Chaos Lab

Tool

Lorenz '96 model

Lorenz's 1996 spatially extended chaos model. N variables sit on a ring; each variable advected by its neighbours and damped, with a constant forcing F. The standard testbed for data-assimilation and ML weather-emulator benchmarks.

F > 8 ⇒ chaos; F > 16 ⇒ stronger.

Lorenz '96

dx_i/dt = (x_{i+1} − x_{i−2}) · x_{i−1} − x_i + F     (ring topology, indices mod N)

Each row in the plot is one snapshot of the lattice; time flows downward.
Travelling waves and turbulent patterns dominate for F > 8.
Lorenz introduced this in 1996 as a testbed for atmospheric data assimilation.

FAQ

Frequently asked questions