Non-equilibrium Particle Dynamics with Unbounded Number of Interacting Neighbors

Abstract

We consider an infinite system of first order differential equations in $${\mathbb {R}}^{\nu }$$ , parameterized by elements x of a fixed countable set $$\gamma \subset {\mathbb {R}}^d$$ , where the right-hand side of each x-equation depends on a finite but in general unbounded number $$n_x$$ of variables (a row-finite system). Such systems describe in particular (non-equilibrium) dynamics of spins $$q_x\in {\mathbb {R}}^{\nu }$$ of a collection of particles labelled by points $$x\in \gamma $$ . Two spins $$q_{x}$$ and $$q_{y}$$ interact via a pair potential if the distance between x and y is no more than a fixed interaction radius. In contrast to the case where $$\gamma $$ is a regular graph, e.g. $$\mathbb {Z}^d$$ , the number $$n_x$$ of particles interacting with particle x can be unbounded in x. Our main example of a “growing” configuration $$\gamma $$ is a typical realization of a Poisson (or Gibbs) point process. Under certain dissipativity-type condition on the right-hand side of our system and a bound on growth of $$n_x$$ , we prove the existence and (under additional assumptions) uniqueness of infinite lifetime solutions with explicit estimates of growth in parameter x and time t. For this, we obtain uniform estimates of solutions to approximating finite systems using a version of Ovsyannikov’s method for linear systems in a scale of Banach spaces. As a by-product, we develop an infinite-time generalization of the Ovsyannikov method.