Oseledets theorem

Oseledets theorem
Name	Oseledets theorem
Field	Ergodic theory; Dynamical systems; Linear algebra
Introduced	1968
Main contributors	Vladimir Oseledets
Related	Multiplicative ergodic theorem; Lyapunov exponent; Furstenberg–Kesten theorem

Oseledets theorem Oseledets theorem gives a structural decomposition for the long-term behavior of linear cocycles over measure-preserving transformations, providing almost-everywhere defined Lyapunov exponents and invariant subspaces in measurable dynamics. Originating in the work of Vladimir Oseledets, the theorem underpins developments connecting ergodic theory, differential equations, probability, and geometry through the study of stability, growth rates, and invariant splittings.

Statement of the theorem

For an invertible measure-preserving transformation on a probability space and a matrix-valued measurable cocycle with integrable logarithm of the norm, Oseledets theorem asserts that for almost every point there exist numbers (Lyapunov exponents) and a filtration of subspaces such that forward iterates grow or decay exponentially at the prescribed rates. In formal terms: given a probability space with an automorphism and a measurable map into GL(n,R) satisfying integrability, there exist real numbers λ1>...>λk and a measurable decomposition R^n = E1(x) ⊕ ... ⊕ Ek(x) invariant under the cocycle, with exponential growth rates λi almost everywhere. The statement ties to classical results such as the multiplicative ergodic theorem, and complements spectral theorems like the Perron–Frobenius theorem, the Birkhoff ergodic theorem, and Oseledets' own predecessors in probability and linear algebra.

Background and definitions

Key objects include a probability space with an ergodic or measure-preserving automorphism, a measurable cocycle taking values in GL(n,R), and the notion of Lyapunov exponents defined as almost-sure limits of scaled logarithms of norms of iterates. The setup references foundational work by Kolmogorov and Sinai in ergodic theory and by Lyapunov in stability theory; it uses integrability conditions related to Kingman’s subadditive ergodic theorem and connections to the Furstenberg–Kesten theorem for random matrix products. Definitions employ measurable selections, Oseledets subspaces, and invariant splittings analogous to stable and unstable manifolds in smooth dynamics studied by Anosov, Smale, and Sinai.

Proof outline and key ideas

The proof combines ergodic averages, subadditive ergodic arguments, and geometric linear-algebraic constructions to extract almost-sure exponential growth rates and invariant subspaces. One begins with Kingman’s subadditive ergodic theorem to obtain limit growth rates for exterior powers and then applies measurable selection theorems to produce invariant flags; the argument uses compactness in projective spaces akin to methods in Furstenberg’s work on random matrices and in Ledrappier’s studies of entropy. Technical steps deploy singular value decompositions, exterior algebra estimates similar to those in Weyl’s and Horn’s results, and measurable dependence results that echo von Neumann’s and Birkhoff’s ergodic theorems.

Examples and applications

Applications span stability analysis for linear differential equations studied by Lyapunov and Perron, random matrix products examined by Furstenberg and Kesten, and smooth dynamical systems including Anosov diffeomorphisms, Axiom A systems of Smale, and flows in homogeneous spaces considered by Ratner. In smooth ergodic theory, Oseledets splittings inform Pesin theory and nonuniform hyperbolicity developed by Pesin and Katok, and they appear in rigidity results connected to Margulis and Zimmer in homogeneous dynamics. Probabilistic and statistical physics applications include products of stochastic matrices in Markov chain contexts associated with Kolmogorov and Chapman, large deviations related to Cramér and Sanov, and Lyapunov exponent computations relevant to Anderson localization studied by Anderson and Thouless.

Extensions and related results

Generalizations include noninvertible and infinite-dimensional cocycles studied by Ruelle, Mañé, and Lian–Lu, multiplicative ergodic theorems for Banach spaces, and refinements such as the Oseledets–Ruelle theorem and Furstenberg’s criteria for simplicity of Lyapunov spectra. Connections appear with Zimmer’s cocycle superrigidity, Ledrappier–Young entropy formulas, and the invariance principles used by Avila, Viana, and Bonatti in nonuniform hyperbolicity and partial hyperbolicity. Further related results encompass simplicity criteria by Guivarc’h and Raugi, continuity questions explored by Bochi and Viana, and stochastic generalizations linked to Kifer and Arnold.

Historical context and attribution

The theorem was proved by Vladimir Oseledets in 1968 and situated within Soviet-era developments in ergodic theory alongside work by Kolmogorov, Sinai, and Rokhlin; it integrated ideas from Lyapunov’s 19th-century stability theory and from mid-20th-century advances by Furstenberg, Kesten, and Kingman. Subsequent decades saw major contributions and extensions by Ruelle, Pesin, Ledrappier, Mañé, and many others, leading to broad influence across dynamical systems, probability, and mathematical physics through collaborations and developments involving Margulis, Zimmer, Katok, and contemporaries.

Category:Dynamical systems