Scaling Methods for Stochastic Chemical Reaction Networks

The asymptotic properties of some Markov processes associated to stochastic chemical reaction networks (CRNs) driven by the kinetics of the law of mass action are analyzed. The scaling regime introduced in the paper assumes that the norm of the initial state is converging to infinity. The reaction rate constants are kept fixed. The purpose of the paper is of showing, with simple examples, a scaling analysis in this context. The main difference with the scalings of the literature is that it does not change the graph structure of the CRN or its reaction rates. Several CRNs are investigated to illustrate the insight that can be gained on the qualitative properties of these networks. A detailed scaling analysis of a CRN with several interesting asymptotic properties, with a bi-modal behavior in particular, is worked out in the last section. Additionally, with several examples, we also show that a stability criterion due to Filonov for positive recurrence of Markov processes may simplify significantly the stability analysis of these networks.

💡 Research Summary

The paper investigates the asymptotic behavior of stochastic chemical reaction networks (CRNs) under a novel scaling regime in which the norm of the initial state ‖x‖ tends to infinity while all reaction rate constants κ_r remain fixed and the network topology is unchanged. This contrasts with the more common volume‑based scalings where both the reaction rates and sometimes the reaction graph depend on a scaling parameter N. By focusing on large initial populations, the authors aim to describe how a CRN returns to a neighborhood of the origin and to uncover qualitative features such as multiple time‑scales, boundary effects, and bi‑modal dynamics that naturally emerge when the rates are not rescaled.

The paper is organized as follows. Section 2 introduces formal definitions of CRNs, the mass‑action propensity functions, and the associated continuous‑time Markov chain X(t) on ℕⁿ. Section 3 studies binary CRNs (complexes contain at most two molecules). Using connections to M/M/∞ queue hitting times, the authors show that for certain initial directions the scaled process X₁(Nt)/N converges to a deterministic linear decay while X₂(t)·N⁻¹ collapses rapidly. Different initial directions may require distinct time‑scales (e.g., N^{1/2}t), illustrating that the limit depends on the direction x/‖x‖.

Section 4 revisits a network introduced by Agazzi and Mattingly. In the original work a sophisticated Lyapunov function was constructed to prove positive recurrence. The present authors demonstrate that Filonov’s theorem—a less‑known result stating that a negative drift after a suitable stopping time suffices for positive recurrence—allows one to prove the same property with a simple linear test function. This dramatically simplifies the stability analysis and also yields a scaling picture of the network’s evolution.

Section 5 contains the most intricate example: a CRN that exhibits bi‑modal behavior. The reactions are
∅ → S₁+S₂ (rate κ₀),
p S₁+S₂ → … (rate κ₁),
p S₁+2 S₂ → … (rates κ₂, κ₃), with p ≥ 2.
The second reaction is blocked when the count of S₁ is below p, introducing a strong boundary effect. Two families of initial conditions are examined.

1. Initial state (N, b) with b fixed. After scaling time by N, the first component satisfies
   X₁(Nt)/N → 1 − t/t_∞, t_∞ = κ₀/(κ₂/κ₃ − 1),
   so the decay is linear in time.

2. Initial state (a, N) with a < p. Scaling time by N^{p‑1} yields convergence of the second component to a stochastic process V(t) on (0, 1] with generator
   A f(x) = r₁ x^{p‑1} ∫₀¹