Total: 1
We study Gaussian concentration inequalities for random fields obtained as finitary codings of i.i.d.\ processes, establishing a natural bridge between finitary codings and concentration inequalities. A finitary coding formalizes how a random field with nontrivial dependence, such as the Ising model, can be generated from an i.i.d.\ source by a shift-equivariant map that determines each output spin from a finite, configuration-dependent window of the input. Gaussian concentration yields uniform sub-Gaussian fluctuation bounds for all local observables. Our main abstract result shows that Gaussian concentration is preserved under finitary codings of i.i.d.\ fields provided the coding volume has finite second moment. The proof relies on a refinement of the classical worst-case bounded-differences inequality of Talagrand and Marton, allowing us to exploit configuration-dependent bounds. Under an additional structural condition that we call the short-range factorization property, satisfied in particular by codings constructed via coupling-from-the-past algorithms, finite first moment of the coding volume already suffices. These moment conditions are sharp. Applications include Gibbs measures and Markov random fields on $\mathbb Z^d$, probabilistic cellular automata, and one-dimensional stochastic processes. For classical lattice models such as the Ising, Potts, and random-cluster models, we obtain sharp criteria for Gaussian concentration coinciding with uniqueness regimes. For countable-state Markov chains, we derive equivalent characterizations, including geometric ergodicity, exponential return-time tails, and the existence of finitary i.i.d.\ codings with exponential tails.