Physica D: Nonlinear Phenomena最新文献

An Excitable Network Attractor (ENA) is a forward-invariant set in phase space that can be used to explain input-driven behaviour of Recurrent Neural Networks (RNNs) trained on tasks involving switching between a discrete set of states. An ENA is composed of two or more attractors and excitable connections that allow transitions from one attractor to another under some input perturbation. The smallest such perturbation that makes a connection between two attractors is called the excitability threshold associated with that connection. The excitability threshold provides a measure of sensitivity of the connection to input perturbations. Errors in performance of such trained RNNs can be related to errors in transitions around the associated ENA. Previous work has demonstrated that ENAs of arbitrary sensitivity and structure can be realised in a RNN by suitable choice of connection weights and nonlinear activation function. In this paper we show that ENAs of arbitrary sensitivity and structure can be realised even using a suitable fixed nonlinear activation function, i.e. by suitable choice of weights only. We show that there is a choice of weights such that the probability of erroneous transitions is very small.

In this work, we define a class of models to understand the impact of population size on opinion formation dynamics, a phenomenon usually related to group conformity. To this end, we introduce a new kinetic model in which the interaction frequency is weighted by the kinetic density. In the quasi-invariant regime, this model reduces to a Kaniadakis–Quarati-type equation with nonlinear drift, originally introduced for the dynamics of bosons in a spatially homogeneous setting. From the obtained PDE for the evolution of the opinion density, we determine the regime of parameters for which a critical mass exists and triggers blow-up of the solution. Therefore, the model is capable of describing strong conformity phenomena in cases where the total density of individuals holding a given opinion exceeds a fixed critical size. In the final part, several numerical experiments demonstrate the features of the introduced class of models and the related consensus effects.

Random collisions of particles occur in various biophysical and physical systems. Inspired by the binding of receptor and ligand on the cell membrane, we devised a method based on stochastic dynamical modeling to quantify the likelihood of two random particles colliding on a circle. We consider the dynamics of a receptor binding to a ligand on the cell membrane, where the receptor and ligand perform different motions and are thus modeled by stochastic differential equations with non-Gaussian noise. We use neural networks based on the Onsager–Machlup function to compute the probability $P_1$ of an unbounded receptor diffusing to the cell membrane. Meanwhile, we compute the probability $P_2$ of the extracellular ligand arriving at the cell membrane by solving the associated nonlocal Fokker–Planck equation. We can then calculate the most probable binding probability by combining $P_1$ and $P_2$. In this way, we conclude with some indication of how the receptors could distribute on the membrane, as well as where the ligand will most probably encounter the receptor, contributing to a better understanding of the cell’s response to external stimuli and communication with other cells.

A minimal model for reservoir computing is studied. We demonstrate that a reservoir computer exists that emulates given coupled maps by constructing a modularised network. We describe a possible mechanism for collapses of the emulation in the reservoir computing by introducing a measure of finite scale deviation. Such transitory behaviour is caused by either (i) an escape from a finite-time stagnation near an unstable chaotic set, or (ii) a critical transition driven by the effective parameter drift. Our approach reveals the essential mechanism for reservoir computing and provides insights into the design of reservoir computer for practical applications.

In this work we consider a new family of algorithms for sequential prediction, Hierarchical Partitioning Forecasters (HPFs). Our goal is to provide appealing theoretical - regret guarantees on a powerful model class - and practical - empirical performance comparable to deep networks - properties at the same time. We built upon three principles: hierarchically partitioning the feature space into sub-spaces, blending forecasters specialized to each sub-space and learning HPFs via local online learning applied to these individual forecasters. Following these principles allows us to obtain regret guarantees, where Constant Partitioning Forecasters (CPFs) serve as competitor. A CPF partitions the feature space into sub-spaces and predicts with a fixed forecaster per sub-space. Fixing a hierarchical partition $H$ and considering any CPF with a partition that can be constructed using elements of $H$ we provide two guarantees: first, a generic one that unveils how local online learning determines regret of learning the entire HPF online; second, a concrete instance that considers HPF with linear forecasters (LHPF) and exp-concave losses where we obtain $O(klogT)$ regret for sequences of length $T$ where $k$ is a measure of complexity for the competing CPF. Finally, we provide experiments that compare LHPF to various baselines, including state of the art deep learning models, in precipitation nowcasting. Our results indicate that LHPF is competitive in various settings.

We numerically analyse solutions of the spherically symmetric gravitational Vlasov–Poisson system close to compactly supported stable steady states. We observe either partially undamped oscillations or macroscopically damped solutions. We investigate for many steady states to which of these behaviours they correspond. A linear relation between the exponents of polytropic steady states and the qualitative behaviour close to them is identified. Undamped oscillations are also observed around not too concentrated King models and around all shells with a sufficiently large vacuum region. We analyse all solutions both at the non-linear and linearised level and find that the qualitative behaviours are identical at both. To relate the observed phenomena to theoretical results, we further include a comprehensive numerical study of the radial particle periods in the equilibria.

We study the nonexistence of multi-dimensional solitary waves for the Euler–Poisson system governing ion dynamics. It is well-known that the one-dimensional Euler–Poisson system has solitary waves that travel faster than the ion-sound speed. In contrast, we show that the two-dimensional and three-dimensional models do not admit nontrivial irrotational spatially localized traveling waves in the $L^1$ space for any traveling velocity and for general pressure laws. Our results provide theoretical evidence for the stability of line solitary waves in multi-dimensional Euler–Poisson flows. We derive some Pohozaev type identities associated with the energy and density integrals. This approach is extended to prove the nonexistence of irrotational multi-dimensional solitary waves for the two-species Euler–Poisson system for ions and electrons.

Residual Dynamic Mode Decomposition (ResDMD) offers a method for accurately computing the spectral properties of Koopman operators. It achieves this by calculating an infinite-dimensional residual from snapshot data, thus overcoming issues associated with finite truncations of Koopman operators (e.g., Extended Dynamic Mode Decomposition), such as spurious eigenvalues. Spectral properties computed by ResDMD include spectra, pseudospectra, spectral measures, Koopman mode decompositions, and dictionary verification. In scenarios where the number of snapshots is fewer than the dictionary size, particularly for exact DMD and kernelized Extended DMD, ResDMD has traditionally been applied by dividing snapshot data into a training set and a quadrature set. We demonstrate how to eliminate the need for two datasets through a novel computational approach of solving a dual least-squares problem. We analyze these new residuals for exact DMD and kernelized Extended DMD, demonstrating ResDMD’s versatility and broad applicability across various dynamical systems, including those modeled by high-dimensional and nonlinear observables. The utility of these new residuals is showcased through three diverse examples: the analysis of a cylinder wake, the study of airfoil cascades, and the compression of transient shockwave experimental data. This approach not only simplifies the application of ResDMD but also extends its potential for deeper insights into the dynamics of complex systems.

Ternary incompressible fluid flows extensively exist in atmospheric science, chemical engineering, and energy and power engineering, etc. The phase-field method is popular in multi-phase fluid modeling thanks to its efficient ability of interface capturing. This work aims to develop an energy dissipation law-preserving and temporally second-order accurate algorithm for a ternary phase-field fluid system. To establish a simple energy estimation, an adapted auxiliary variable approach is used to transform the original model into its equivalent form. Later, a second-order backward difference strategy is used to design the fully decoupled and linear time-marching scheme. To improve the consistency between original and modified discrete energy functionals, a practical energy correction technique is presented. We analytically prove the discrete energy dissipation property and show that the energy estimation can be easily established without considering the complex treatments of nonlinear and coupling terms. To facilitate the interested readers, we briefly describe the numerical implementation in each time step. The numerical tests indicate that the proposed method not only has desired accuracy, but also satisfies the energy stability even if a larger time step is used. Moreover, the proposed method can well simulate various ternary fluid phenomena, such as the liquid lens, phase separation, droplet dynamics, Kelvin–Helmholtz instability, and billowing cloud.

Basins of attraction (BoAs) are crucial for evaluating quality of a response and unraveling reliability of complex systems and mechanism of nonlinear phenomena. As a global strategy, however, it will pose a significant challenge to quantify a high-dimensional BoAs due to the curse of dimensionality and the insufficiency of data. This paper proposes a boundary-oriented progressive learning method based on the state space discretization, which aims to perform the classification of dynamics using few samples needed for learning while still achieving high efficiency and accuracy. Using pattern recognition network, training samples are purposefully extracted from the discretized and limited region that covers cells of boundary, disregarding the region outside of it. The region is then refined and identified iteratively to enhance discriminability of data model. This method does not seek to approximate the structure of boundary by refine cells, but rather regards cells as a framework of training neural network. The three typical examples are illustrated to show the power of the proposed method. The results demonstrate that the higher the dimensions, the better cost-effectiveness when compared to state-of-the-art approaches. The performance is even improved by more than 4 orders of magnitude on the computing loads when coping with a formidable six-dimensional BoA with satisfactory accuracy. Also, we discuss how the boundary-oriented progressive learning can improve the overall accuracy and robustness of the data model. Furthermore, this idea has the potential to efficiently handle other tasks of classification of dynamics beyond BoA, from a perspective of engineering.