2026-10-16 | | Total: 14
Reservoir computing has established itself as a computationally efficient alternative to fully trained recurrent neural networks using a fixed high-dimensional nonlinear substrate to map temporal input into a feature space for a simple readout. Here, we argue that reservoir computing can be realized entirely within a coarse-grained, symbolic state space, and that constrained random walks on ordinal-pattern transition networks provide a natural instance of this realization, replacing the reservoir's continuous state space with a discrete and directly data-derived one. We develop a formal correspondence between the two frameworks, show that the echo-state property has a natural discrete analog in the ergodicity and spectral gap of the walk's underlying Markov chain, and illustrate the correspondence with a minimal example and an empirical one-step-ahead prediction task evaluated across five paradigmatic model systems. These systems exhibit dynamics that correspond to strictly periodic, quasiperiodic, chaotic, or stochastic signals. We argue that this symbolic reservoir perspective allows a redefinition of the reservoir substrate itself, replacing the requirement of a continuous, randomly generated dynamical system with any sufficiently ergodic, data-derived symbolic process. This opens a route toward interpretable, data-adaptive reservoirs whose internal structure is inferred rather than randomly assigned, with implications for surrogate-data methods and reservoir design more broadly.
Stochasticity in complex systems is commonly treated either as external noise added to deterministic dynamics or as an effective stochastic description of deterministic chaos. A framework in which stochastic dynamics arise endogenously from deterministic chaos and subsequently feed back on the deterministic dynamics, thereby fundamentally reshaping system dynamics, remains rare. Here, we develop such a framework. Through the interaction between deterministic and stochastic dynamics, alternative stable states, critical transitions, adaptive capacity, stochasticity, and resilience become dynamically interconnected within a single model. Furthermore, the framework establishes a new mechanism of long transients, driven by two key factors: (i) time-scale separation and (ii) time-delay-induced cancellation. Stochasticity plays a dual stabilizing-destabilizing role in long transients: it promotes their persistence while also facilitating their eventual escape and collapse. This dual role allows resilience to be quantified dynamically through escape and stochastic recapture dynamics. Finally, we establish the generality of the mechanism mathematically and numerically across a broad class of one-dimensional nonlinear systems, suggesting that it reflects a general dynamical principle.
We propose a general statistical framework for characterizing erratic functions through the statistics of their extrema, with the aim of identifying statistical signatures associated with chaotic processes. It is based on analyzing the extrema points of erratic functions associated with these processes. One is instructed to determine the probability distribution functions ( PDF)s of the locations of the extrema points, the spacings between the locations, in particular the nearest neighbors, and the ratios between adjacent spacings. We then fit these PDFs with PDFs known to relate to chaotic behavior like of random matrix theory. The extrema considered establish a cloud of points in general in $D$ dimensions for erratic function of several variables. For certain systems one finds also extrema curves. We further determine the ``thermodynamic" description of the erratic functions by computing the Shannon or differential entropy and by equating the PDFs to the Gibbs one. With this map we determine the ``energy function" and for certain cases also the ``temperature". We analyze three benchmark spacing PDFs: the Bernoulli shift map, the Poisson and the Gaussian beta ensemble and several others. We want to emphasize that the map to thermodynamics can be applied to variety of chaotic processes including non-physics ones. We distinguish marginal spacing distributions from joint distributions of consecutive spacings. The latter probe correlations and statistical organization that are invisible in one-spacing PDFs and may provide stronger signatures for distinguishing structured chaotic-like behavior from uncorrelated irregularity.
Cooperative behavior and its regulatory mechanisms are central to complex systems and evolutionary game theory. In natural populations, strategy updates and endogenous payoff environments interact bidirectionally, shaping cooperation. We propose a strategy-payoff environmental-feedback model for four canonical games: Prisoner's Dilemma, Snowdrift, Stag-Hunt, and Harmony. Mathematical derivation reduces the high-dimensional coupled system to an analytically equivalent canonical replicator equation. Bifurcation diagrams, phase portraits, and simulations show how initial cooperation, temptation to defect, and punishment jointly regulate cooperation. Stag-Hunt and Harmony display bistability and initial-state dependence; Prisoner's Dilemma exhibits robust dynamics with stable mixed equilibria. Temptation and punishment drive sequential cross-type game transitions at critical thresholds. Sufficient punishment suppresses defection, shifts bifurcation thresholds, delays destabilization of cooperative steady states, and reshapes these transitions. Numerical tests confirm equivalence and stability switching. Spatial lattice simulations agree with theoretical predictions, supporting robustness. This work reveals regulatory mechanisms from endogenous payoff feedback and initial-condition sensitivity, providing a basis for understanding and optimizing collective prosocial behavior.
How densely must a network of identical Kuramoto oscillators be wired before synchrony is its only stable state? The answer is a critical connectivity $μ_c$ (the minimum degree divided by $n-1$), known to satisfy $11/16\leμ_c<3/4$. We raise the lower bound from $11/16=0.6875$ to $μ_c\ge(4+2q^*)/(7+q^*)=0.6929077\ldots$, where $q^*\approx0.6506$ is a root of an explicit polynomial of degree 8. The networks approaching this bound carry stable states that hug the four-cluster state: three neighboring clusters each split into two slightly separated halves, and the four crossed links between neighboring halves (those whose phase gap deviates most from a quarter turn) have densities approaching $q^*$. We build these networks from a coarse-grained cluster model, verify stability with interval arithmetic, and blow the model up into large networks with an exact minimum degree. The states are fragile and easily missed by numerical searches that impose stability margins.
We consider the \textsc{ABC}-equation $Au_xu_{yt} + Bu_yu_{xt} + Cu_tu_{xy} = 0$, $A$, $B$, $C\in\mathbb{R}$, which was proved to be integrable in the case $A + B + C = 0$ (the Veronese web equation). Our study reveals ten more cases that admit integrable structures that, to the best of our knowledge, where not known before. The techniques is based essentially on the theory of the tangent and cotangent coverings.
We develop a constrained-power variational framework for positive homoclinic standing waves in one-dimensional discrete nonlinear Schrödinger equations with cooperative finite-range interactions. Starting from periodic lattices, we construct constrained energy maximizers and use the associated projected gradient flow to establish positive invariance of the cone of nonnegative, even, and unimodal profiles. A central result is that the monotone ordering of the interaction coefficients, \(a_1\ge a_2\ge\cdots\ge a_M\ge0\), with \(a_1>0\), provides the structural condition ensuring preservation of this cone under finite-range coupling. Under a strict maximum condition above the linear spectral edge, we obtain a uniform frequency gap and period-independent exponential localization. These estimates provide the compactness required for the infinite-period limit and yield positive, even, unimodal, and exponentially localized homoclinic standing waves as strong \(\ell^2(\mathbb Z)\)-limits of periodic constrained maximizers. Numerical computations for several interaction ranges and cooperative kernels illustrate the effects of extended coupling on dispersion and localization and support the periodic-to-homoclinic convergence predicted by the analysis.
Simulating chaotic planetary system evolution requires resolving small time-scales due to close encounters, scattering, collisions, and more. Simulations typically assign fixed timesteps to all particles, however, which is inefficient and wastes compute resources in a chaotically evolving system; moreover, it may under-resolve the simulation. Hybrid methods are a blunt tool for resolving small time-scales as they do not allow for fine-grained adaptation. Assigning individual timesteps, which vary in time, to planetary bodies usually violates desirable properties such as symplecticity and reversibility. In this work, we develop an "almost" exactly time-reversible direct $N$-body integrator, MTR, in which each pair of planets gets its own timestep, which itself can vary as the system evolves. In contrast to our previous work, we achieve this adaptation on the fly, independent of a global timestep. We provide a C version of the code and test it on several problems, demonstrating no energy error drift, as expected of a reversible method. At large particle numbers, up to $N = 975$, we compare against the state-of-the-art GPLUM code, which we can use in direct Hermite mode (for which it is not optimized) or tree mode (for which it is). MTR provides a speedup of up to two orders of magnitude compared to the Hermite mode, while agreeing with the distribution of orbits. In our default configuration, MTR's speed matches or improves on the tree mode's speed, by up to an order magnitude, despite the tree mode's faster and approximate force calculation.
We investigate a single-component Abelian-Higgs model with an eighth-order scalar potential whose vacuum manifold consists of two disconnected circles. The model supports two species of vortices and domain walls interpolating between the two vacuum components. We find type-1.5-like behavior characterized by nonmonotonic intervortex interactions and bound states of spatially separated vortices. Inner-vacuum vortices form clusters through short-range repulsion and long-range attraction, whereas outer-vacuum vortices form vortex-filled bubbles in which a closed domain wall confines multiple inner-vacuum-vortex cores. We construct the domain wall analytically, derive the asymptotic vortex-wall interactions, and analyze the fluctuation spectra of coincident multivortices to identify their splitting instabilities.
Pantograph equations are functional differential equations with proportional delay, typically involving terms of the form $x(at)$, where $0<a<1$. These equations have been studied for several decades and have applications in areas ranging from number theory to electrodynamics. In this work, we investigate pantograph difference equations as discrete dynamical maps of the form \[ x_{i+1}=F(x_{i})+G(x_{h(i)}), \qquad i\geq 0, \] where $h(i)\leq i$. In particular, we consider $h(i)=\lfloor ai \rfloor$ and $h(i)=\lfloor ai^α\rfloor$, with $0<a<1$ and $0<α<1$. We distinguish between homogeneous pantograph difference equations, for which $F(x)=0$, and inhomogeneous pantograph difference equations, for which $F(x)\ne 0$. We first analyze linear pantograph difference equations and establish sufficient conditions for asymptotic stability and for nonconvergence to zero. For proportional delays, analytical stability and instability regions are combined with multi-horizon parameter scans to construct regime diagrams in the $(λ,δ)$-plane. We then investigate pantograph versions of the logistic and tent maps. In the homogeneous case, the dynamics evolves through step-like plateaus whose lengths increase at least exponentially, leading to slow convergence and finite-time multistability within computationally accessible time scales. In the inhomogeneous case, numerical experiments show finite-time exponential separation of nearby trajectories, consistent with chaotic dynamics.
This paper considers a nonlocal advection--reaction--diffusion equation with logistic growth. We investigate how the interaction kernel affects the criticality of Turing bifurcations and the associated multistability. We first prove that nonhomogeneous stationary solutions bifurcate from the positive homogeneous stationary solution. Center manifold reduction yields an amplitude equation near the bifurcation point. This equation provides criteria for supercritical and subcritical bifurcations in terms of the Fourier transform of the interaction kernel. We also derive a sufficient condition for subcriticality. This condition can be satisfied by sufficiently localized attractive interactions. Based on these analytical results, we investigate multistability through numerical bifurcation analysis. The numerical results show that the stable homogeneous stationary solution can coexist with stable nonhomogeneous stationary solutions or stable time-periodic solutions.
We consider the full distribution $P_N(A)$ of the dynamical observable $A = - \sum_{i=1}^N \ln x_i$ where $x_1, \dots, x_N$ is a sequence generated by a chaotic map $f(x)$ via $x_{n+1} = f(x_n)$. We study the particular cases where $f(x)$ is the doubling, tent or logistic map, and focus on the limit $N \gg 1$. Typical fluctuations of $A$ are Gaussian with mean $\left\langle A\right\rangle$ and variance which are both of order $\sim N$. We find that the left tail of the distribution ($A < \left\langle A\right\rangle$) scales as $P_{N}(A) \sim e^{-NI(A/N)}$, and calculate the rate function $I(a)$ which depends on the map $f(x)$. Interestingly, we find that the right tail of the distribution ($A > \left\langle A\right\rangle$) follows an anomalous scaling: $P_{N}(A) \sim e^{-N^{1/3} F((A-\left\langle A\right\rangle) / N^{2/3})}$. The large-deviation function $F(y)$ has a universal form for the three maps, differing only in the scaling constants. At a critical value $y=y_c$, it exhibits a corner singularity which has the character of a condensation transition. The transition separates between a subcritical phase $0< y<y_c$ in which the fluctuations are Gaussian and arise homogeneously from all of the $x_i$'s, and a supercritical phase $y>y_c$ in which a small fraction of the $x_i$'s give a macroscopic contribution to the fluctuation. Despite this anomalous behavior of the right tail, we find that the cumulants of the distribution $P_N(A)$ grow linearly with $N$ at large $N$, and calculate the associated coefficients.
Synchronization of coupled oscillators on complex networks underlies critical functions in natural and engineered systems, including power-grid operation. While synchronization stability on fixed network topologies has been studied extensively, less is known about how an added transmission link affects stability depending on where it is placed relative to generation and consumption. We address this question using simplified ring-network power-grid models in which producer and consumer nodes are spatially organized into clusters. Long-range transmission links, which we refer to as shortcuts, are added between non-adjacent nodes, and synchronization stability is quantified by basin stability in second-order coupled-oscillator dynamics. We compare shortcut placements across four scenarios varying network size, power allocation, power-injection heterogeneity, and producer--consumer arrangement. When producers and consumers form two spatially separated clusters, center-to-center shortcuts do not reduce the mean basin stability in the configurations examined and increase it by up to 75\% relative to the shortcut-less baseline. Peripheral shortcuts instead reduce mean basin stability by up to 28\% in some configurations despite the added connection, producing Braess-like degradation. In alternating multi-block networks, however, no single shortcut placement provides the same substantial enhancement, indicating that the center-to-center advantage does not directly generalize to more complex spatial arrangements. Representative time-domain simulations further show faster decay of frequency deviations for center-to-center than for peripheral shortcuts. These results demonstrate that the effect of added connectivity depends critically on where the new link is placed relative to the spatial organization of generation and consumption.
We construct a seven-cycle example for a periodically forced scalar equation with a continuous three-zone piecewise-linear nonlinearity. Symbolic integration and linear elimination reduce the four-flight matching conditions to two equations in the residence times. The competing exterior visits produce an additional turn of the critical response and lead to the rational parameters $(A,B,μ)=(9,3/2,181363/100000)$. At these parameters, finite Taylor estimates give three sign changes of the return displacement. Scalar order, analyticity along transverse itineraries, and half-period reflection then establish at least seven distinct isolated periodic orbits on the phase cylinder.