Lorenz '96 model

Lorenz/Emmanuel 1996 chaotic model usually used to test data assimilation systems.

Description

X = lorenz4D(tf,F,pert)

In general, the Lorenz model is: (cyclical)
dX[j]/dt=(X[j+1]-X[j-2])*X[j-1]-X[j]+F
We use
J=40; %the number of variables
h=0.05; %the time step, equivalent to 6 hours

For the imput: tf is final time, F is the perturbation generally chosen to be 8 and pert is the perturbation (in the 20th variable) which is of the order of 10^-3.

• MATLAB release: MATLAB 7.3 (R2006b)

[Javier Amezcua]

By the way, the model is solved using Runge-Kutta 4 step method.

[Mike]

 

[Mike]

needs "nargin" early on in case someone doesn't know default parameters.

needs global used to relate changes in J in main program to J values in subroutines.

Why not use the built-in ode45?