Ergodic Theory

Core idea

Ergodic theory studies measure-preserving dynamics through invariant events, time averages, recurrence, and mixing. These reading paths connect its probability foundations with Koopman representations, operator algebras, and arithmetic examples.

Start here

1. A probability-preserving system specifies the space, its measurable events, a probability measure, and a transformation.
2. The finite uniform example computes the three-point and general permutation cases explicitly.
3. Ergodicity says every invariant event has probability zero or one.
4. The Koopman representation turns the dynamics into a unitary action on functions.
5. The fixed-function criterion connects these viewpoints; Birkhoff's theorem identifies typical time averages.

Branches

• Probability spaces and dynamical systems — Start with measurable events and probabilities, then add transformations, invariant events, factors, and flows.
• Averaging, recurrence, and decomposition — The averaging theorems identify limits; recurrence forces returns; decomposition resolves a system into ergodic components.
• Koopman operators, representations, and spectrum — Translate dynamics into linear operators. Follow fixed vectors to ergodicity, then eigenfunctions, spectral measures, mixing, and continuous-time generators.
• Finite, toral, and Gaussian examples — Work from permutations to rotations and mixing maps, then transport cube dynamics into Gaussian coordinates.
• Number-field translations — Construct the Archimedean torus and its character lattice before comparing one translation with the joint additive action of the ring of integers.
• Noncommutative dynamics and orbit spaces — The function algebra and its state encode classical probability. Crossed products retain orbit motion; the rotation algebra links this construction to foliations and noncommutative differentiation.
• Entropy and growth along orbits — Entropy measures information in orbit observations. Subadditive and linear cocycles describe growth, leading to Kingman and Oseledets.
• Arithmetic, Hamiltonian, and data applications — Follow recurrence to arithmetic progressions, volume preservation to Hamiltonian Koopman dynamics, and finite observations to data-driven operator approximations.

Three conventions to keep in view

For the regularity and domain questions behind Lie-group generators, see Smooth and distribution vectors: a reading path.

• For forward time averages use $U_Tf=f\circ T$. A left group representation uses $\kappa(g)f=f\circ T_{g^{-1}}$. The inverse makes the order of group multiplication correct.
• A single number-field translation has a trace-integrality criterion. The joint ring-of-integers action has a different invariant-character test.
• The factor criterion for crossed products requires essential freeness. The regular crossed product uses an additional group-coordinate Hilbert space.