https://arxiv.org/api/7tixN/GQRfeeXRYUsBVyaWHCbB8 2026-06-10T09:41:41Z 81779 165 15 http://arxiv.org/abs/2602.14885v2 2026-06-03T11:34:12Z

Recurrent neural networks (RNNs) provide a theoretical framework for understanding computation in biological neural circuits, yet classical results, such as Hopfield's model of associative memory, rely on symmetric connectivity that restricts network dynamics to gradient-like flows. In contrast, biological networks support rich time-dependent behaviour facilitated by their asymmetry. Here we introduce a general framework, which we term drift-diffusion matching, for training continuous-time RNNs to represent arbitrary, nonlinear stochastic differential equations (SDEs), with given drift and diffusion coefficients, within a low-dimensional latent subspace. Allowing asymmetric connectivity, we show that RNNs can faithfully embed the drift and diffusion of a given SDE, including nonlinear and nonequilibrium dynamics such as chaotic attractors. As an application, we construct RNN realisations of stochastic systems that transiently explore various attractors through both input-driven switching and autonomous transitions driven by nonequilibrium currents, which we interpret as models of associative and sequential (episodic) memory. To elucidate how these dynamics are encoded in the network, we introduce decompositions of the RNN based on its asymmetric connectivity and its time-irreversibility. Our results extend attractor neural network theory beyond equilibrium, showing that asymmetric neural populations can implement a broad class of dynamical computations within low-dimensional manifolds, unifying ideas from associative memory, nonequilibrium statistical mechanics, and neural computation.

2026-02-16T16:15:59Z 25 pages, 16 figures Ramón Nartallo-Kaluarachchi Renaud Lambiotte Alain Goriely http://arxiv.org/abs/2606.04748v1 2026-06-03T11:29:29Z

The low-to-high-density transition in compressed silica glass is investigated using percolation theory. Large-scale molecular dynamics simulations of SiO$_2$ glasses, with system sizes of up to 10$^6$ atoms and pressures ranging from 0 to 35 GPa, were carried out to investigate the emergence of structural motifs and their growth to system-spanning length scales under compression. On this basis, we introduced long-range descriptors that complement conventional local and medium-range structural measures. The results reveal critical percolation transitions of SiO$_Z$-SiO$_Z$ clusters with increasing coordination number $Z$. The critical exponents slightly deviate from the standard (random) correlation, a behavior that seems to be more pronounced for higher coordinated polyhedra than for tetrahedra, suggesting a possible rigidity percolation mechanism. SiSi$_z$-SiSi$_z$ clusters were also analyzed using the non-bonded approach. Bonded and non-bonded approaches complement each other in a particularly illuminating way for describing pressure-induced structural transformations and common mechanisms shared by bonded glasses, such as SiO$_2$, and non-bonded glasses, such as amorphous ice.

2026-06-03T11:29:29Z 15 pages, 11 figures, 2 tables Julien Perradin Simona Ispas Ricardo V. Paredes Anwar Hasmy Bernard Hehlen http://arxiv.org/abs/2604.09768v2 2026-06-03T10:49:27Z

Ordering at arbitrarily high temperature - entropic order - has been argued to take place in a class of generalized Ising models parameterised by a real interaction parameter $p$ when $p\ge 1$. We give a rigorous proof of this conjecture. We further show that on arbitrary graphs, these models solve graph packing problems - crucially, the Maximum Independent Set optimisation problem. Due to the NP-hardness of this packing problem on generic graphs, some lattice systems will exhibit glassy phases. We call this phenomenon $entropic$ $glass$.

2026-04-10T18:00:05Z Main document: 5 pages, 3 figures Suppemental Material: 7 pages, 2 figures. This version has some minor revisions Enrico Andriolo Mendel Nguyen Emily Richards Tin Sulejmanpasic http://arxiv.org/abs/2510.25711v2 2026-06-03T10:20:39Z

We study numerically the enveloping $f$-function of the fluctuation term in eigenstate thermalization hypothesis (ETH) statement. We concentrate on the energy (or entropy) dependence of this function in two-dimensional systems. Our numerical results show that it is, in general, a monotonically increasing function of the entropy. This is in agreement with the general expectation that fluctuations increase with increasing entropy. We show that the $f$-function locally flattens with increasing system-size. The flattening rate is directly proportional to the system size. We also show that the flattening rate is directly proportional to the particle number for systems of same spatial size. This variation of the $f$-function is important for physics at subleading order of the system-size. So, it is relevant for intermediate-size systems (upto a few hundred qubits) which are experimentally accessible. One exception we found is that the $f$-function of the order parameter of a thermal phase transition defy the monotonic behaviour.

2025-10-29T17:20:10Z Major revision emphasizing the numerical nature of the work, otherwise same results as previous version Nilakash Sorokhaibam Anjan Daimari http://arxiv.org/abs/2602.09795v2 2026-06-03T10:02:17Z

Schelling's model of segregation demonstrates that even in the absence of social or governmental interventions, individuals with mild in-group preferences can self-organize into strongly segregated neighborhoods. Many variants of this celebrated model have been proposed by assuming agents tend to increase their satisfaction. Complementary to this traditional, utility-based approach, we model residential moves using satisfaction-independent reaction rates in a spatially extended chemical reaction network. The resulting model exhibits an emergent phenomenon: despite symmetric in-group preferences, the system undergoes a tipping transition at a critical preference level, beyond which one agent type dominates. We characterize this asymmetric phase transition in details using mean-field analysis, numerical simulations and finite size scaling methods. We find that while the transition shares key features with the Ising universality class, such as $\mathbb{Z}_2$ symmetry breaking and similar exponent ratios, the full set of critical exponents does not match any known universality class.

2026-02-10T14:01:50Z v1: 10+6 pages, 7+3 figures. v2: 11+8 pages, 7+3 figures, improved finite-size scaling robustness analysis. Comments welcome Fabio van Dissel Tuan Minh Pham Wout Merbis http://arxiv.org/abs/2606.04651v1 2026-06-03T09:22:27Z

We consider an unsupervised Hebbian network where the pairwise interactions among neurons are built on noisy realizations of hidden ground-truth vectors. Unlike classical Hopfield models, designed as memory devices, this class of networks can be employed to extract latent structure and generalize beyond the "training" set. By combining random matrix theory and replica methods, we derive the asymptotic spectrum of the corresponding interaction matrix and show that the onset of generalization is controlled by a sharp spectral transition. Depending on the quality and the size of the accessible dataset, the spectrum displays either two separated bulks, encoding informative and noisy directions, or a merged single-bulk phase where such distinction is lost. We show that, when coupled with regularization, the emergence of such a spectral split predicts the network's capability to reconstruct the ground-truth vectors from corrupted samples.

2026-06-03T09:22:27Z Elena Agliari Paulo Duarte Mourão Alberto Fachechi Pierpaolo Vivo http://arxiv.org/abs/2606.04533v1 2026-06-03T07:17:37Z

A vibrational paradigm of atomic dynamic in dense fluids is known to provide useful insight on the transport and thermodynamic properties of fluids in three dimensions. In this paper, a vibrational model is generalized to describe the excess entropy of two-dimensional (2D) fluids. A simple practical implementation of this model is demonstrated to deliver accurate results for various systems, such as one-component plasmas with Coulomb and logarithmic interactions, a 2D fluid of dipole particles, and a 2D Yukawa fluid. The applicability limits, relevance to three-dimensional fluids, relations to other 2D phenomena, and potential practical applications are briefly discussed.

2026-06-03T07:17:37Z 11 pages, 7 figures Physical Review E 113, 014104 (2026) Sergey Khrapak 10.1103/14db-xhrv http://arxiv.org/abs/2405.00215v4 2026-06-03T07:13:19Z

We introduce a non-equilibrium version of the Caldeira-Leggett model in which a quantum particle is strongly coupled to a set of engineered reservoirs. The reservoirs are composed by collections of squeezed and displaced thermal modes, in contrast to the standard case in which the modes are assumed to be at equilibrium. The model proves to be very versatile. Strongly displaced/squeezed reservoirs can be used to generate an effective time dependence in the system Hamiltonian and can be identified as sources of pure work. In the case of squeezing, the time dependence is stochastic and breaks the fluctuation-dissipation relation, this can be reconciled with the second law of thermodynamics by correctly accounting for the energy used to generate the initial non-equilibrium conditions. To go beyond the average description and compute the full heat statistics, we treat squeezing and displacement as generalized Hamiltonians on a modified Keldysh contour. As an application of this technique, we show the quantum-classical correspondence between the heat statistics in the non-equilibrium Caldeira-Leggett model and the statistics of a classical Langevin particle under the action of squeezed and displaced colored noises. Finally, we discuss thermodynamic symmetries of the heat generating function, proving a fluctuation theorem for the energy balance and showing that the conservation of energy at the trajectory level emerges in the classical limit.

2024-04-30T21:41:34Z 46 pages, 6 figures. Some typos corrected, acknowledgments added Vasco Cavina Massimiliano Esposito http://arxiv.org/abs/2606.04331v1 2026-06-03T01:15:09Z

The van der Waals interaction has been extensively studied for colloidal forces and resultant emergent phenomena such as colloidal stability, aggregation, and suspension rheology, but the effect of electrolytes on this interaction, especially at intermediate and high electrolyte concentrations, remains incompletely understood. We have extended the Lifshitz theory for van der Waals interactions in pure water to alkali nitrate solutions at arbitrary concentrations by developing a dielectric response model for alkali nitrate solutions that is based on electronic structure calculations of the molecular constituents. Due to their importance in catalysis, ceramics, and coating technologies, the Hamaker constants for rutile, boehmite, and alumina nanoparticles suspended in alkali nitrate solutions are calculated as a function of salt concentration. Contrary to prevailing assumptions, increasing the concentration of sodium (Na), potassium (K), and rubidium (Rb) nitrate solutions causes appreciable increases of the Hamaker constants relative to pure water instead of decreases, whereas cesium nitrate (CsNO3) has almost no effect on the Hamaker constant. We discussed the influence of the solution molar volume, the polarizability of the dissolved ions, and optical properties of the interacting particles in the context of previously published work. Our study indicates a non-vanishing role of van der Waals interactions on colloidal stability at intermediate and high electrolyte concentrations, leading to physical insights on emergent phenomena associated with nanoparticles.

2026-06-03T01:15:09Z Micah P. Prange Jaehun Chun Gregory K. Schenter Elias Nakouzi Yihui Wei Aurora E. Clark Kevin M. Rosso Carolyn I. Pearce http://arxiv.org/abs/2606.05231v1 2026-06-03T00:19:08Z

In open nonequilibrium systems, the thermodynamic entropy of a subsystem is not generally a Lyapunov function. Even during relaxation toward equilibrium, it may decrease temporarily because of exchanges with external reservoirs. This raises a basic question: what thermodynamic quantity, if any, organizes irreversible relaxation in an open system? We address this question using an explicit open-piston model coupled to both a pressure reservoir and a heat bath. The reversible sector is formulated as a Nambu rotational flow generated by the extended energy and the subsystem entropy, while the irreversible sector is written as a gradient flow generated by a dissipation potential $S_{NB}$. In the adiabatic reversible limit, the Nambu bracket produces the oscillatory piston motion on the intersection of conserved level surfaces. After coupling to a heat bath and adding friction, the subsystem entropy $S$ can exhibit nonmonotonic oscillations, whereas $S_{NB}=S-H_{1}/T_{b}$ increases monotonically under the proposed positive-semidefinite dissipative structure. We show that this monotonicity is not a consequence of identifying $S_{NB}$ with thermodynamic entropy. Rather, it follows from two geometric conditions: the reversible Nambu flow preserves $S_{NB}$, and the irreversible dynamics can be written as a positive-semidefinite gradient flow generated by $S_{NB}$. The open-piston model therefore provides a minimal macroscopic realization in which thermodynamic entropy, dissipation potential, reversible temporal order, and irreversible relaxation can be separated explicitly.

2026-06-03T00:19:08Z 30 pages, 9 figures So Katagiri http://arxiv.org/abs/2604.24783v2 2026-06-02T20:53:11Z

We propose a general approach, named by us hyperstatistics, to treat complex systems, in which Boltzmann-Gibbs statistics breaks down in domains of the system. Hyperstatistics preserves the concavity of nonadditive $q$-entropy. We obtain analytical closed-form expressions for the here proposed $q$-generalized Boltzmann factor $B_q$ considering uniform, $γ$, Log-normal, F, and the $q$-$γ$ probability distribution functions. Remarkably, for all investigated distribution functions, $B_q$ reduces to a $q$-exponential-type function. To demonstrate the applicability of hyperstatistics, we use a table top experiment of the discharge of a capacitor considering $γ$-distributed relaxation times, the pressure decay over time associated with the pumping of $^4$He lines of a closed cycle cryostat, midrapidity data for $p$-Pb collisions at the LHC, as well as data set for acceleration distribution in turbulent systems. Furthermore, we deduce the power-law-like dielectric response using the $q$-$γ$-distribution function. Our proposal is applicable to systems with inherent non-Boltzmann-Gibbsian statistics in domains of the system.

2026-04-23T22:30:25Z 23 pages, 5 figures, 1 table. Supplementary material upon request Lucas Squillante Samuel M. Soares Constantino Tsallis Mariano de Souza http://arxiv.org/abs/2605.26083v2 2026-06-02T20:49:44Z

Colloidal gels assembled from nanoparticles (NPs) are a versatile class of soft network-based materials capable of rich dynamic, mechanical, and even optical or magnetic responses to stimuli. Understanding how their hierarchically organized processes relate to macroscopic network properties remains a broad and unresolved problem in soft matter physics. The mechanisms of gel formation can depend sensitively on the pathway and the nature of NP interactions, thus far preventing a unified theoretical bridge between nanoscopic interactions and structural evolution and network dynamics. Indirect measurement of dynamics using light-scattering techniques provides an experimental means to quantify underlying particle and network motion. X-ray photon correlation spectroscopy (XPCS) has emerged as a powerful tool for probing nanoscopic motion in nanoparticle gels, but alone cannot resolve the full spatiotemporal spectrum of dynamics that drive gelation, aging, and network mechanical properties. While in situ rheo-XPCS enables simultaneous probing of nanoscale and bulk mechanical responses, complementary light scattering, microscopy, or simulations can extend spatiotemporal characterization and, consequently, understanding of NP gel network physics. Implementing a modular model platform with tunable primary nanoparticle features allows systematic variation of nanoscopic characteristics that drive emergent gel responses and inform the development of theoretical models for a wide range of soft, dynamic, nanostructured materials. The rapid expansion of XPCS capabilities at fourth-generation light sources, combined with complementary tools and robust model systems, positions the field to move beyond descriptive fundamental studies toward the design of nanoparticle gels with adaptive and programmable behaviors.

2026-05-25T17:47:52Z 77 pages, 7 figures, Perspective for Physical Review E (accepted for publication) William D. Brackett Zachary M. Sherman Felix Lehmkühler Thomas M. Truskett Delia J. Milliron 10.1103/xn5z-cwjz http://arxiv.org/abs/2606.04085v1 2026-06-02T18:00:01Z

Wilsonian renormalization is usually formulated in momentum space, but on curved backgrounds momentum shells have no invariant meaning. We replace them by an intrinsic spectral cutoff, ordering modes by the covariant Laplacian and setting the cutoff resolution by the system size in renormalization group (RG) units. For a scalar field on $S^3$, this yields a covariant, momentum-free RG flow whose trace is an exact sum over spherical harmonics. The standard flat-space flow is recovered when the sphere is large compared with the coarse-graining scale. As a nontrivial test, the compact spectral flow realizes Wilson-Fisher universality without momentum shells: the interacting fixed point survives at finite resolution, has one relevant direction, and approaches its flat-space counterpart smoothly, with critical exponents only weakly affected by the compact spectrum.

2026-06-02T18:00:01Z 4 pages, 1 figure Alfio M. Bonanno Renata Ferrero Giovanni Oglialoro http://arxiv.org/abs/2606.03956v1 2026-06-02T17:45:55Z

We investigate operator spreading in random quantum circuits with gates drawn from orthogonal-invariant or symplectic-invariant ensembles, revealing several key distinctions from the well-studied unitary-invariant case. We find that the ensemble-averaged Pauli-string weights relax to a ternary-valued structure, instead of the binary structure of unitary-invariant circuits. For orthogonal- or symplectic-invariant circuits, the domain wall separating trivial and scrambled regions has a finite width even for Haar-random gates, whereas domain walls are sharp for Haar-distributed random unitary circuits. We further find a fundamental dichotomy between random circuits with two-qubit gates from the two disconnected components of the orthogonal group: While the butterfly velocity for the special orthogonal ensemble lies between zero and the Haar value, the negative-determinant sector exhibits a non-zero lower bound for any gate distribution. Moreover, for qudit size $q=2$, the butterfly velocity can exceed that of the Haar-random ensemble.

2026-06-02T17:45:55Z 33 pages, 8 figures Zhiyang Tan Piet W. Brouwer http://arxiv.org/abs/2606.03848v1 2026-06-02T16:27:04Z

We establish a reverse-time denoising theory for quantum diffusions of continuously measured quantum systems. Starting from the stochastic Schrödinger equation of a forward noising dynamics, we derive the exact reverse-time dynamics for quantum trajectories, whose law coincides with the time-reversal of the original process. We prove that the denoising dynamics is a physically admissible quantum diffusion, with the same measurement-induced noise but a state-dependent feedback Hamiltonian, a direct analogue of the "score function" of generative classical diffusion models. This provides a principled framework for converting samples of a simple distribution into those of a more complex ensemble of quantum states. We show how the denoising dynamics can be directly learnt from forward trajectory data, and how to exploit purification to initialise the denoising process.

2026-06-02T16:27:04Z Maël Bompais Mădălin Guţă Juan P. Garrahan