Abstract
Chaos is associated with stochasticity, complex, irregular motion, and extreme sensitivity to initial conditions. These pseudo-random features lead chaotic systems to enormous applications such as random number generators, image encryption, and secure communication. In general, the concept of chaos is never associated with similarity. However, we found the chaotic systems belonging to one chaos family (OCF) have similar dynamic behavior, which is a novel characteristic of chaos. In this work, three classical chaotic system families are studied: the Lorenz family, the Chua family, and the hyperbolic sine family. These systems contain different derived chaotic systems, different order chaotic systems, and different kinds of chaotic systems (chaos and hyper-chaos). Their Phase Space Plots (PSPs) demonstrate that there exists a strong correlation within an OCF. Moreover, we found that high-order/dimensional chaotic systems inherit all dynamic behavior of lower ones, and the similarity decreases as the order/dimension goes higher, which is analogous to genetic processes in biology. All of these features are quantitatively evaluated by PPMCC and SSIM.
Introduction
In the study of nonlinear dynamics, chaos is typically defined by its "pseudo-random" features: the butterfly effect, ergodicity, and long-term unpredictability. Because of these traits, researchers have generally assumed there is no meaningful correlation between different chaotic systems. This paper challenges that traditional view by demonstrating that systems within the same "family" exhibit remarkably similar global structures. While two chaotic trajectories might diverge locally, the topological structure of their attractors remains cohesive within a family.
Methodology
The researchers employed a mix of numerical simulation and image-processing metrics to evaluate three major families: the Lorenz family, the Chua family, and the hyperbolic sine family.
- Numerical Simulation: The 4th-order Runge-Kutta (RK4) method was used to calculate trajectories for systems ranging from 3D to 10D.
- Phase Space Plots (PSPs): Trajectories were visualized as 2D projections (e.g., $x-y$ or $x-z$ planes) to treat the attractors as visual data.
- Quantitative Evaluation: Two primary metrics were used to measure similarity:
- PPMCC (Pearson Product-Moment Correlation Coefficient): Measures linear correlation between data sets.
- SSIM (Structural Similarity Index): Repurposed from image processing to compare the topological structure of the phase space plots.
The Genetic Analogy of Chaos
A striking insight from this research is that chaotic systems undergo a process analogous to biological inheritance. A 5D or 10D system isn't merely a more complex version of a 3D system; it actually "inherits" the fundamental geometric structures of its lower-dimensional precursors. This suggests that the "DNA" of a chaotic family is preserved even as the system's dimensionality increases.
Key Results
The Chua Family
The Chua family demonstrates the highest level of structural preservation. The 3D, 4D, and 5D versions showed extremely high correlation, with several PPMCC and SSIM scores reaching near 1.0. This indicates that the phase space structures remain nearly identical across dimensions.
The Hyperbolic Sine Family
While inheritance exists, it is subject to "similarity decay." In the hyperbolic sine family, the 3D system's traits are clearly visible in the 10D system, but the quantitative correlation drops significantly as complexity increases. PPMCC scores fell from approximately 0.97 (4D vs 3D) to 0.24 (10D vs 3D), illustrating how the "genetic" signal weakens in higher dimensions.
The Lorenz Family
Derived systems like the Chen and Lü systems maintain high structural correlation with the original Lorenz attractor (SSIM scores around 0.76–0.78). This proves that these systems are mathematically cohesive, though the Lorenz family exhibits more "diversity" in its dynamic expressions compared to the Chua family.
Conclusion and Future Work
This work establishes that chaotic systems within the same family are not isolated islands of randomness but are linked by structural inheritance. Future research will focus on providing rigorous mathematical proofs for these empirical observations and refining the metrics used to evaluate chaotic topology. Understanding these similarities could lead to more robust applications in secure communication and system modeling.
Summary updated