Kinetic simulated annealing optimization with entropy-based cooling rate

2026-02-17T12:43:55Z

We present a modified simulated annealing method with a dynamical choice of the cooling temperature. The latter is determined via a closed-loop control and is proven to yield exponential decay of the entropy of the particle system. The analysis is carried out through kinetic equations for interacting particle systems describing the simulated annealing method in an extended phase space. Decay estimates are derived under the quasi-invariant scaling of the resulting system of Boltzmann-type equations to assess the consistency with their mean-field limit. Numerical results are provided to illustrate and support the theoretical findings.

Phase Transitions in Neural Networks Pruning

2026-02-16T22:07:50Z

Deep neural networks are strongly over-parameterized, often containing far more weights than required for their task. Although such redundancy can aid optimization, it leads to inefficient deployment and high computational cost, motivating model compression techniques. Among these, network pruning provides a clear and effective route to sparsity. We study pruning from a statistical-physics perspective, interpreting performance degradation under weight removal as a phase transition. Focusing on magnitude-based pruning with fine-tuning, we show that deep networks undergo a sharp transition from a cooperative, functional phase to a disordered phase with collapsed performance. This transition is characterized by scaling laws consistent with second-order critical behavior, with connectivity as the control parameter. Our findings suggest universal pruning-induced criticality across architectures and datasets. Finally, we show that there exists a large class of subnetworks sharing the same nodes' degrees with similar learning ability, thus linking model performance to its topological properties.

Finding the Edge of Chaos in a Ferromagnet: Quantifying the "Complexity" of 2D Ising Phase Transitions with Image Compression

2026-02-16T20:45:15Z

The data-driven characterization of the ``complexity'' present in dynamical systems remains an open problem with broad applications across the physical sciences. We investigate the ``structural complexity'' of the 2D ferromagnetic Ising model, a paradigmatic system exhibiting a second-order phase transition at a certain critical temperature which is often cited as a canonical example of complex morphology. We define a quantitative metric for this structural complexity, $\mathcal C_s$, through the lens of algorithmic information theory by approximating the Kolmogorov complexity of lattice configurations via standard lossless image compression algorithms. We regularize our proposed metric, $\mathcal C_s$, by comparing the compressibility of a configuration to that of its pixel-wise sorted and randomly shuffled counterparts. We arrive at a definition of $\mathcal C_s$ as a product of two components representing the systems departure from perfect order and disorder respectively which we then plot as a function of temperature. Our numerical simulations reveal a distinct peak in $\mathcal C_s$ at the known critical temperature $T_c$. This result demonstrates that such information-theoretic measures can act as sensitive, model-agnostic indicators of criticality, directly quantifying the emergence of complex structure at the boundary between order and chaos, opening the door to data-driven applications in domains where analytic solutions are unavailable.

Memorisation and forgetting in a learning Hopfield neural network: bifurcation mechanisms, attractors and basins

2026-02-15T19:10:32Z

Despite explosive expansion of artificial intelligence based on artificial neural networks (ANNs), these are employed as "black boxes'', as it is unclear how, during learning, they form memories or develop unwanted features, including spurious memories and catastrophic forgetting. Much research is available on isolated aspects of learning ANNs, but due to their high dimensionality and non-linearity, their comprehensive analysis remains a challenge. In ANNs, knowledge is thought to reside in connection weights or in attractor basins, but these two paradigms are not linked explicitly. Here we comprehensively analyse mechanisms of memory formation in an 81-neuron Hopfield network undergoing Hebbian learning by revealing bifurcations leading to formation and destruction of attractors and their basin boundaries. We show that, by affecting evolution of connection weights, the applied stimuli induce a pitchfork and then a cascade of saddle-node bifurcations creating new attractors with their basins that can code true or spurious memories, and an abrupt disappearance of old memories (catastrophic forgetting). With successful learning, new categories are represented by the basins of newly born point attractors, and their boundaries by the stable manifolds of new saddles. With this, memorisation and forgetting represent two manifestations of the same mechanism. Our strategy to analyse high-dimensional learning ANNs is universal and applicable to recurrent ANNs of any form. The demonstrated mechanisms of memory formation and of catastrophic forgetting shed light on the operation of a wider class of recurrent ANNs and could aid the development of approaches to mitigate their flaws.

Wavefront propagation in a bistable dual-delayed-feedback oscillator: analogy to networks with nonlocal interactions

2026-02-13T17:34:15Z

In the present research, a bistable delayed-feedback oscillator with two delayed-feedback loops is shown to replicate a network of bistable nodes with nonlocal coupling. It is demonstrated that certain aspects of the nonlocal interaction impact on wavefront propagation identified in networks of bistable elements are entirely reproduced in the dynamics of a single oscillator with two delays. In particular, adding the second delayed-feedback loop allows speeding up both deterministic and stochastic wavefront propagation, achieving stabilization of propagating fronts at lower noise intensity and preventing fronts from noise-induced destruction occurring in the presence of single delayed-feedback. All the revealed effects are studied in numerical simulations and confirmed in physical experiments, showing an excellent correspondence.

Topological Reorganization and Coordination-Controlled Crossover in Synchronization Onset on Regular Lattices

2026-02-13T16:24:34Z

The transition to global synchronization in coupled dynamical systems is governed by the interplay between coupling strength and structural topology. Although abrupt, first-order-like synchronization transitions have been extensively reported in heterogeneous networks, it is unclear whether comparable accelerated onset behavior can emerge purely from coordination geometry in spatially homogeneous, regular lattices. In this study, we investigate large-scale ($N=10^5$) stochastic Stuart-Landau oscillator networks defined on regular lattices with controlled coordination number. Using topological data analysis (TDA), simplicial-complex characterization, and optimal-transport-based geometric diagnostics, we identify a coordination-controlled crossover in synchronization onset dynamics at approximately $z_{c} \approx 7$ within the class of regular lattices considered. Low-coordination lattices ($z < z_{c}$) exhibit persistent $H_2$ topological features in the dynamical amplitude field that correlate with delayed coherence and surface-limited propagation. In contrast, higher-coordination lattices ($z > z_{c}$) display rapid fragmentation of these features, reduced interface roughness, and predominantly positive Ricci curvature. This is consistent with enhanced path redundancy and improved transport efficiency. In this regime, the global order parameter exhibits accelerated exponential-like growth during the onset stage. Throughout this work, abrupt synchronization refers specifically to this exponential onset behavior rather than to thermodynamic first-order hysteresis. Our results demonstrate that increasing coordination density induces a qualitative reorganization of higher-order topological structure that strongly correlates with synchronization efficiency in regular lattice systems.

Inferring Coupled Stuart-Landau Equations from Waveforms

2026-02-13T15:37:30Z

We present a data-driven framework to infer phase-amplitude equations of coupled limit-cycle oscillators directly from waveform measurements. Exploiting the universality of the Stuart-Landau normal form near a supercritical Hopf bifurcation, we reconstruct a near-identity transformation from two independent observables of an isolated oscillator and infer the intrinsic Stuart-Landau parameters. Using this reconstructed transformation, we then estimate linear coupling coefficients from paired measurements. The method accurately recovers parameters for coupled van der Pol oscillators, providing a quantitative benchmark. Applied to a high-dimensional hydrodynamic system of two coupled collapsible-channel oscillators, the inferred Stuart-Landau model captures bistability between in-phase and anti-phase synchronization and reveals that the anti-phase state is destabilized through a Neimark-Sacker bifurcation. Our approach enables quantitative prediction of synchronization transitions involving amplitude dynamics from experimentally accessible waveform data.

Modified Axelrod Model Showing Opinion Convergence And Polarization In Realistic Scale-Free Networks

2026-02-12T18:44:08Z

Axelrod model is an opinion dynamics model such that each agent on a square lattice has a finite number of possible nominal opinions on a finite number of issues that are usually called features in the field. Moreover, its dynamics between two agents is assimilative in the sense that the number of agreeing features between them never decreases upon interaction. Here we modify this model to study opinion convergence, polarization and more importantly to find ways to reduce opinion polarization in an already polarized population. We do so by changing or adding several elements from complex network and continuous opinion dynamics research. First, we put agents in a scale-free network. Second, we adopt the bounded confidence model by representing our agent's opinions by numbers in $[-1,1]$ those distances follow the standard Euclidean metric. Third, our rules allow both convergence and divergence of their resultant opinions after a pair of agents interacts. As a result, our modified model offers a more comprehensive exploration of opinion dynamics. Computer simulation results of our model show scaling behavior and a notable trend in opinion polarization on all features in the majority of reasonable simulation parameters. To mitigate this polarization, we introduce empathetic agents that work actively to reduce opinion differences. However, our findings indicate limited success in the approach for the most effective way is to change the behavior of a significant portion of highly connected agents. This research contributes to the understanding of opinion dynamics within society and highlights the nuanced complexities that arise when considering factors such as network structure and continuous opinion values. Our results prompt further exploration and open avenues for future investigations into effective methods of reducing opinion polarization.

Scaling Laws for Function Diversity and Specialization Across Socioeconomic and Biological Complex Systems

2026-02-12T18:31:53Z

Function diversity, the range of tasks individuals perform, and specialization, the distribution of function abundances, are fundamental to complex adaptive systems. In the absence of overarching principles, these properties have appeared domain-specific. Here, we introduce an empirical framework and a mathematical model for the diversification and specialization of functions across disparate systems, including bacteria, federal agencies, universities, corporations, and cities. We find that the number of functions grows sublinearly with system size, with exponents from 0.35 to 0.57, consistent with Heaps' Law. In contrast, cities exhibit logarithmic scaling. To explain these empirical findings, we generalize the Yule-Simon model by introducing two key parameters: a diversification parameter that characterizes how existing functions inhibit the creation of new ones, and a specialization parameter that describes how a function's attractiveness depends on its abundance. Our model enables cross-system comparisons, from microorganisms to metropolitan areas. The analysis suggests that what drives the creation of new functions depends on the system's goals and structure: federal agencies tend to ensure comprehensive coverage of necessary functions; cities tend to slow the creation of new occupations as existing ones expand; and cells occupy an intermediate position. Once functions are introduced, their growth follows a remarkably universal pattern across all systems.

Information bounds production in replicator systems

2026-02-12T12:28:02Z

Environmental fluctuations can shape replicator dynamics, with important consequences for both prebiotic and modern ecosystems. However, it remains unclear how simple replicators can acquire and use information about fluctuating environments, given that such information processing is often assumed to require sophisticated mechanisms for sensing and control. Here, we show that even simple replicator networks can increase productivity by exploiting environmental information in a functional way. Using a model of autocatalytic replicators in a flow reactor, we derive an information-theoretic decomposition of productivity, with separate contributions from environmental uncertainty, side information, and distribution mismatch. We derive optimal strategies and universal bounds on the benefit of information and compare our findings with existing work, including ``Kelly gambling'' in information theory. By applying our framework to a model of real-world molecular replicators, we demonstrate the benefits of internal memory and propose an experimental setup for detecting functional information in a minimal chemical system.

Kinetic renormalization of auroral turbulence

2026-02-11T12:07:13Z

Driven-dissipative systems often exhibit self-organization in the form of coherent dissipative structures. However, observing such critical states in natural plasmas remains elusive, leading to the traditional view that the fine structure of Earth's auroral ionosphere is shaped by local turbulent flows. Here we report the discovery of a self-organizing regime in Earth's ionosphere. We identify this by modeling the sum of saturation electric fields in the turbulent auroral electrojets as a stochastic variable that renormalizes into noise-enabled transport, via explicitly derived Bohm diffusion. This constitutes an effective field-theory for Farley-Buneman turbulence in the Martin-Siggia-Rose formalism for renormalization group theory, for which we provide strong empirical evidence. Using a composite radar-GPS power spectrum of plasma turbulence, we resolve a scale-invariant cascade that exhibits a characteristic kinetic Alfvén $k^{-8/3}$-signature across four orders of magnitude in $k$. What is more, a large statistical analysis of how the turbulence responds to magnetospheric driving reveals a clear tendency for the observed number density of turbulent waves to scale linearly with driving power, matching the predictions made by our field theory's overdamped equations of motion, which offer closed-form calculations of macroscopic transport relations that are uniquely suitable for sub-grid parameterization in space weather modeling. This establishes geospace storms as opportunities to observe non-equilibrium phase transitions imposing global constraints on collision-dominated systems.

Defect states as a precursor of the chimera states in a ring of non-locally coupled oscillators

2026-02-11T05:03:19Z

We investigate the transition from synchronized to chimera states in a ring of non-locally coupled phase oscillators. Our focus is on the intermediate defect states, where solitary waves in the phase gradient profile travel at a constant speed. These traveling defects serve as a dynamical precursor for the nucleation of chimera clusters. The fraction of samples exhibiting defect states increases with the phase delay $α$ and peaks at $α_{c}$, where the system crosses over to asynchronous states filled with chimera clusters. While the traveling speed, number, and width of these defects increase with $α$, the total spatial extent of the defects remains robust against the system size $N$. These results shed new light on the emergence of chimera states in frustrated coupled oscillators.

Time delay in the 1d swarmalator model

2026-02-08T23:16:02Z

We study the 1d swarmalator model augmented with time delayed coupling. Along with the familiar sync, async, and phase wave states, we find a family of unsteady states where the order parameters are time periodic, sometimes with clean oscillations, sometimes with irregular vacillations. The unsteady states are born in two ways: via a Hopf bifurcation from the phase wave, and a zero eigenvalue bifurcation from the async state. We find both of these boundary curves analytically. A surprising result is that stabilities of the async and sync states are independent of the delay τ; they depend only on the coupling strength.

Extreme vulnerability to intruder attacks destabilizes network dynamics

2026-02-08T16:26:48Z

Consensus, synchronization, formation control, and power grid balance are all examples of virtuous dynamical states that may arise in networks. Here, we focus on how such states can be destabilized from a fundamental perspective; namely, we address the question of how one or a few intruder agents within an otherwise functioning network may compromise its dynamics. We show that a single adversarial node coupled via adversarial connections to one or more other nodes is sufficient to destabilize the entire network, which we prove to be more efficient than targeting multiple nodes. Then, we show that concentrating the attack on a single low-degree node induces the greatest instability, challenging the common assumption that hubs are the most critical nodes. This leads to a new characterization of the vulnerability of a node, which contrasts with previous work, and identifies low-indegree nodes (as opposed to the hubs) as the most vulnerable components of a network. Our results are derived for linear systems but hold true for nonlinear networks, including those described by the Kuramoto model. Finally, we derive scaling laws showing that larger networks are less susceptible, on average, to single-node attacks. Overall, these findings highlight an intrinsic vulnerability of technological systems, such as autonomous networks, sensor networks, power grids, and the Internet of Things, with implications that extend also to the realm of complex social and biological networks.

The effect of role-based resource allocation on epidemic dynamics

2026-02-08T02:13:16Z

We propose a coupled dynamical model of resource allocation and epidemic spread, inspired by the hierarchical structure of real-world therapeutic resource allocation. In this framework, network nodes are assigned distinct roles as either resource allocators or resource recipients. As the average number of links per recipient from allocators increases, the prevalence exhibits one of four distinct response patterns across conditions: monotonically increasing, monotonically decreasing, U-shaped trend, or a sudden decrease with large fluctuations. A mechanistic analysis uncovers three central insights: (i) a trade-off between efficient resource allocation and infection risk faced by allocators, (ii) the critical need to avoid resource redundancy when therapeutic efficiency is high, and (iii) the emergence of cascade-induced bistability in the coupled system.