Sunteți pe pagina 1din 7

IDC621: Non-Linear Dynamics

The Lorenz System

Nikhil Kumar MS08035

Abstract In this paper we study the Lorenz system, nding solutions to the coupled dierential equations describing the system. We also study a number of bifurcation diagrams, depicting the (one of the parameter involved in the model) dependence of the attractor.

Introduction.

The Lorenz system is a system of ordinary dierential equations (the Lorenz equations) rst studied by Edward Lorenz. It is notable for having chaotic solutions for certain parameter values and initial conditions. In particular, the Lorenz attractor is a set of chaotic solutions of the Lorenz system which, when plotted, resemble a buttery or gure eight.

History.

The Lorenz system of dierential equations arose from the work of meteorologist/mathematician Edward N. Lorenz, who was studying thermal variation in an air cell underneath a thunder. As he was computing numerical solutions to the system of three dierential equations that he came up with, he noticed that initial conditions with small dierences eventually produced vastly dierent solutions. What he observed was sensitivity to initial conditions, a characteristic of chaos. His observations led him to further study of the system, and since that time, about 1963, the Lorenz system has become one of the most widely systems of ODEs because of its wide range of behaviors.

The Lorenz system.

The system of dierential equations Lorenz use was : x = (y x) y = x( z) y z = xy z (1) (2) (3)

where , and are positive parameters which denote the physical characteristic of air ow. The variable x corresponds to the amplitude of convective current of air cell, y to the temperature dierence between rising and falling currents, and to z to the deviation of the temperature from the normal temperature in the cell. Even though a denition of chaos has not been agreed upon by 2

mathematicians, two properties that are generally agreed to characterize it are: sensitivity to initial conditions and the presence of limit cycles which repeatedly double their period as is varied in one direction until the orbits begin to wander chaotically. We will explore these dynamics and other behaviors of the Lorenz system.

4
4.1

Analysis of the system.


Steady States and Stability.

Solving equations (1)-(3) at equilibrium, i.e.: (y x ) = 0 x ( z ) y = 0 x y z = 0, which yields that the steady states are: (x , y , z ) = (0, 0, 0) & ( ( 1), ( 1), 1). This pair is stable only if = ++3 , which can hold only for positive if 1 > + 1. For < 1, all solutions are attracted to the origin. At = 1, the two equilibrium points appear with a period doubling bifurcation. Further, they are stable until some , the gure below shows the unstable manifold of the origin for = 10, = 10, = 8/3 which end up as part of the stable manifold of the two equilibrium points.

4.2

Jacobian and solutions.

The system of equations (1)-(3) can be represented as X = F (X) has the Jacobian matrix, 0 DF (x, y, z) = z 1 x . At the steady state (0,0,0), the y x 0 Jacobian DF(0,0,0) is 1 0 , which is a block diagonal. The 0 0 1s (1s)2 +4rs . For r < 1, where eigenvalues are b, 2 2 + 4sr < (1 + s), all three eigenvalues are negative. For r > 1, we (1 s) have one positive eigenvalue and one negative eigenvalue, of which the positive eigenvalue belongs to unstable manifold, which is in turn part of the Lorenz attractor. At the two other steady state, the eigenvalues are the roots of a polynomial of degree 3. For > b + 1 and 1 < < = (( + b + 3)/( b 1), all eigenvalues have a negative real part and the two xed points are stable. At = , a Hopf bifurcation happens, the two stable xed points collide, each with an unstable cycle and become unstable. For = 10, b = 8/3 we have = 470/19 = 24.7. For large r parameters, the attractor can be single periodic orbit. Some periodic solutions are knots.

Attractors and Bifurcation.

one normally assumes , , > 0, but usually = 10, = 8/3, and is varied. The system exhibits chaotic behavior for = 28 but displays knotted periodic orbits for other values of . For example, with = 99.96 it becomes a T(3,2) torus knot(as shown in Plot-2). A Saddle-node bifurcation occurs at ( 1) = 0. When = 0 and ( 1) 0, the equations generate three critical points. The critical points at (0,0,0) correspond to no convection, and the critical points at ( ( 1), ( 1), 1) correspond to steady convection. This ++3 pair is stable only if < 1 , which can hold only for positive if > + 1. When = 28, = 10, and = 8/3, the Lorenz system has chaotic solutions (not all solutions are chaotic). The set of chaotic solutions make up the Lorenz attractor, a strange attractor and a fractal of Hausdor dimension between 2 and 3 (see Plot-1). Grassberger(1983) has estimated the Hausdor dimension to be 2.06 0.01 and the correlation dimension to be 2.05 0.01.1
P.Grassberger and I. Procaccia(1983). Measuring the strangeness of strange attractors.Physica D 9(1-2):189-208.
1

Plot-1. Chaotic solution of the Lorenz system , the Lorenz attractor.

Plot-2. T(3,2) torus knot obtained for = 99.96.

Plot-3. At = 24.74 = 470/19, the unstable cycles collide with the stable equilibrium points, and render them unstable. This is called subcritical Hopf bifurcation.

Results.

Although only a few plots have been added here, but there other plots which can be of interest, as: For 0 < < 1, the origin is the only equilibrium point. At = 1, a pitchfork bifurcation takes place, the origin becomes unstable and two stable equilibrium points appear. Between = 0.99524 and = 100.795, one observes a series of period doubling bifurcations of stable periodic points (one has to start with the larger value and decrease r). These bifurcations are analogue to the Feigenbaum scenario. The system has sensitive dependence on the initial condition and divergence of orbits can be clearly observed by varying t, keeping , , xed.

References.
1. Strogatz, Steven H.(1994). Nonlinear Systems and Chaos. Perseus publishing. 2. P.Grassberger and I. Procaccia(1983). Measuring the strangeness of strange attractors. Physica D 9(1-2):189-208. 3. lorenz, E.N.(1963). Deterministic nonperiodic ow. J. Atmos. Sci. 20 (2):130-141.

Appendix 1. Simulating the Lorenz system.


Following is a code written in SAGE to simulate the Lorenz system. import numpy as np from numpy import * from scipy.integrate import odeint import scipy def lorenz_int(initial, t): x = initial[0] y = initial[1] z = initial[2] sigma = 10 rho = 28 beta = 8.0/3 x_dot = sigma * (y - x) y_dot = x * (rho -z) - y z_dot = x * y - beta* z return [x_dot, y_dot, z_dot] #initial conditions. initial = [0, 1, 1.05] t = scipy.arange(0, 200, 0.01) lorenz_sol = odeint(lorenz_int, initial, t) x = [i[0] for i in lorenz_sol] y = [i[1] for i in lorenz_sol] z = [i[2] for i in lorenz_sol] DataOut = column_stack((x,y,z)) savetxt(out.dat, DataOut) Change the value of variable rho, to get the solutions of the Lorenz system for dierent values of . Now plot the data le out.dat using GNUPlot to get the bifurcation diagrams for the system.

S-ar putea să vă placă și