You are on page 1of 5

Name: Fan Zhang

EAS6490-HW1

August 28, 2012

Solutions of Lorenz Attractor

After increasing the initial condition of X slightly (0.2), both X and Y diverge after some time. Although we have exact Y conditions, it still diverges as X initial condition changes. This is because the systme is non-linear. Small changes in X initial conditions result in big dierences in the future state. Variable Y stuidied in isolation of X appears to develop random uctuations. The system is chaotic.

Fig. 1.Time series of X, Y and Z. Blue: initial conditions (X = 10, Y = 10, Z = 10); red: small changes on X initial condition (X = 10 + 0.2, Y = 10, Z = 10).

Phase space diagram

The 3-D phase space diagram of X, Y and Z shows a clear pattern (buttery pattern), which demonstrates that the trajectory is predictable. 1

Name: Fan Zhang

EAS6490-HW1

August 28, 2012

If a small error introduced into X initial condtion, the buttery pattern remains: there are small dierences between trajectories with dierent initial conditions. We can still predict the genral behavior of the system. However, in the 2-D phase space diagram of X and Z, it seems that multiple trajectories pass through the same point (in fact, in 3-D, trajectory never intersect). The visible intersection of trajectories indicates apparent randomness of the system, due to the lack of observations of Y. In this context, X and Z are signal, and Y is the source of noise.

Fig. 2.Upper: The 3-D phase space diagram of X, Y and Z; lower: the 2-D phase space diagram of X and Z.

Name: Fan Zhang

EAS6490-HW1

August 28, 2012

3
3.1

Probability Density Function (PDF), Joint and Conditional PDF


PDF for X and Z

The Probability Distribution Function (PDF) of X and Z oer a global view of statistics of X and Z. We could tell the occurence probability of X/Z at a certain value.

Fig. 3.Probability Distribution Function (PDF) of X and Z

3.2

Joint PDF of XZ

The Joint PDF of X and Z tells the probability of the system at certain values of X and Z. It is shown (Fig. 4) that the Joint PDF of X and Z has higher intensity in the center, from which we could infer that trajectories slow down in the center. Joint PDF adds another constraint to the system.

Name: Fan Zhang

EAS6490-HW1

August 28, 2012

Fig. 4.The Joint PDF of X and Z.

3.3

Joint PDF of XZ with Y conditions

a) The Conditional Joint PDF of X and Z tells the occurence probability of X and Z on certain values of Y. When Y has certain values, the Joint PDF of X and Z are constrained to a specic pattern, which is part of the original Joint PDF of X and Z. b) Compared to the 2-D phase spcae diagram of X and Z, the Joint PDF not only shows the potential values and pattern of X and Z, but also reveals the possibility of X and Z at certain values. The Joint PDF provides more insightful information about the system.

Fig. 5.Conditional Joint PDF of X and Z on1 < Y < 1 and 8 < Y < 10. P.S. There is probably a minor mistake in the matlab script ( EXAM P LE lorenz.m) from the class website, in which the line 115: tmp1 = [tmp1; xout2(in, 2)]; should be tmp1 = 4

Name: Fan Zhang

EAS6490-HW1

August 28, 2012

[tmp1; xout2(in, 1)];. This has impact on computing the Conditional Joint PDF of X and Z; consequently the appearence of the gures.

You might also like