arXiv subjects

Toshio Aoyagi

Publications and source records attributed to Toshio Aoyagi.

At least 19 recordsLinked to original sources

arXiv2026

Data-driven reconstruction of spatiotemporal phase dynamics for traveling and oscillating patterns via Bayesian inference

Building on the phase reduction theory formulated for reaction-diffusion systems with spatial translational symmetry, we develop a data-driven method that reconstructs the spatiotemporal phase dynamics of traveling and oscillating patterns. Spatiotemporal phase dynamics are described by spatial and temporal phases that represent the position and oscillation of the pattern, respectively. Using Bayesian inference, our method directly reconstructs phase equations from time-series data. When tested on simulation data from coupled Gray-Scott models exhibiting traveling breathers, the method accurately reconstructs the deterministic part of the phase equations in the weak-noise regime, in which the phase dynamics converge to a linearly stable fixed point.

nlin.AO↗

arXiv2026

A Phase Description of Mutually Coupled Chaotic Oscillators

The synchronization of rhythms is ubiquitous in both natural and engineered systems, and the demand for data-driven analysis is growing. When rhythms arise from limit cycles, phase reduction theory shows that their dynamics are universally modeled as coupled phase oscillators under weak coupling. This simple representation enables direct inference of inter-rhythm coupling functions from measured time-series data. However, strongly rhythmic chaos can masquerade as noisy limit cycles. In such cases, standard estimators still return plausible coupling functions even though a phase-oscillator model lacks a priori justification. We therefore extend the phase description to the chaotic oscillators. Specifically, we derive a closed equation for the phase difference by defining the phase on a Poincaré section and averaging the phase dynamics over invariant measures of the induced return maps. Numerically, the derived theoretical functions are in close agreement with those inferred from time-series data. Consequently, our results justify the applicability of phase description to coupled chaotic oscillators and show that data-driven coupling functions retain clear dynamical meaning in the absence of limit cycles.

nlin.CD↗

arXiv2025

Gaussian process regression with additive periodic kernels for two-body interaction analysis in coupled phase oscillators

Since many physical laws -- from classical mechanics to electromagnetism -- are formulated as two-body interactions, the same perspective naturally extends to biological and social dynamics. Here we focus on rhythmic phenomena, where phase reduction theory shows that synchronization dynamics can be universally described by coupled phase oscillators. Estimating the interaction functions of such systems from data offers a direct path to understanding and predicting such dynamics. Existing Fourier-series-based methods encounter difficulties with limited or biased data. To overcome this, we employ Gaussian process regression with additive periodic kernels. In our approach, we incorporate information about the estimation target into the statistical model in advance by designing kernel functions that capture the characteristics -- additivity and $2π$-periodicity -- of the coupling functions. Furthermore, owing to the Bayesian framework, our method enables the evaluation of uncertainty in the estimation results. We validate our approach on Van der Pol, FitzHugh-Nagumo, and spiking neural models. Our approach outperforms Fourier-series baselines in both error and stability under biased phase sampling. This enables data-driven studies of rhythm dynamics across a broader range of datasets. Furthermore, it makes a first step toward a statistically grounded, data-driven approach to general many-body systems with two-body interactions.

nlin.AO↗

arXiv2025

Neuronal correlations shape the scaling behavior of memory capacity and nonlinear computational capability of reservoir recurrent neural networks

Reservoir computing is a powerful framework for real-time information processing, characterized by its high computational ability and quick learning, with applications ranging from machine learning to biological systems. In this paper, we investigate how the computational ability of reservoir recurrent neural networks (RNNs) scales with an increasing number of readout neurons. First, we demonstrate that the memory capacity of a reservoir RNN scales sublinearly with the number of readout neurons. To elucidate this observation, we develop a theoretical framework for analytically deriving memory capacity that incorporates the effect of neuronal correlations, which have been ignored in prior theoretical work for analytical simplicity. Our theory successfully relates the sublinear scaling of memory capacity to the strength of neuronal correlations. Furthermore, we show this principle holds across diverse types of RNNs, even those beyond the direct applicability of our theory. Next, we numerically investigate the scaling behavior of nonlinear computational ability, which, alongside memory capacity, is crucial for overall computational performance. Our numerical simulations reveal that as memory capacity growth becomes sublinear, increasing the number of readout neurons successively enables nonlinear processing at progressively higher polynomial orders. Our theoretical framework suggests that neuronal correlations govern not only memory capacity but also the sequential growth of nonlinear computational capabilities. Our findings establish a foundation for designing scalable and cost-effective reservoir computing, providing novel insights into the interplay among neuronal correlations, linear memory, and nonlinear processing.

cond-mat.dis-nn↗

arXiv2024

Setting of the Poincaré section for accurately calculating the phase of rhythmic spatiotemporal dynamics

The synchronization analysis of limit-cycle oscillators is prevalent in many fields, including physics, chemistry, and life sciences. It relies on the phase calculation that utilizes measurements. However, the synchronization of spatiotemporal dynamics cannot be analyzed because a standardized method for calculating the phase has not been established. The presence of spatial structure complicates the determination of which measurements should be used for accurate phase calculation. To address this, we explore a method for calculating the phase from the time series of measurements taken at a single spatial grid point. The phase is calculated to increase linearly between event times when the measurement time series intersects the Poincaré section. The difference between the calculated phase and the isochron-based phase, resulting from the discrepancy between the isochron and the Poincaré section, is evaluated using a linear approximation near the limit-cycle solution. We found that the difference is small when measurements are taken from regions that dominate the rhythms of the entire spatiotemporal dynamics. Furthermore, we investigate an alternative method where the Poincaré section is applied to the time series obtained through orthogonal decomposition of the entire spatiotemporal dynamics. We present two decomposition schemes that utilize the principal component analysis. For illustration, the phase is calculated from the measurements of spatiotemporal dynamics exhibiting target waves or oscillating spots, simulated by weakly coupled FitzHugh-Nagumo reaction-diffusion models.

nlin.AO↗

arXiv2024

Suppression of chaos in a partially driven recurrent neural network

The dynamics of recurrent neural networks (RNNs), and particularly their response to inputs, play a critical role in information processing. In many applications of RNNs, only a specific subset of the neurons generally receive inputs. However, it remains to be theoretically clarified how the restriction of the input to a specific subset of neurons affects the network dynamics. Considering RNNs with such restricted input, we investigate how the proportion, $p$, of the neurons receiving inputs (the "inputs neurons") and the strength of the input signals affect the dynamics by analytically deriving the conditional maximum Lyapunov exponent. Our results show that for sufficiently large $p$, the maximum Lyapunov exponent decreases monotonically as a function of the input strength, indicating the suppression of chaos, but if $p$ is smaller than a critical threshold, $p_c$, even significantly amplified inputs cannot suppress spontaneous chaotic dynamics. Furthermore, although the value of $p_c$ is seemingly dependent on several model parameters, such as the sparseness and strength of recurrent connections, it is proved to be intrinsically determined solely by the strength of chaos in spontaneous activity of the RNN. This is to say, despite changes in these model parameters, it is possible to represent the value of $p_c$ as a common invariant function by appropriately scaling these parameters to yield the same strength of spontaneous chaos. Our study suggests that if $p$ is above $p_c$, we can bring the neural network to the edge of chaos, thereby maximizing its information processing capacity, by amplifying inputs.

q-bio.NC↗

arXiv2022

Extracting phase coupling functions between collectively oscillating networks directly from time-series data

Many real-world systems are often regarded as weakly coupled limit-cycle oscillators, in which each oscillator corresponds to a dynamical system with many degrees of freedom that have collective oscillations. One of the most practical methods for investigating the synchronization properties of such a rhythmic system is to statistically extract phase coupling functions between limit-cycle oscillators directly from observed time-series data. In Particular, using a method that combines phase reduction theory and Bayesian inference, the phase coupling functions can be extracted from the time-series data of even just one variable in each oscillatory dynamical system with many degrees of freedom. However, it remains unclear how the choice of the observed variables affects the statistical inference for the phase coupling functions. In this study, we examine the influence of observed variable types on the extraction of phase coupling functions using some typical dynamical elements under various conditions. We demonstrate that our method can consistently extract the macroscopic phase coupling functions between two phases representing collective oscillations in a fully locked state, regardless of the observed variable types; for example, even using one variable of any element in one system and the mean-field value over all the elements in another system. We also study the case of globally coupled phase oscillators in a partially locked state. Our results reveal directional asymmetry in the robustness of extracting the macroscopic phase coupling function between two networks. For instance, when an asynchronous oscillator in network $A$ and the macroscopic collective oscillation of network $B$ is observed, the macroscopic phase coupling function from network $A$ to network $B$ can be extracted more robustly than in the opposite direction.

nlin.AO↗

arXiv2021

Phase description of chaotic oscillators

This paper presents a phase description of chaotic dynamics for the study of chaotic phase synchronization. A prominent feature of the proposed description is that it systematically incorporates the dynamics of the non-phase variables inherent in the system. Taking these non-phase dynamics into account is essential for capturing the complicated nature of chaotic phase synchronization, even in a qualitative manner. We numerically verified the validity of the proposed description in application to the Rössler and Lorenz oscillators, and we found that our method provides an accurate description of the characteristic distorted shapes of the synchronization regions for these chaotic oscillators. Furthermore, the proposed description allows us to systematically identify and describe the origin of this distortion.

nlin.CD↗

arXiv2019

Interaction Mechanisms Quantified from Dynamical Features of Frog Choruses

Interaction mechanism in the acoustic communication of actual animals is investigated by combining mathematical modeling and empirical data. Here we use a deterministic mathematical model (a phase oscillator model) to describe the interaction mechanism underlying the choruses of male Japanese tree frogs (Hyla japonica) in which the male frogs attempt to avoid call overlaps with each other due to acoustic communication. The mathematical model with a general interaction term is identified by a Bayesian approach from multiple audio recordings on the choruses of three male frogs. The identified model qualitatively reproduces the stationary and dynamical features of the empirical data, supporting the validity of the model identification. In addition, we quantify the magnitude of attention paid among the male frogs from the identified model, and then analyze the relationship between the attention and behavioral parameters by using a statistical model. The analysis demonstrates the biologically valid relationship about the negative correlation between the attention and inter-frog distance, and also indicates the existence of a behavioral strategy that the male frogs selectively pay attention towards a less attractive male frog so as to utilize the advantage of their attractiveness for effective mate attraction.

nlin.AO↗

arXiv2018

A role of asymmetry in linear response of globally coupled oscillator systems

The linear response is studied in globally coupled oscillator systems including the Kuramoto model. We develop a linear response theory which can be applied to systems whose coupling functions are generic. Based on the theory, we examine the role of asymmetry introduced to the natural frequency distribution, the coupling function, or the coupling constants. A remarkable difference appears in coexistence of the divergence of susceptibility at the critical point and a nonzero phase gap between the order parameter and the applied external force. The coexistence is not allowed by the asymmetry in the natural frequency distribution but can be realized by the other two types of asymmetry. This theoretical prediction and the coupling-constant dependence of the susceptibility are numerically verified by performing simulations in $N$-body systems and in reduced systems obtained with the aid of the Ott-Antonsen ansatz.

nlin.AO↗

arXiv2018

Use of recurrent infomax to improve the memory capability of input-driven recurrent neural networks

The inherent transient dynamics of recurrent neural networks (RNNs) have been exploited as a computational resource in input-driven RNNs. However, the information processing capability varies from RNN to RNN, depending on their properties. Many authors have investigated the dynamics of RNNs and their relevance to the information processing capability. In this study, we present a detailed analysis of the information processing capability of an RNN optimized by recurrent infomax (RI), which is an unsupervised learning scheme that maximizes the mutual information of RNNs by adjusting the connection strengths of the network. Thus, we observe that a delay-line structure emerges from the RI and the network optimized by the RI possesses superior short-term memory, which is the ability to store the temporal information of the input stream in its transient dynamics.

cs.NE↗

arXiv2016

Nonstandard transitions in the Kuramoto model: A role of asymmetry in natural frequency distributions

We study transitions in the Kuramoto model by shedding light on asymmetry in the natural frequency distribution, which has been assumed to be symmetric in many previous studies. The asymmetry brings two nonstandard bifurcation diagrams, with the aid of bimodality. The first diagram consists of stationary states, and has the standard continuous synchronization transition and a subsequent discontinuous transition as the coupling strength increases. Such a bifurcation diagram has been also reported in a variant model, which breaks the odd symmetry of the coupling function by introducing the phase lag. The second diagram includes the oscillatory state emerging from the partially synchronized state and followed by a discontinuous transition. This diagram is firstly revealed in this study. The two bifurcation diagrams are obtained by employing the Ott-Antonsen ansatz, and are verified by direct $N$-body simulations. We conclude that the asymmetry in distribution, with the bimodality, plays a similar role to the phase lag, and diversifies the transitions.

nlin.AO↗

arXiv2016

Dynamics of two populations of phase oscillators with different frequency distributions

A large variety of rhythms are observed in nature. Rhythms such as electroencephalogram signals in the brain can often be regarded as interacting. In this study, we investigate the dynamical properties of rhythmic systems in two populations of phase oscillators with different frequency distributions. We assume that the average frequency ratio between two populations closely approximates some small integer. Most importantly, we adopt a specific coupling function derived from phase reduction theory. Under some additional assumptions, the system of two populations of coupled phase oscillators reduces to a low-dimensional system in the continuum limit. Consequently, we find chimera states in which clustering and incoherent states coexist. Finally, we confirm consistent behaviors of the derived low-dimensional model and the original model.

nlin.CD↗

arXiv2015

Learning in Neural Networks Based on a Generalized Fluctuation Theorem

Information maximization has been investigated as a possible mechanism of learning governing the self-organization that occurs within the neural systems of animals. Within the general context of models of neural systems bidirectionally interacting with environments, however, the role of information maximization remains to be elucidated. For bidirectionally interacting physical systems, universal laws describing the fluctuation they exhibit and the information they possess have recently been discovered. These laws are termed fluctuation theorems. In the present study, we formulate a theory of learning in neural networks bidirectionally interacting with environments based on the principle of information maximization. Our formulation begins with the introduction of a generalized fluctuation theorem, employing an interpretation appropriate for the present application, which differs from the original thermodynamic interpretation. We analytically and numerically demonstrate that the learning mechanism presented in our theory allows neural networks to efficiently explore their environments and optimally encode information about them.

cond-mat.dis-nn↗

arXiv2014

Direct extraction of phase dynamics from fluctuating rhythmic data based on a Bayesian approach

Employing both Bayesian statistics and the theory of nonlinear dynamics, we present a practically efficient method to extract a phase description of weakly coupled limit-cycle oscillators directly from time series observed in a rhythmic system. As a practical application, we numerically demonstrate that this method can retrieve all the interaction functions from the fluctuating rhythmic neuronal activity exhibited by a network of asymmetrically coupled neurons. This method can be regarded as a type of statistical phase reduction method that requires no detailed modeling, and as such, it is a very practical and reliable method in application to data-driven studies of rhythmic systems.

nlin.AO↗

arXiv2012

Scale-Free Structure Emerging from Co-Evolution of a Network and the Concentration of Diffusive Resources on It

Co-evolution exhibited by a network system, involving the intricate interplay between the dynamics of the network itself and the subsystems connected by it, is a key concept for understanding the self-organized, flexible nature of real-world network systems. We propose a simple model of such co-evolving network dynamics, in which the diffusion of a resource over a weighted network and the resource-driven evolution of the link weights occur simultaneously. We demonstrate that, under feasible conditions, the network robustly acquires scale-free characteristics in the asymptotic state. Interestingly, in the case that the system includes dissipation, it asymptotically realizes a dynamical phase characterized by an organized scale-free network, in which the ranking of each node with respect to the quantity of the resource possessed thereby changes ceaselessly. Our model offers a unified framework for understanding some real-world diffusion-driven network systems of diverse types.

physics.soc-ph↗

arXiv2011

Multistable attractors in a network of phase oscillators with three-body interaction

Three-body interactions have been found in physics, biology, and sociology. To investigate their effect on dynamical systems, as a first step, we study numerically and theoretically a system of phase oscillators with three-body interaction. As a result, an infinite number of multistable synchronized states appear above a critical coupling strength, while a stable incoherent state always exists for any coupling strength. Owing to the infinite multistability, the degree of synchrony in asymptotic state can vary continuously within some range depending on the initial phase pattern.

nlin.AO↗

arXiv2009

A Weighted Spike-Triggered Average of a Fluctuating Stimulus Yielding the Phase Response Curve

We demonstrate that the phase response curve (PRC) can be reconstructed using a weighted spike-triggered average of an injected fluctuating input. The key idea is to choose the weight to be proportional to the magnitude of the fluctuation of the oscillatory period. Particularly, when a neuron exhibits random switching behavior between two bursting modes, two corresponding PRCs can be simultaneously reconstructed, even from the data of a single trial. This method offers an efficient alternative to the experimental investigation of oscillatory systems, without the need for detailed modeling.

nlin.AO↗