Lorenz System — Mathematics Knowledge System

Overview

The Lorenz system is a three-dimensional system of ordinary differential equations originally derived by Edward Lorenz in 1963 as a simplified model of atmospheric convection. It is the prototypical example of a chaotic dynamical system, exhibiting sensitive dependence on initial conditions (the butterfly effect) and a strange attractor — a fractal set of Hausdorff dimension approximately 2.06 toward which nearly all trajectories are attracted. The system has three fixed points (for the classic parameter values), two of which are unstable spirals surrounded by the lobes of the butterfly-shaped attractor. Despite being only three-dimensional and fully deterministic, its long-term behavior is practically unpredictable.

Intuition

The Lorenz attractor looks like a pair of butterfly wings or owl eyes: trajectories spiral around one lobe for a while, then unpredictably switch to spiraling around the other lobe. How long they stay on each lobe is sensitive to initial conditions. Yet the attractor itself is robust: nearly all trajectories eventually find their way to these two lobes and never leave the neighborhood. The three variables represent: x = rate of convective overturning, y = temperature difference between ascending and descending fluid, z = deviation from linear vertical temperature profile. The parameters sigma (Prandtl number), rho (Rayleigh number ratio), and beta (geometric factor) control the behavior.

Formal Definition

Definition

The Lorenz equations are a system of three coupled nonlinear ODEs with three positive parameters sigma, rho, and beta.

\dot{x} = \sigma(y - x)

Lorenz equation 1 (x = convective intensity)

\dot{y} = x(\rho - z) - y

Lorenz equation 2 (y = horizontal temperature difference)

\dot{z} = xy - \beta z

Lorenz equation 3 (z = vertical temperature variation)

\sigma = 10, \quad \rho = 28, \quad \beta = 8/3 \quad \text{(Lorenz's classic values)}

Classic parameter values

Notation

Notation	Meaning
$\sigma$	Prandtl number (ratio of fluid viscosity to thermal diffusivity)
$\rho$	Rayleigh number ratio (normalized convection intensity)
$\beta$	Geometric factor of the convection region
$C^\pm$	Two nontrivial fixed points of the Lorenz system for rho > 1

Theorems

Theorem 1: Fixed Points of the Lorenz System

For rho < 1 the origin is the only fixed point and is globally stable. For rho > 1, two additional fixed points C^+ = (\sqrt{ \beta(\rho-1) }, \sqrt{ \beta(\rho-1) }, \rho-1) and C^- = (-\sqrt{ \beta(\rho-1) }, -\sqrt{ \beta(\rho-1) }, \rho-1) exist. For the classic values rho=28, these are unstable foci.

Theorem 2: Global Attracting Region

All trajectories of the Lorenz system enter and remain in the ellipsoidal region defined by V(x,y,z) = x^2 + y^2 + (z -- rho -- sigma)^2 <= C for sufficiently large C. The system is dissipative: volumes in phase space contract at constant rate --(sigma + 1 + beta) < 0.

Theorem 3: Strange Attractor (Tucker 2002)

For sigma=10, rho=28, beta=8/3, the Lorenz system has a robust strange attractor: a compact invariant set with a dense orbit, sensitive dependence on initial conditions, and Hausdorff dimension approximately 2.06. This was proved rigorously by Warwick Tucker using interval arithmetic in 2002.

Worked Examples

1
Set all derivatives to zero. From x' = sigma(y-x) = 0: y = x.
$\sigma(y - x) = 0 \implies y = x$
2
From z' = xy - beta*z = 0 with y=x: x^2 = beta*z, so z = x^2/beta.
$z = \frac{x^2}{\beta}$
3
From y' = x(rho-z) - y = 0 with y=x and z=x^2/beta: x(rho - x^2/beta) - x = 0. For x ≠ 0: rho - x^2/beta - 1 = 0, so x^2 = beta*(rho-1).
$x^2 = \beta(\rho - 1)$
4
For rho=28, beta=8/3: x^2 = (8/3)(27) = 72, so x = ±6*sqrt(2). Then y = ±6*sqrt(2), z = 27.
$C^\pm = (\pm 6\sqrt{2},\, \pm 6\sqrt{2},\, 27)$

✓ Answer

Three fixed points: origin (0,0,0) and C± = (±6sqrt(2), ±6sqrt(2), 27) for rho=28, beta=8/3.

Practice Problems

Mediumfree response

For the Lorenz system with rho < 1, show the origin is globally asymptotically stable using the Lyapunov function V = x^2/sigma + y^2 + z^2.

Easyfree response

Explain in plain language why the Lorenz system cannot have periodic orbits for the parameter values rho=28, sigma=10, beta=8/3.

Common Mistakes

Common Mistake

Treating the Lorenz attractor as a 2D surface.

The Lorenz attractor is a fractal with Hausdorff dimension approximately 2.06, not a smooth surface; it has fine structure at all scales.

Common Mistake

Thinking the Lorenz system's chaotic behavior means it is random.

The Lorenz equations are perfectly deterministic; chaos arises from sensitive dependence on initial conditions, not from any randomness in the equations.

Common Mistake

Treating the nontrivial fixed points C+ and C- as attractors for rho=28.

For rho=28, C+ and C- are unstable foci; trajectories spiral away from them and are attracted to the strange attractor, not to these fixed points.

Common Mistake

Believing a numerical simulation faithfully tracks the exact trajectory from given initial conditions for all time.

Due to floating-point round-off and sensitive dependence, long simulations track some true Lorenz trajectory but not necessarily the one starting from the nominal initial condition; only the attractor geometry is faithfully reproduced.

Quiz

Historical Background

Edward Lorenz, a meteorologist at MIT, derived his famous system in 1963 while studying a twelve-variable weather model. He truncated it to three modes of a Fourier expansion of the Navier-Stokes equations for Rayleigh-Benard convection. While rerunning a simulation from a printout with rounded values, he noticed catastrophically different outcomes — the first clear demonstration of chaos in a scientific model. His 1963 paper in the Journal of Atmospheric Sciences went largely unnoticed for a decade until the chaos revolution of the 1970s. Guckenheimer and Williams proved in 1979 that the Lorenz system does indeed have a strange attractor (called the geometric Lorenz attractor), and Warwick Tucker gave a computer-assisted proof in 2002 that the actual Lorenz system has a robust strange attractor.

1963
Lorenz derives the three-equation system and discovers chaos by accident
Edward Lorenz
1972
Lorenz popularizes the butterfly effect in his talk 'Predictability: Does the Flap of a Butterfly's Wings in Brazil Set Off a Tornado in Texas?'
Edward Lorenz
1979
Guckenheimer and Williams prove the geometric Lorenz attractor is truly chaotic
John Guckenheimer, Robert Williams
2002
Tucker proves by computer-assisted proof that the Lorenz attractor is a robust strange attractor
Warwick Tucker

Summary

The Lorenz system is a three-dimensional ODE derived from atmospheric convection; for rho=28, sigma=10, beta=8/3 it exhibits chaos.
It has three fixed points for rho > 1: the origin (unstable) and two symmetric points C± (also unstable for rho=28).
The system is dissipative (div F = -(sigma+1+beta) < 0), so phase space volumes shrink exponentially and an attractor of zero volume must exist.
The Lorenz strange attractor is a fractal set of Hausdorff dimension approximately 2.06; typical trajectories spend forever spiraling around it without repeating.

References

BookStrogatz, S. H. Nonlinear Dynamics and Chaos. Westview Press, 2015. Chapter 9.
WebsiteWikipedia — Lorenz system