Downward conditional monotonicity gives survival and extinction for contact processes in random environments

Abstract: The concept of downward conditional monotonicity for the Markov-modulated Poisson process (MMPP) is introduced and used to derive the optimal stochastic domination of a standard Poisson point process. The maximum arrival rate for the Poisson process which allows this domination to exist is shown to be related to an eigenvalue extracted from the generator matrix of the quasi-birth--death (QBD) formulation of the MMPP. This allows derivation of survival and extinction regimes for a large family of contact processes whose infection and recovery rates vary over time according to an underlying random environment with a finite number of states. Direct comparison with standard contact processes which dominate from above and below accomplishes this.

• The paper introduces downward conditional monotonicity (DCM) for Markov-modulated Poisson processes, providing explicit inequalities to ensure stochastic domination.
• It establishes a spectral framework linking Poisson intensities to eigenvalues, enabling precise coupling of random environments with standard Poisson processes.
• It derives explicit survival and extinction criteria for contact processes in dynamic environments, unifying previous results and informing epidemic modeling.

Downward Conditional Monotonicity and Extinction/Survival Criteria for Contact Processes in Random Environments

Introduction and Overview

This paper investigates the long-term behavior—survival versus extinction—of contact processes on lattices when both infection and recovery rates fluctuate dynamically according to an independent finite-state Markov process (a Markov-modulated Poisson process, MMPP). This extends the classical contact process framework by incorporating more realistic temporal environmental randomness and establishes new sharp conditions under which stochastic domination arguments, essential for comparing such complex dynamics to standard Poisson processes, apply. Central to this analysis is the introduction of downward conditional monotonicity (DCM) for MMPPs, a conditional ordering property enabling tractable stochastic bounds.

Main Contributions

Formalization of Downward Conditional Monotonicity for MMPPs

A key conceptual innovation is the DCM property for MMPPs: informally, it asserts that if the environmental process begins in a distribution dominated from below by a certain equilibrium ("no arrival") distribution $v^*$, this domination persists (in a conditional sense) at later times, conditioned on no arrivals. Formally, DCM can be checked in terms of the generator $T$ of the environmental Markov process and the vector of state-dependent arrival rates $\alpha$, leading to an explicit and verifiable system of inequalities encapsulated in equation (1.8) of the abstract. All two-state MMPPs are automatically DCM; for $|S|\geq 3$, DCM characterizes a broad family of processes relevant for applications.

Stochastic Domination Theorem and Coupling with Poisson Processes

Using DCM, the paper proves that for coupled MMPPs, the largest Poisson intensity $\lambda$ such that a simple Poisson process can be stochastically dominated by the MMPP is computable as the minimal eigenvalue $\alpha^*$ of the matrix $D_\alpha-T$, where $D_\alpha$ is the diagonal matrix of state-dependent rates. This gives an optimal threshold for coupling and yields explicit, sharp bounds for stochastic comparisons. The main result (Theorem 1.2) asserts that, under DCM and monotonicity, $PPP(\lambda)$ is dominated by $N(X)$ (the MMPP's arrivals) iff $T$0, while $T$1 is dominated by $T$2 iff $T$3. These results subsume previous specialized cases in the literature and extend to arbitrary initial environmental distributions.

Survival and Extinction Regimes for Contact Processes in Markovian Random Environments

Applying the above structural results, the paper develops explicit, new survival and extinction criteria for the contact process driven by a sitewise, independently evolving environmental Markov process (CPMRE). The main result (Theorem 1.3) provides sufficient conditions for both almost sure extinction and positive probability of strong survival, formulated in terms of the eigenvalues ($T$4 and $T$5) associated with the infection and recovery MMPPs and the critical parameter $T$6 for the classic contact process on $T$7.

These conditions are:

• Survival: If the infection process is DCM and $T$8, the CPMRE survives strongly.
• Extinction: If the recovery process (after "permutation reversal") is DCM and $T$9, the CPMRE goes extinct.

These criteria are shown to be robust to arbitrary initial environmental distributions by using strict domination properties and careful "delaying arguments"—waiting for environments to reach stationarity before exploiting the structural stochastic domination.

Analytical Examples and Numerical Bounds

The paper provides explicit computations for two- and three-state random environments. For example, in the two-state case, closed-form expressions for $\alpha$0 and $\alpha$1 as functions of the environmental transition rates show precisely when survival and extinction occur as environmental parameters vary. In the three-state model, the DCM criterion is analyzed directly, and implications for parametric phase transitions are discussed.

Mathematical and Technical Strengths

• Fine-grained coupling tools: The analysis builds upon and generalizes stochastic domination results for point processes, providing explicit expressions and constructive couplings.
• Spectral Characterization: Connecting the optimal Poisson domination rate to eigenvalues of $\alpha$2 leverages matrix-analytic methods from queueing and interacting particle systems.
• Robustness to Environment: The conditions account for arbitrary initial distributions for the environmental process and avoid pathologies by leveraging strict stochastic ordering and the monotonicity structure.

The paper's technical machinery—particularly the incremental "mass-shift" arguments between probability vectors, and the delicate limiting arguments utilizing perturbed (strictly monotone) models—provides a means to bypass the more ad hoc comparison methods prevalent in prior work (e.g., block construction or percolation techniques).

Implications and Future Directions

The identification of DCM as a sufficient structural property for coupling provides a unifying framework extensible to broader classes of interacting particle systems with dynamically disordered environments. From a theoretical viewpoint, the explicit spectral conditions enable the prediction of phase transitions in heterogeneous spatial models. These results clarify, for the first time, a sharp, computable relationship between environmental dynamic parameters and population-level behavior in spatial epidemic models.

Numerical implications are strong: for any computable finite $\alpha$3-state Markov environment, the critical coupling can be explicitly determined via standard linear algebra routines, allowing for precise numerical phase diagrams.

Open theoretical directions include:

• Exploring the gap between sufficient DCM-based survival/extinction criteria and possibly necessary conditions.
• Extending to non-monotone or infinite-state random environments.
• Characterizing possible weak survival regimes (beyond strong survival/extinction).
• Applying similar domination techniques to more general interacting particle systems (beyond the contact process), e.g., voter models or exclusion processes.

Applied implications are most relevant for modeling epidemic spread and information transmission in temporally variable and locally heterogeneous environments, where environmental fluctuations cannot be treated as static percolations or mean-field perturbations.

Conclusion

The paper establishes DCM as a central property governing the tractability of survival and extinction questions for the contact process in dynamically disordered environments. Through eigenvalue analysis and structured coupling arguments, it provides explicit, practical, and robust criteria for survival/extinction, grounded in the theory of MMPPs and stochastic domination. These results bridge previously disconnected areas—interacting particle systems, queueing theory, and point process coupling—offering new tools for both rigorous analysis and computation in the study of stochastic spatial processes in random environments (2606.10257).