Chaotic Reorganization of Rugged Energy Landscapes

21 concepts12 papers

Imagine trying to find the lowest point in a vast mountain range during a heavy fog. A natural strategy is to walk downhill. If you get stuck in a shallow valley, you might shake the system—add a little thermal energy—to hop over small ridges, and then slowly lower the temperature to settle into the true global minimum. This intuition underpins simulated annealing and optimization across physics and computer science.

But what if lowering the temperature doesn't just reduce your ability to hop over walls? What if changing the temperature completely redraws the map?

In complex, disordered systems like spin glasses, the free-energy landscape undergoes a radical, unpredictable reshuffling with even the slightest thermal fluctuation. A valley that was a deep, stable global minimum at one temperature can abruptly become a high-energy, irrelevant plateau at another. This phenomenon is known as temperature chaos.

The Mechanism: Entropy Fluctuations and Level Crossings

To understand why the landscape morphs, we must look at what defines a "valley" in a thermal system. We are not just minimizing pure potential energy $E$; we are minimizing free energy:

$$F = E - TS$$

where $T$ is temperature and $S$ is entropy.

In a simple system, entropy changes smoothly. But in a disordered system, every local minimum represents a completely different structural configuration with its own unique, highly irregular arrangement of microstates. Olivier Martin demonstrated via the Random-Energy Random-Entropy (Re-Rem) model that state-to-state entropy fluctuations drive temperature chaos. Crucially, these fluctuations scale positively with the system size $N$. When a system is large, a tiny shift in $T$ multiplies across all degrees of freedom, causing the $TS$ term to swing wildly from one state to another.

Free Energy (F)
  ^
  |    / \       State A (Dominant at high T)
  |   /   \     
  |  /     \     State B (Dominant at low T)
  | /       \   
  |/         \___________ Level Crossing
  +----------------------------------------> Decreasing Temperature (T)

This reshuffling does not happen as a gradual, continuous melting of hills and valleys. Instead, L. A. Fernández et al. showed that temperature chaos is driven by rare, sudden events akin to quantum level crossings. As temperature shifts, the free-energy lines of distinct states cross. An isolated, previously sub-dominant valley suddenly and discontinuously becomes the new global minimum, completely upending the equilibrium configuration of the system.

The Breakdown of Adiabatic Tracking

This chaotic reorganization shatters a fundamental assumption in non-equilibrium physics: that we can track a state continuously as it cools.

Using a mathematical framework called the "state-following" formalism, Florent Krzakala (see also, /paper/art_98b42cd5192e4033afb8b655eab5d44d) proved that adiabatic evolution frequently fails in rugged landscapes. If you locate an optimal Gibbs state at a higher temperature and try to follow it as you cool the system, that state will often become highly suboptimal or vanish entirely.

Because the landscape is constantly scrambling itself, the final state of a system depends entirely on its exact path through time and temperature:

  • Slow Cooling vs. Quenching: As Federico Ricci‐Tersenghi demonstrated, a slow cooling schedule allows a system to explore and settle into deeper, more stable energy layers than a sudden thermal quench (such as the rapid protocols studied by Chiara Cammarota et al. and Mark S. Conradi et al.).
  • Structural Shattering: Without a continuous manifold of near-optimal states to buffer these shifts—a property known as marginal stability, analyzed mathematically by Mark Sellke—the landscape shatters into isolated, trapped minima, making the ground state highly elusive.

Macroscopic Proof: Rejuvenation and Memory

For decades, temperature chaos was a compelling theoretical construct that eluded clear simulation-based validation because it requires massive system sizes to manifest. Recently, researchers bypassed these limits using dedicated hardware like the Janus supercomputer to observe how this microscopic chaos manifests macroscopically through two bizarre behaviors: rejuvenation and memory.

Imagine aging a spin glass at a temperature $T_1$. The system slowly relaxes into a local minimum, "aging" as its correlations grow.

If you suddenly drop the temperature to $T_2$, the system undergoes rejuvenation: it acts as if it is completely young again, restarting its relaxation process from scratch. Because temperature chaos has reorganized the landscape, the configuration achieved at $T_1$ is completely irrelevant to the new $T_2$ landscape.

Temperature Profile:         Aging at T1 -------> Drop to T2 (Rejuvenation) -------> Return to T1 (Memory)
System State:                [State 1 Grows]      [State 2 Grows; State 1 Frozen]    [State 1 Resumes]

However, if you heat the system back up to $T_1$, it exhibits memory: it instantly recalls its old aged state, picking up right where it left off.

Enrico Calore et al. demonstrated that this dual behavior is governed by the interplay of multiple microscopic length scales. When the temperature shifts, a new, independent correlated state begins to grow at the new temperature, while the old configuration remains dynamically frozen in the background. As I. Paga et al. showed by quantifying these memory coefficients, the system does not lose its past information; rather, temperature chaos temporarily uncorrelates the active landscape from the frozen past state.

Beyond Physics: Disorder Chaos

The mathematics of this extreme sensitivity are not confined to temperature. Any complex system with a high degree of frustration and many competing variables can exhibit an analogous phenomenon called disorder chaos.

Michael Benzaquen et al. showed that this exact structural instability occurs when optimizing financial portfolios or modeling ecological equilibria. A minute change in asset correlations or species interaction strengths completely reshuffles the optimal solution. Temperature chaos is a physical manifestation of a broader mathematical truth: in rugged landscapes, the optimal configuration is fundamentally unstable to minor changes in environmental parameters.

Key Foundational Work

Chaotic temperature dependence in a model of spin glasses (2002) — contributed 2 concept elements to this direction.
Cooling-schedule dependence of the dynamics of mean-field glasses (2004) — contributed 1 concept elements to this direction.
History & citationVersion 2 · 2026-10-04

Current · Permanent link

  1. Version 2 (latest) · 2026-10-04 · 12 papers
  2. Version 1 · 2026-10-04 · 22 papers
Cite this direction

Latest version (a living page):

@misc{lacuna_temperature_chaos_5f96b0a7,
  title = {temperature chaos},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/temperature-chaos/rdr_5f96b0a7e0e24a9e9230f7dd43e34e08},
  note = {Latest version of a living research direction; cite a version for a fixed text. Accessed 2026-11-19},
}

This exact version (version 2, frozen text):

@misc{lacuna_temperature_chaos_5f96b0a7_v2,
  title = {temperature chaos},
  author = {{Lacuna}},
  howpublished = {Lacuna research direction},
  year = {2026},
  url = {https://lacuna.tiptreesystems.com/direction/temperature-chaos/rdr_5f96b0a7e0e24a9e9230f7dd43e34e08/v/2},
  note = {Version 2, frozen 2026-10-04. Accessed 2026-11-19},
}

See something wrong? Report a mistake