Boundary Hopf bifurcations in three-dimensional Filippov systems

This paper investigates boundary Hopf bifurcations in three-dimensional Filippov systems, demonstrating that the complex dynamics of associated grazing-sliding bifurcations reduce to a two-parameter family of piecewise-linear maps, for which the authors derive explicit formulas and provide a comprehensive numerical characterization of potential chaotic attractors.

The Gist

Imagine you are driving a car on a road that suddenly changes its rules. On the left side of a line painted on the road, the car accelerates automatically. On the right side, it brakes automatically. The line itself is the "switching surface."

This paper is about what happens when a car (or any system) is driving in a perfect circle (a limit cycle) and that circle gets bigger and bigger until it just barely touches that line. This moment of touching is called a Grazing Bifurcation.

But this paper studies a very specific, rare, and complex moment: a Boundary Hopf Bifurcation.

Here is the story of the paper, broken down into simple concepts:

1. The Setup: The Two-Mode System

Think of a thermostat.

• Mode A (Left): If the room is cold, the heater turns on full blast.
• Mode B (Right): If the room is hot, the AC turns on full blast.
• The Switch: The line where the temperature is "just right."

In many real-world systems (like pest control, harvesting fish, or mechanical brakes), the rules change abruptly when a threshold is crossed. This is a Filippov System.

2. The Drama: The Hopf Bifurcation

Imagine the temperature in the room starts to oscillate. It goes up and down in a perfect circle.

• The Hopf Bifurcation: This is the moment the room stops settling at a steady temperature and starts oscillating wildly. The "circle" of temperature changes is born.
• The Boundary: Now, imagine this oscillating circle grows until it hits the "Switching Surface" (the line where the rules change).

3. The Collision: Grazing-Sliding

When the oscillating circle hits the line, two things can happen:

• The Bounce: It hits the line and bounces off.
• The Slide: It hits the line and gets "stuck" sliding along it for a while before jumping back into the air. This is the Grazing-Sliding Bifurcation.

The paper asks: What happens to the system right after this collision? Does the system settle into a new, stable rhythm? Does it become chaotic (random and unpredictable)? Or does it fly apart?

4. The Big Discovery: The "Magic Map"

The author, D.J.W. Simpson, discovered that even though these systems look incredibly complicated (involving 3 dimensions, sliding, and sudden jumps), they can all be simplified into a 2D Map.

Think of this map as a video game level.

• The game has two zones: Left and Right.
• The rules for moving in the Left zone are different from the Right zone.
• The paper proves that near this specific "Boundary Hopf" moment, only two numbers (let's call them $\tau_L$ and $\tau_R$) determine the entire future of the system.

It's like saying that no matter how complex your car's engine is, if you are driving on this specific road, your fate depends entirely on two dials: Steering Sensitivity and Brake Sensitivity.

5. The Three Possible Outcomes

By analyzing these two "dials," the author mapped out three distinct futures for the system:

• Scenario A: The Chaotic Dance (The "Messy Room")
  If the dials are set to certain values, the system becomes chaotic. The temperature (or population, or speed) jumps around wildly and never repeats the same pattern. It's like a room where the heater and AC fight each other so violently that the temperature never settles.

  • Real-world example: A pest control model where the population fluctuates unpredictably, making it hard to manage.

• Scenario B: The Stable Slide (The "Smooth Glide")
  If the dials are set differently, the system finds a new, stable rhythm. It hits the line, slides along it for a bit, and then continues its cycle perfectly. It's a "sliding limit cycle."

  • Real-world example: A food chain where a predator population stabilizes by only hunting when prey is abundant, creating a predictable, safe cycle.

• Scenario C: The Ejection (The "Crash")
  In some settings, the system has no stable rhythm at all. When the cycle hits the line, the system is "ejected" into a completely different part of the world. It might crash into a new, strange attractor far away.

  • Real-world example: A harvesting model where, if you set the threshold wrong, the fish population doesn't just fluctuate; it collapses or explodes into a completely different state.

6. Why This Matters

The paper provides a formula. Before this, scientists had to run massive, slow computer simulations to guess what would happen when a system hit a switching surface.

Now, the author gives a "cheat code." You can take the physical parameters of your system (like the speed of a car or the birth rate of a pest), plug them into his formula, and instantly know:

1. Will it be chaotic?
2. Will it be stable?
3. Will it crash?

The "Virtual Counterpart" Trick

One of the coolest parts of the paper is the math trick used to solve it. To calculate what happens when the system hits the line, the author invented a "Virtual Counterpart."

Imagine you are trying to calculate the path of a ball hitting a wall. Instead of calculating the messy bounce, you imagine a "ghost ball" that passes through the wall. You calculate where the ghost ball would be, and then you simply measure the difference between the ghost and the real ball. This "ghost" method simplified a math problem that usually requires pages of complex equations down to a few clean lines.

Summary

This paper is a user manual for chaos. It tells us that when complex systems (like ecosystems or machines) hit a "tipping point" where they switch rules, their behavior isn't random. It follows a hidden, simple map. By understanding just two numbers, we can predict whether the system will dance chaotically, slide smoothly, or crash and burn.

Technical Summary

1. Problem Statement

The paper investigates the local dynamics of three-dimensional piecewise-smooth ordinary differential equations (Filippov systems) near a specific codimension-two bifurcation known as the Boundary Hopf Bifurcation.

• Context: In Filippov systems, the state space is divided by a switching surface $\Sigma$. A Boundary Hopf bifurcation occurs when an equilibrium point of one smooth piece of the system lies exactly on $\Sigma$ and undergoes a Hopf bifurcation (where a pair of complex conjugate eigenvalues crosses the imaginary axis).
• The Challenge: While codimension-one bifurcations (like standard Hopf or saddle-node) are well-understood, codimension-two points act as "organizing centers" for the global dynamics. Specifically, the unfolding of a boundary Hopf bifurcation involves a curve of grazing-sliding bifurcations, where a limit cycle born from the Hopf bifurcation collides with the switching surface.
• The Gap: In general, the local dynamics near grazing-sliding bifurcations are described by high-dimensional, piecewise-linear maps with many parameters. The author seeks to determine if, for 3D systems, this complexity can be reduced to a manageable, lower-dimensional form that allows for a complete characterization of the resulting attractors (which may be chaotic).

2. Methodology

The author employs a combination of analytical bifurcation theory, asymptotic analysis, and numerical continuation.

• System Formulation: The study focuses on systems of the form $\dot{X} = F_L(X)$ for $H(X)<0$ and $\dot{X} = F_R(X)$ for $H(X)>0$, where the equilibrium $X^*$ of $F_R$ lies on the switching surface $\Sigma$ ($H(X^*)=0$) at the bifurcation point.
• Coordinate Transformation and Blow-up: To analyze the limit as the bifurcation parameters approach zero, the author performs an affine coordinate change and a "spatial blow-up." This scales the neighborhood of the limit cycle to an $O(1)$ region, allowing the Jacobian of the system to be converted into real Jordan form.
• Poincaré Map Decomposition: The dynamics are analyzed via a 2D Poincaré map $P$. The map is decomposed into two pieces:
  • $P_R$: Orbits that do not touch the sliding region.
  • $P_L$: Orbits that include a sliding segment along $\Sigma$.
• Discontinuity Map Derivation: A critical component is the derivation of the discontinuity map (or grazing map), which accounts for the correction caused by the sliding segment. The author presents a novel, simplified derivation of the linear term of this map for $n$-dimensional systems. Instead of calculating points directly, the method uses a "virtual counterpart" approach, computing the difference between the actual trajectory and a virtual extension of the flow. This significantly reduces the algebraic complexity.
• Normal Form Reduction: By evaluating the traces and determinants of the Jacobians of $P_L$ and $P_R$ at the grazing point, the author reduces the infinite-dimensional problem to a two-parameter family of 2D border-collision normal forms.
• Numerical Analysis: The reduced normal form is analyzed numerically to map out the bifurcation diagram of the $(\tau_L, \tau_R)$ parameter plane, identifying regions of fixed points, period-doubling, and chaos.
• Application to Models: The theoretical results are applied to three specific models: a pedagogical example, a pest control model, and a food chain model with threshold harvesting.

3. Key Contributions

A. Theoretical Reduction to a 2-Parameter Family

The primary theoretical contribution is proving that for 3D Filippov systems, the complex dynamics near a boundary Hopf bifurcation are governed by a specific two-parameter family of the 2D border-collision normal form:
$$\begin{bmatrix} x \\ y \end{bmatrix} \mapsto \begin{cases} \begin{bmatrix} \tau_L x + y - 1 \\ 0 \end{bmatrix}, & x \le 0 \\ \begin{bmatrix} \tau_R x + y - 1 \\ (1-\tau_R)x \end{bmatrix}, & x \ge 0 \end{cases}$$
This reduction is possible because:

1. Sliding motion reduces the dimension of the flow on the switching surface.
2. The stability of the limit cycle is degenerate at the Hopf point (Floquet multiplier $\to 1$).
   The parameters $\tau_L$ and $\tau_R$ are explicitly derived in terms of the system's physical quantities (eigenvalues, Lie derivatives, and vector field values at the equilibrium).

B. New Derivation of the Discontinuity Map

The paper provides a new, simpler derivation (Appendix B) for the linear term of the discontinuity map associated with grazing-sliding bifurcations. By utilizing a virtual counterpart (a trajectory extended smoothly across the switching surface) and calculating displacements, the author avoids the need to retain higher-order terms that eventually vanish, streamlining the asymptotic analysis for $n$-dimensional systems.

C. Comprehensive Characterization of the Normal Form

The paper performs a detailed numerical bifurcation analysis of the reduced normal form. It classifies the $(\tau_L, \tau_R)$ plane into distinct regions:

• Stable Fixed Points: Corresponding to stable limit cycles with sliding segments.
• Period-Two Solutions: Stable cycles with two points.
• Chaotic Attractors: Finite unions of line segments where the dynamics are transitive and chaotic.
• No Attractor: Regions where orbits diverge.
  The author identifies specific bifurcation curves (e.g., where the attractor splits into components or gains "limbs") and maps the transition from smooth limit cycles to chaotic behavior.

4. Results

• Explicit Formulas: The paper derives explicit formulas for the limiting values of the normal form parameters ($\tau_L, \tau_R, \delta_R$) at the codimension-two point in terms of the original system's Jacobian eigenvalues ($\pm i\beta_0, \gamma_0$) and geometric vectors.
• Dynamical Scenarios:
  • Scenario 1 (Chaotic): If the parameters fall into specific regions (e.g., $\tau_L \approx -1.86, \tau_R \approx 1.28$), the grazing-sliding bifurcation creates a chaotic attractor immediately.
  • Scenario 2 (Stable Sliding): If parameters are in the region of stable fixed points (e.g., $\tau_L \approx -0.02, \tau_R \approx 1$), the limit cycle remains stable but acquires a sliding segment.
  • Scenario 3 (Explosion/Divergence): If parameters correspond to "no attractor" (e.g., $\tau_L \approx 1.54, \tau_R \approx 1$), the grazing-sliding bifurcation destroys the local limit cycle, ejecting orbits to a distant chaotic attractor or a different basin of attraction.
• Model Applications:
  • Pest Control: Demonstrated that threshold-based control can lead to a stable limit cycle with sliding motion (intermittent control application).
  • Food Chain Harvesting: Showed that harvesting based on prey thresholds can cause a "jump" from a stable limit cycle to a large-amplitude chaotic attractor, a phenomenon distinct from the soft emergence of chaos seen in other scenarios.

5. Significance

• Predictive Power: The work provides a rigorous framework for predicting the qualitative behavior of complex 3D systems (mechanical, ecological, or control systems) near critical switching points. It allows researchers to determine a priori whether a system will exhibit chaos, stability, or divergence upon crossing a grazing-sliding curve.
• Simplification of Complexity: By reducing the problem to a 2-parameter map, the paper makes the "extraordinarily rich dynamical behaviour" of grazing-sliding bifurcations tractable for analysis and classification.
• Methodological Innovation: The "virtual counterpart" derivation technique for discontinuity maps offers a powerful tool for future research in non-smooth dynamics, potentially simplifying the analysis of other sliding bifurcations (like adding-sliding).
• Practical Implications: The results are directly applicable to engineering (impact mechanics, relay control) and ecology (pest management, harvesting strategies), highlighting how small changes in control thresholds can lead to catastrophic shifts in system behavior (e.g., sudden onset of chaos or population collapse).