibrium behavior.
New Class of Active Matter: It defines Informational Active Matter, a category distinct from traditional active matter, where the energy source is the environment's fluctuations and the control mechanism is information processing.
Biological and Synthetic Applications:
- Biology: Offers a mechanism for understanding how biological systems (e.g., cell tissues, bacterial swarms) maintain order and function in noisy, fluctuating environments without expending excessive metabolic energy.
- Synthetic Systems: Provides design principles for microrobots and programmable colloids. It suggests that future swarms could be guided by reinforcement learning to exploit turbulence or vibration for navigation and pattern formation, rather than fighting against it.
Measurement-Induced Transitions: It establishes a classical analog to quantum measurement-induced phase transitions, suggesting that the interplay between measurement and dynamics is a universal mechanism for generating order in complex systems.

In summary, this work demonstrates that information is a physical resource capable of driving macroscopic phase transitions and sustaining non-equilibrium states in many-body systems, provided the agents can measure their environment and adjust their interaction cross-sections accordingly.