Emergence of solitary and chimera states in adaptive pendulum networks under diverse learning rules

This study demonstrates that adaptive pendulum networks governed by Hebbian or STDP learning rules spontaneously generate diverse collective states, including a novel delay-free solitary state driven purely by phase-lag variations, and systematically maps these dynamical transitions using incoherence measures and stability analysis.

The Gist

Imagine a large group of pendulums hanging from a ceiling, all swinging back and forth. In a normal physics experiment, if you push them all at once, they might swing together in perfect unison, or they might swing completely randomly.

But in this research, the scientists added a special "social rule" to these pendulums. They made the strength of the connection between them change based on how they are swinging. It's like if two pendulums swing together, they become best friends and hold hands tightly. If they swing in opposite directions, they might push each other away.

The researchers wanted to see what happens when these swinging pendulums "learn" from each other using two different types of social rules:

1. The "Frenemy" Rule (Hebbian): "If we swing together, we get closer." (Like the saying, "Birds of a feather flock together.")
2. The "Timing" Rule (STDP): A more complex rule based on exactly when one swings relative to the other, similar to how neurons in our brains learn.

Here is what they discovered, explained through simple analogies:

1. The Magic of the "Phase Lag" (The Delay)

The scientists introduced a tiny "delay" or "lag" in how the pendulums react to each other. Think of this like a group of dancers where everyone is slightly out of step with the music.

• The Surprise: Usually, to get a pendulum to swing alone while everyone else swings together, you need to add a delay, a random noise, or a special external push.
• The Discovery: This paper found that you don't need any of that. Just by tweaking that tiny "lag" in the timing, the system spontaneously creates a Solitary State.
  • Analogy: Imagine a choir where everyone sings the same note perfectly, except for one person who suddenly decides to sing a completely different tune. In this study, that one "loner" appeared naturally just because of the timing rules, without anyone telling them to leave the group.

2. The Different "Dance Parties" (The States)

Depending on how they set the rules and the timing, the pendulums formed eight different types of "dance parties":

• Two-Cluster (The Split Party): The group splits perfectly in half. One half swings left-to-right, the other half swings right-to-left. They are perfectly synchronized within their own group but opposite to the other group.
• Solitary (The Loner): As mentioned, almost everyone swings together, but one or two "rebels" break away and do their own thing.
• Multi-Antipodal (The Wheel): The group splits into three or more distinct teams, all spinning in a circle, each team offset from the next like slices of a pizza.
• Chimera (The Half-Sleeping City): This is a famous phenomenon in physics. Imagine a city where the left side is a bustling, synchronized metropolis, but the right side is a chaotic, sleeping village. Both exist at the same time in the same network.
• Splay (The Rotating Wave): Imagine a stadium "wave." Everyone stands up and sits down, but not at the same time. Neighbor A stands, then Neighbor B, then Neighbor C. It creates a perfect, smooth circle of motion where everyone is doing the same thing, just at a slightly different time.
• Splay-Chimera (The Mixed Wave): Part of the stadium does the perfect "wave," while the other part is just jumping around randomly.

3. The "Brain" Connection

Why does this matter? The rules the scientists used (Hebbian and STDP) are actually how human brains learn.

• Hebbian: "Neurons that fire together, wire together."
• STDP: The brain strengthens connections based on the precise timing of signals.

By studying these swinging pendulums, the scientists are actually building a simple model to understand how complex patterns emerge in the brain. They are showing that you don't need a complicated brain to get complex behaviors; you just need a network of simple units that can "learn" from each other.

4. The Big Takeaway

The most exciting part of this paper is the Solitary State.

• Old Thinking: To get a "loner" in a synchronized group, you needed to break the rules (add delays, make the pendulums different, or shake the table).
• New Thinking: You can get a "loner" just by changing the timing of the interaction. The system creates its own "rebel" naturally.

This suggests that in nature (like in fireflies flashing or neurons firing), complex behaviors like a single unit breaking away might happen simply because of how the timing of their interactions is set, without needing external chaos.

In summary: The researchers took a bunch of swinging pendulums, gave them the ability to "learn" from each other like a brain, and discovered that by simply adjusting the timing of their interactions, they could create a whole zoo of behaviors—from perfect harmony to chaotic loners and everything in between. It's a new way to understand how order and chaos can coexist in nature.

Technical Summary

Here is a detailed technical summary of the paper "Emergence of solitary and chimera states in adaptive pendulum networks under diverse learning rules."

1. Problem Statement

The study addresses the collective dynamics of networks of identical oscillators where the coupling strengths are not static but evolve dynamically based on the oscillators' states (adaptive networks). While previous research has extensively explored synchronization, chimera states, and cluster formation in adaptive networks, most studies rely on first-order phase oscillators or require specific conditions such as time delays, nonlocal coupling, or external perturbations to induce complex states like solitary states (where one or few oscillators desynchronize from a coherent group).

The specific problem investigated is whether solitary states and other complex collective patterns can emerge in a globally coupled network of second-order pendulum oscillators driven purely by adaptive learning rules (Hebbian and Spike-Timing-Dependent Plasticity) and phase lag, without the need for delays, heterogeneity, or external forcing.

2. Methodology

Model Formulation

The authors propose a network of $N$ identical pendulum oscillators governed by second-order differential equations with adaptive coupling:

1. Oscillator Dynamics:
   $$\ddot{\phi}_i + \dot{\phi}_i = \nu - \frac{1}{N} \sum_{j=1}^{N} k_{ij} \sin(\phi_i - \phi_j + \alpha)$$
   Where $\phi_i$ is the phase, $\dot{\phi}_i$ is angular velocity, $\nu$ is intrinsic frequency, and $\alpha$ is the phase-lag parameter.
2. Adaptive Coupling:
   $$\dot{k}_{ij} = -\epsilon (\sin(\phi_i - \phi_j + \beta) + k_{ij})$$
   Where $k_{ij}$ is the coupling strength (bounded by $|k_{ij}| \leq 1$), $\epsilon \ll 1$ separates timescales, and $\beta$ controls the adaptation rule.
   • Hebbian Rule: $\beta = -\pi/2$ ("neurons that fire together wire together").
   • STDP (Spike-Timing-Dependent Plasticity): $\beta = 0$.
   • Anti-Hebbian Rule: $\beta = \pi/2$.

Numerical and Analytical Techniques

• Simulation: The system was integrated using a fourth-order Runge-Kutta algorithm with $N=100$ oscillators. Initial conditions were randomized, and simulations ran for long transients to reach steady states.
• Characterization Metrics: To distinguish dynamical states, the authors employed two complementary incoherence measures based on local standard deviations across spatial bins:
  1. Frequency-based ($S$): Based on time-averaged frequencies.
  2. Phase-based ($S_\sigma$): Based on instantaneous phases.
• Discontinuity Measure ($\eta$): Introduced to distinguish between solitary states (single desynchronized oscillator) and multi-antipodal clusters (multiple groups), which share similar incoherence values.
• Analytical Stability: Linear stability analysis was performed on the Two-Cluster (TC) state to derive stability boundaries analytically.

3. Key Contributions

1. Delay-Free Solitary States: The most significant contribution is the demonstration that solitary states emerge spontaneously in a globally coupled adaptive network without time delays, nonlocal coupling, or external noise. This is induced purely by variations in the phase-lag parameter ($\alpha$) under Hebbian adaptation.
2. Diverse State Landscape: The study identifies eight distinct collective states arising from the interplay between phase lag ($\alpha$) and adaptation rules ($\beta$):
   • Two-Cluster (TC)
   • Solitary State (SS)
   • Multi-Antipodal Cluster (MAC)
   • Chimera (CHI)
   • Splay (SP)
   • Splay-Cluster (SPC)
   • Splay-Chimera (SPCHI)
   • Incoherent (INC)
3. Novel Hybrid States: The discovery of Splay-Chimera and Splay-Cluster states under STDP rules, which combine uniform phase gradients (splay) with partial synchronization or incoherence.
4. Analytical Validation: The derivation of an analytical stability condition for the Two-Cluster state, confirming its stability for $\beta \in (-\pi, 0)$, which aligns perfectly with numerical phase diagrams.

4. Results

Dynamical Transitions under Hebbian Adaptation ($\beta = -\pi/2$)

As the phase lag $\alpha$ increases, the system undergoes a systematic sequence of transitions:

• TC ($\alpha \approx 0.05\pi$): Two synchronized clusters oscillating in antiphase.
• SS ($\alpha \approx 0.12\pi$): Transition to a solitary state where one oscillator detaches while the rest remain synchronized. This occurs without external forcing.
• MAC ($\alpha \approx 0.25\pi$): Emergence of three or more antipodal clusters.
• CHI ($\alpha \approx 0.46\pi$): Coexistence of synchronized and desynchronized subpopulations (Chimera).

Dynamical Transitions under STDP ($\beta = 0$)

• SPCHI ($\alpha \approx 0.05\pi$): A hybrid state with a splay-ordered coherent domain and an incoherent domain.
• SPC ($\alpha \approx 0.1\pi$): Multiple clusters, each exhibiting internal splay-like ordering.
• SP ($\alpha \approx 0.25\pi$): A fully splay state where phases are uniformly distributed ($2\pi/N$) with identical frequencies but zero global order parameter.

Anti-Hebbian Adaptation ($\beta = \pi/2$)

• The network transitions to a completely Incoherent (INC) state for all values of $\alpha$, destabilizing all synchronization.

Multistability

The system exhibits extreme multistability, where multiple attractors (e.g., TC, SS, MAC) coexist for identical parameter sets depending on initial conditions. The phase diagrams reveal complex regions of coexistence (e.g., regions where TC, SP, and MAC states can all be stable).

5. Significance and Implications

• Theoretical Advancement: This work challenges the prevailing view that solitary states require nonlocal coupling or time delays. It establishes that phase lag combined with adaptive plasticity is a sufficient mechanism for generating these states in globally coupled systems.
• Biological Relevance: The model bridges the gap between abstract oscillator theory and biological systems (e.g., neuronal networks, firefly synchronization). The use of STDP and Hebbian rules suggests that similar mechanisms could drive complex synchronization patterns in neural circuits without requiring structural heterogeneity.
• Memory and Switching: The observation of extreme multistability implies that adaptive networks can store multiple collective states simultaneously. This has potential applications in neuromorphic computing and understanding memory encoding, where switching between states could represent information processing.
• Minimal Framework: The study demonstrates that a minimal model of adaptive pendulums is sufficient to reproduce a rich variety of complex dynamical phenomena, providing a robust framework for future studies on self-organized synchronization and multistable transitions.

In conclusion, the paper provides a comprehensive mapping of the dynamical landscape of adaptive pendulum networks, revealing that the interplay between phase lag and learning rules alone is sufficient to generate a spectrum of complex states, including the novel delay-free emergence of solitary states.