Oscillatory Motion Simple pendulum: linear Hooke s Law restoring force for small angular deviations d 2 θ dt 2 = g l θ θ l Oscillatory solution θ(t) =θ 0 sin(ωt + φ) F with characteristic angular frequency Ω = g/l
Oscillatory Motion Simple pendulum: linear Hooke s Law restoring force for small angular deviations d 2 θ dt 2 = g l θ small angle approximation θ l Oscillatory solution θ(t) =θ 0 sin(ωt + φ) F with characteristic angular frequency Ω = g/l
Oscillatory Motion Important features: oscillations are perfectly regular in time and continue forever (no decay) θ l angular frequency is independent of the mass m and amplitude θ 0 F
Oscillatory Motion Usual trick for numerical solution: decompose the secondorder ODE into a system of first-order equations dω dt = g l θ dθ dt = ω Convert to a system of difference equations by discretizing the time variable, t t i = i t
Oscillatory Motion Simple pendulum is a Hamiltonian system described by an angular coordinate where I = ml 2 θ and angular momentum is the moment of inertia: L = Iω H = L2 2I + 1 2 IΩ2 θ 2 Hamilton s equations reproduce the first-order pair: L = H θ = IΩ2 θ θ = H L = L I dω dt = g l θ dθ dt = ω
Oscillatory Motion Hamiltonian system implies conservation of energy and preservation of the symplectic symmetry Neither are satisfied with Euler updates (small errors accumulate over each cycle) Important to use higher-order integration methods
Dissipation Simple pendulum is highly idealized More realistic models might include friction or damping terms proportional to the velocity d 2 θ dt 2 = g l θ q dθ dt Boring: Analytic solution shows that oscillations die out θ(t) e qt/2
Dissipation Depending on the strength of the parameter q, the behaviour may be under-, over-, or critically damped (Ω ] θ(t) =θ 0 e sin[ qt/2 2 1 4 q2) 1/2 t + φ [ θ(t) =θ 0 exp q/2 ± ( 1 4 q2 Ω 2) ] 1/2 t θ(t) = ( θ 0 + Ct ) e qt/2
Dissipation + Driving Force We now add a driving force to the problem Assume a sinusoidal form with strength F D and angular frequency Ω D d 2 θ dt 2 = g l θ q dθ dt + F D sin Ω D t Pumps energy into or out of the system and competes with the natural frequency when Ω D Ω
Dissipation + Driving Force In this case, the steady state solution is sinusoidal in the driving frequency: θ(t) =θ 0 sin(ω D t + φ) But the amplitude has a resonance when the driving frequency is close to the natural frequency: θ 0 = F D (Ω2 Ω 2 D )2 +(qω D ) 2
Dissipation + Driving Force In this case, the steady state solution is sinusoidal in the driving frequency: θ(t) =θ 0 sin(ω D t + φ) But the amplitude has a resonance when the driving frequency is close to the natural frequency: natural frequency θ 0 = F D (Ω2 Ω 2 D )2 +(qω D ) 2 transient dies out
Non-linearity Small angular approximation is no longer appropriate Pendulum may swing completely around its pivot θ No analytical solution with a sinusoidal restoring force F d 2 θ dt 2 = g l sin θ q dθ dt + F D sin Ω D t
Chaotic Motion When F D is sufficiently large, the motion has no simple long-time behaviour The motion never repeats and is said to be chaotic But it is not random
Chaotic Motion What does it mean to be non-repeating, unpredictable, and yet still deterministic? Remember: the behaviour is unique and governed by the specification of the initial value problem
Chaotic Motion For chaotic systems, infitesimal variations in the initial conditions lead to different long-time behaviour For example, two identical chaotic pendulums with nearly identical initially conditions will show exponential growth in the angular distance θ θ 1 θ 2
Transition to Chaotic Motion Regular and chaotic regions distinguished by a change of sign of the Lyapunov exponent λ θ e λt λ < 0 λ > 0
Lyapunov Exponents In general, there is a Lyapunov exponent associated with each phase-space degree of freedom δ 1 (t) δ 2 (t). δ N (t) = M(t) δ 1 (0) δ 2 (0). δ N (0) M(t) =U T exp λ 1 t λ 2 t... U λn t For a conservative system: N λ k =0 k=1 For a dissipative system: N λ k < 0 k=1
Lyapunov Exponents In general, there is a Lyapunov exponent associated with each phase-space degree of freedom δ 1 (t) δ 2 (t). δ N (t) = M(t) δ 1 (0) δ 2 (0). δ N (0) For a conservative system: M(t) =U T exp N k=1 λ k =0 λ 1 t λ 2 t... U λn t Unitary transformation: For a dissipative system: N k=1 λ k < 0 matrix of Eigenvectors of M
The View from Phase Space Chaotic path through phase space still exhibits structure There are many orbits that are nearly closed and persist for one or two cycles
Strange Attractor Poincaré section: only plot points at times in phase with the driving force Ω D t =2nπ for integer n For a wide range of initial conditions, trajectories lie on this surface of points, known as a strange attractor
Strange Attractor Poincaré section: only plot points at times in phase with the driving force Ω D t =2nπ for integer n For a wide range of initial conditions, trajectories lie on this surface of points, known as a strange attractor fractal structure
Period Doubling For the pendulum, the route to chaos is via period doubling The system shows response at a subharmonic Ω D /2
Period Doubling Bifurcation diagram plots a Poincaré section versus driving force Regularity in windows of period 2 n Feigenbaum delta: δ n F n F n 1 F n+1 F n Universal value δ 4.669