Classification and Threshold Dynamics of Stochastic Reaction Networks

The paper proves the 1D positive recurrence conjecture for weakly reversible mass-action SRNs and even establishes exponential ergodicity on each positive irreducible class (PIC). It does so by (i) giving threshold criteria for one-dimensional mass-action SRNs (Theorem 4.6), (ii) showing that endotactic networks satisfy R−>R+ (Theorem 4.11), and (iii) deducing that weakly reversible networks (being endotactic) are positive recurrent with exponentially ergodic stationary distributions on every PIC (Corollary 4.12). These claims appear explicitly in the paper’s abstract and Section 4.3 . The candidate solution mirrors this structure: it reduces the dynamics on each PIC to a 1D CTMC with polynomial jump rates, uses highest-degree dominance (forward/backward) to get a Foster–Lyapunov drift, and invokes weakly reversible ⇒ endotactic to satisfy the dominance, thus obtaining positive recurrence and exponential ergodicity. Notationally, the model’s d± correspond to the paper’s R±, and its leading-coefficient discussion aligns with the paper’s α(c),β(c) framework (Section 4.2), where for d=1 the relevant quantities simplify and the classification applies uniformly on PICs . Minor issues: the model mislabels the corollary number in one place (it is Cor. 4.12 in this version) and slightly overstates that catalysts do not affect leading-coefficient sums (they do rescale them on a PIC, though the strict degree dominance R−>R+ coming from endotacticity makes that tie-breaker unnecessary in this setting). Overall, the logic and conclusions match the paper’s proof strategy and results.