Abstract:

This lecture provides a comprehensive overview of classical chaos, beginning with a review of integrable systems and their phase space structure of nested tori. The core of the discussion is the Kolmogorov–Arnold–Moser (KAM) theorem, which describes the fate of these tori when a non-integrable perturbation is introduced. The theorem states that sufficiently irrational tori survive by deforming, while rational tori are destroyed, leading to a mixed phase space of regular islands and chaotic seas. The lecture introduces the Poincaré surface of section as a crucial tool for visualizing these complex dynamics in two dimensions. Various physical systems are used as examples, including coupled quartic oscillators, billiard systems (integrable like the circle, and chaotic like the stadium), and the double pendulum, to illustrate the transition from regular to chaotic motion. The lecture culminates in a detailed analysis of one-dimensional maps, focusing on the logistic map and its period-doubling route to chaos. The concept of sensitive dependence on initial conditions is quantified by the Lyapunov exponent, and the logistic map is shown to be analytically solvable at its fully chaotic point via a transformation to the "doubling map," whose left-shift dynamics in binary provides a clear and intuitive mechanism for information loss and chaotic behavior.

From Integrability to Chaos: Key Concepts and Examples

• 00:00:16 - Integrable Systems: The lecture begins by recapping integrable systems, defined as having as many independent constants of motion as degrees of freedom. Their dynamics are confined to N-dimensional tori in phase space.
• 00:02:32 - The KAM Theorem: This central theorem explains what happens when an integrable system is perturbed. It states that tori with "sufficiently irrational" frequency ratios survive the perturbation (they deform but are not destroyed), while tori with rational frequency ratios are destroyed.
• 00:13:59 - Poincaré Surface of Section: A visualization technique to study systems with two or more degrees of freedom. By plotting the system's state (e.g., position x and momentum px) only when another coordinate crosses a specific plane (e.g., y=0), the 4D phase space is reduced to a 2D map, making the dynamics easier to analyze. In an integrable system, this creates clear, nested curves (tori), while in a chaotic system, it can produce a splatter of points.
• 00:21:28 - Example: Quartic Oscillators: This system is used to show the transition from an integrable case (with regular, nested tori) to a non-integrable, "mixed" phase space with both stable islands and chaotic regions, and finally to a fully chaotic state where a single orbit appears to fill the entire available phase space.
• 00:36:39 - Example: Billiard Systems: A free particle bouncing inside a 2D boundary. The system's integrability depends on the shape of the boundary.
  • Integrable Billiards: Include the circle (conserves angular momentum) and the rectangle (conserves momentum components separately).
  • Chaotic Billiards: Include the stadium billiard and the Sinai billiard (a square with a circular scatterer), which are proven to be fully chaotic.
• 00:50:07 - Sensitive Dependence on Initial Conditions: This is the hallmark of chaos. Two initial conditions that are infinitesimally close will diverge exponentially fast over time. This is contrasted with integrable systems, where the divergence is much slower (polynomial). The triple pendulum is used as a dramatic visual example.
• 00:54:55 - One-Dimensional Maps: To simplify the study of chaos, the lecture shifts to 1D maps like xn+1 = f(xn). The famous logistic map, xn+1 = r * xn * (1 - xn), is introduced as a model that exhibits a "period-doubling" route to chaos as the parameter r is increased.
• 01:05:24 - The Lyapunov Exponent (λ): This is the quantitative measure of chaos. It represents the average exponential rate at which nearby trajectories separate. A positive Lyapunov exponent (λ > 0) is a defining feature of a chaotic system.
• 01:17:45 - Solvability and the Doubling Map: The logistic map at its most chaotic point (r=4) is shown to be analytically solvable. A change of variables (x = sin²(πθ)) transforms the logistic map into the much simpler doubling map: θn+1 = 2θn (mod 1).
• 01:23:47 - The Mechanism of Chaos: Left-Shift Dynamics: The doubling map reveals the underlying mechanism of chaos. When the variable θ is written in binary, each iteration of the map corresponds to a "left shift," moving all bits one position to the left and discarding the most significant digit. This process systematically destroys information about the initial condition, providing a clear explanation for the sensitive dependence and unpredictability of the system.