Lecture 20 - Tue 2020 11 10 - Virtual % ============================================================================== Left to the students: READ THE EXAMPLE of a driven-damped pendulum in section 8.5 book. The example involves three bifurcations of limit cycles [Hopf, Homo-clinic, and infinite period], and illustrates the use of the Poincare Map. % % ---------------------------------------------------------------------------- % % Flows in the 2-Torus. % ---------------------------------------------------- % Torus as square with glued opposite sides. Simple example \dot{theta_j} = omega_j; j=1, 2. Show periodicity if omega_1/omega_2 is rational. Trefoils when omega_1/omega_2 = 2/3 or 3/2 [project Torus on a plane, along main axis]. Case omega_1/omega_2 irrational. Orbit is ergodic. Precessing trefoil when omega_1/omega_2 close to 2/3 Example with nonlinear coupling: \dot{theta_1} = omega_1 + k_1*sin(theta_2-theta_1) \dot{theta_2} = omega_2 + k_2*sin(theta_1-theta_2) Show phase locking and compute compromise frequency. More generic couplings do not necessarily yield phase locking, but they can still produce synchronization [same period, but phase difference wobbles]. In theta_1 -- theta_2 phase plane orbits attracted to one where theta_2 advances a multiple of 2*pi when theta_1 advances some other multiple [reminiscent of the picture for the periodic orbits in the driven-damped pendulum]. This way can produce "simple" periodic orbits, or trefoils, or ... % % ---------------------------------------------------------------------------- % ALMOST PERIODIC % ------------------------------------------------------------ % Define almost periodic via "almost" repeatable. There are several, not quite equivalent definitions. Example: y = y(t) is almost periodic iff, for every epsilon, there is an "almost period" T > 0 such that |y(t+T) - y(t)| < epsilon. Relationship with Fourier series, can write: y(t) = sum a_n exp(i*omega_n*t) [A] Quasi-periodic as case where Fourier frequencies are generated by a finite set of frequencies: Any of the omega's in [A] has the form omega = \sum n_j Omega_j for some integers n_j and a finite set of frequencies Omega_1, ..., Omega_N. Note how y gets more and more complex as the number of base frequencies grow. Landau hypothesis on Turbulence as a series of Hopf-on-a-Hopf. Leading to turbulence as an almost periodic process. Note: problem 1: almost periodic leads to orbit separation that grows linearly, not exponentially. problem 2: such sequence of bifurcations, generally, cannot happen in a dissipative system. Will explain this later. ******************* problem 3: spectrum of an almost periodic function is still discrete, not continuous. Will explain this later. *************************** Landau's hypothesis, however, is related to the Ruelle Takens "route to chaos". In this route the first Hopf-on-a-Hopf happens, creating a quasi-periodic sln. with two periods [which tracks a 2-D Torus in phase space]. But when the third frequency appears, instead of three period quasi-periodic --- with a 3-D Torus associated; chaos ensues. Geometrically, the T2 torus "crumbles". If time permits, point out relationship with "small divisors" and KAM theory: Perturbing \ddot{x} + omega^2*x = 0, i.e.: \ddot{x} + omega^2*x = f(x), leads to a nonlinear oscillator with a single frequency [equal to omega in the infinitesimal amplitude limit]. Think of Poincare-Lindstead. However \ddot{x_1} + omega_1^2*x_1 = f(x_1, x_2) \ddot{x_2} + omega_2^2*x_2 = g(x_1, x_2) generates all the frequencies n1*omega_1 + n2*omega_2 and if omega_1/omega_2 is irrational, arbitrarily close to resonance terms are generated [which generate the "small divisors"]. The result is that some solutions survive but most do not [KAM theory ... explain]. Note that the scenario above is "sort of" what happens beyond the two period solution. % % ============================================================================== EOF