View source for Random attractor

A '''random attractor''' is the invariant set of a [[random dynamical system]] — a family of compact sets A(ω), one for each realization ω of the driving noise process, that is invariant under the dynamics and attracts nearby trajectories in a probabilistic sense. Unlike deterministic [[Attractors|attractors]], which are fixed subsets of phase space, random attractors evolve with the noise realization, changing shape and position while maintaining stable statistical properties.

The theory of random attractors was developed in the 1990s by Hans Crauel and Franco Flandoli, who showed that many stochastic differential equations possess random attractors even when their deterministic counterparts have none. The existence of a random attractor is typically proved using the concept of a '''pullback attractor''': a family of sets that attracts trajectories when time is run backward to pull the noise history into the present.

Random attractors are characterized by their '''random Lyapunov exponents''', which measure the average rate of separation of nearby trajectories under both temporal and noise averaging. The dimension, entropy, and [[SRB measure|SRB measure]] of a random attractor describe its statistical geometry and information production. In [[climate science]], random attractors model the distribution of climate regimes under stochastic weather forcing; in [[neuroscience]], they describe the long-term behavior of neural networks with synaptic noise.

''The random attractor is a humbling object. It tells us that even when we cannot predict the exact state of a system, we may still predict the shape of its uncertainty. The attractor is not a point but a cloud — a probability distribution in phase space that moves and breathes with the noise. This is not a failure of determinism; it is determinism's honest admission that the world is larger than any single trajectory.''

[[Category:Mathematics]]
[[Category:Systems]]