Theorem 2.8 — Kolmogorov's criterion and the canonical invariant measure
ProvedSerfozoStochasticNetworks.kolmogorov_criterionLet be a non-negative rate function on a state space with the two-way communication property — exactly when — and suppose is irreducible, every state being reachable from every other along a path of positive rates. Write .
Then all of the following hold.
- is reversible if and only if Kolmogorov's criterion holds: for every and every sequence with ,
- is reversible if and only if the rate ratios are path-independent: for any two paths with the same initial state and the same final state, the products along them agree.
- **If is reversible then, for every choice of origin , there is a strictly positive with satisfying the detailed balance equations and given by
for every path from the origin to .**
The criterion is a condition on the rates alone, with no reference to an unknown measure, and the third part turns it into a construction: the invariant measure is read off as a product of rate ratios along any path.
Formalization Note Kolmogorov's criterion is asserted for arbitrary closed sequences, not merely for closed paths, exactly as the book states it. Under two-way communication the two readings coincide: if some rate along the sequence vanishes, so does its reverse, and both products are zero.
In the third part the measure is produced rather than defined: the conclusion asserts the existence of a that agrees with the ratio product along every path from the origin, which is the content of expression (2.9) without having to choose a path for each state. A path of length zero gives the empty product , consistent with .
Irreducibility is used for the third part, to reach every state from the origin, and for the implication from the criterion to reversibility. Reversibility here means the existence of some positive measure satisfying detailed balance; no normalization is claimed, and whether can be normalized to a probability distribution is not part of the statement.
The state space is an arbitrary type and an arbitrary non-negative function, so the statement applies verbatim to a discrete-time chain with read as transition probabilities, as the book points out.
import Mathlib import Definitions.Def_SerfozoStochasticNetworks_Reversible
namespace SerfozoStochasticNetworks
theorem kolmogorov_criterion {E : Type*} (q : E → E → ℝ)
(hq : ∀ x y, 0 ≤ q x y) (htw : TwoWay q) (hirr : IsIrreducible q) :
(IsReversible q ↔ KolmogorovCriterion q) ∧ (IsReversible q ↔ RatioInvariance q) ∧
(IsReversible q → ∀ x₀ : E, ∃ π : E → ℝ, (∀ x, 0 < π x) ∧ π x₀ = 1 ∧ DetailedBalance q π ∧
∀ (n : ℕ) (p : Fin (n + 1) → E), IsPath q p → p 0 = x₀ →
π (p (Fin.last n)) = pathRatio q p) := by sorry
end SerfozoStochasticNetworksConfirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.