[Paper Review] Markov Evolution of Continuum Particle Systems with Dispersion and Competition

Dmitri Finkelshtein, Yuri Kondratiev+2 more|arXiv (Cornell University)|Dec 5, 2011

Stochastic processes and statistical mechanics3 citations

TL;DR

This paper constructs a Markov evolution for infinite particle systems in $\mathbb{R}^d$ with dispersal and competition, showing that correlation functions evolve in a scale of Banach spaces and remain sub-Poissonian—indicating no clustering—when dispersion is subordinate to competition. The key result establishes stochastic stability and moment closure via a rigorous semigroup approach in configuration space.

ABSTRACT

We construct birth-and-death Markov evolution of states(distributions) of point particle systems in $\mathbb{R}^d$. In this evolution, particles reproduce themselves at distant points (disperse) and die under the influence of each other (compete). The main result is a statement that the corresponding correlation functions evolve in a scale of Banach spaces and remain sub-Poissonian, and hence no clustering occurs, if the dispersion is subordinate to the competition.

Motivation & Objective

• To develop a rigorous Markov process for infinite particle systems in continuous space $\mathbb{R}^d$ with birth-and-death dynamics driven by dispersal and competition.
• To address the closure problem in spatial moment equations by constructing a stochastic semigroup for correlation functions in a Banach space framework.
• To prove that correlation functions remain sub-Poissonian under competition-dominant conditions, ensuring no clustering occurs.
• To establish the existence and regularity of the stochastic semigroup via the sun-dual approach and $L^1$-type estimates.
• To provide a microscopic foundation for mesoscopic kinetic equations in spatial ecology by linking individual-based models to scaling limits.

Proposed method

• Model particle dynamics using a Markov generator with dispersal kernel $a_+(x,y)$ and competition kernel $a_-(x,y)$, defining birth and death rates via additive intensity.
• Define the state space as probability measures on configuration space $\Gamma(\mathbb{R}^d)$, with dynamics governed by Fokker-Planck-Kolmogorov-type equations.
• Use correlation functions $k_t^{(p)}$ to describe the evolution, evolving in a scale of Banach spaces $\mathcal{K}_\alpha$ with exponential weight $e^{-\alpha p}$.
• Construct the stochastic semigroup via the sun-dual of the generator, ensuring strong continuity and positivity preservation.
• Apply $L^1$-estimates and majorization techniques to control the growth of correlation functions, using the inequality $|G_t^{D}|^{(p)} \leq \|G_t^D\|_\alpha e^{-\alpha p}$.
• Prove convergence of correlation functions in $\mathcal{K}_{\alpha^*}$ by estimating integrals over expanding domains $\Lambda_n$ and using relative compactness arguments.

Experimental results

Research questions

• RQ1 Under what conditions does the Markov evolution of a continuum particle system with dispersal and competition preserve sub-Poissonian correlation functions?
• RQ2 How can the infinite-dimensional Fokker-Planck-Kolmogorov equation be rigorously solved in a Banach space framework for correlation functions?
• RQ3 What is the role of the competition kernel in preventing clustering and ensuring moment closure in individual-based models?
• RQ4 Can the stochastic semigroup for the correlation functions be constructed via the sun-dual method and shown to be strongly continuous?
• RQ5 How does the scaling limit of the system preserve chaos, and what conditions ensure convergence to a mesoscopic kinetic equation?

Key findings

• The correlation functions $k_t^{(p)}$ evolve in a scale of Banach spaces $\mathcal{K}_\alpha$, ensuring well-posedness and regularity of the dynamics.
• If dispersion is subordinate to competition, the correlation functions remain sub-Poissonian, implying no spatial clustering occurs.
• The stochastic semigroup generated by the Markov process is strongly continuous and conserves positivity in the Banach space setting.
• The construction relies on $L^1$-type estimates and majorization via $\|G_t^D\|_\alpha$, with convergence controlled by exponential weights $e^{-\alpha p}$.
• The limit $\Lambda_n \to \mathbb{R}^d$ and $N_l \to \infty$ yield uniform control over integrals, proving convergence of the semigroup in the strong operator topology.
• The method establishes a rigorous link between individual-based models and mesoscopic kinetic equations by preserving chaos under scaling.