Pesin entropy formula

The Pesin entropy formula, proved by Yakov Pesin in 1977, is the foundational theorem connecting the geometric instability of a smooth dynamical system to its information-theoretic complexity. The formula states that for a C^2 diffeomorphism preserving a measure with non-zero Lyapunov exponents, the Kolmogorov-Sinai entropy h_μ equals the sum of the positive Lyapunov exponents:

h_μ = Σ_{λ_i > 0} λ_i

This equality is not an approximation. It is an exact identity that holds for a broad class of systems, including those with hyperbolic structure and, via the work of Pesin and others, for systems with non-uniform hyperbolicity. The formula resolves a question that had been open since the 1960s: what is the relationship between the rate at which a system generates information and the rate at which nearby trajectories separate?

The Formula and Its Conditions

The Pesin entropy formula requires three conditions: the system must be a C^2 diffeomorphism (twice continuously differentiable), it must preserve a finite measure μ, and the Lyapunov exponents must be non-zero almost everywhere. The C^2 condition is essential: the formula fails for C^1 systems, as shown by counterexamples due to Pugh. The non-zero exponent condition ensures that the system is genuinely chaotic, not merely quasi-periodic or integrable.

The proof relies on the construction of local unstable manifolds and the absolute continuity of the foliation they generate. The key insight is that entropy measures the rate at which the system refines partitions, and this refinement is driven entirely by the expansion in unstable directions. There is no contribution from contracting or neutral directions.

Significance and Extensions

The Pesin entropy formula was the first rigorous bridge between the geometric theory of dynamical systems and the information theory of Kolmogorov and Sinai. It showed that the "information cost" of predicting a chaotic system is not an abstract statistical quantity but a direct consequence of the system's differential-geometric instability. The formula was later extended to systems with singularities, partially hyperbolic systems, and infinite-dimensional dynamics by Pesin and his collaborators.

The Ledrappier-Young formula generalizes Pesin's result by decomposing entropy according to the partial dimensions of the unstable manifolds. Where Pesin gives the total entropy as the sum of exponents, Ledrappier and Young show that each exponent contributes in proportion to the dimension of the corresponding invariant measure. This dimensional refinement is not merely a technical improvement; it reveals that entropy is a weighted sum, and the weights are geometric.

The Pesin entropy formula is often cited as a triumph of rigorous mathematics over physical intuition. But the deeper truth is the opposite: it is a triumph of physical intuition over the prejudice that information and geometry are separate domains. The formula proves that the rate at which a system becomes unpredictable is identical to the rate at which its phase space stretches. Information and geometry are not two languages describing the same thing. They are the same language, spoken with different accents.