Convergence of the one-dimensional contact process with two types of particles and priority (Q2692537)

The paper studies a two-type contact process in one dimension with priority. The `classical' (one-type) contact process is an interacting particle system whose state space is \(\{0,1\}^{\mathbb{Z}}\): occupied (infected) sites infect nearby sites at rate \(\lambda\) and die out (are removed) at rate \(1\). This paper studies a version of this process with two types of particles, such that type-1 particles have priority over (cannot be replaced by) type-2 particles on \(\mathbb{Z}_{\leq 0}\), with the opposite priority holding on \(\mathbb{Z}_{>0}\). This process can be coupled to the original contact process via the Harris construction. For the one-contact process it was previously known that there is a critical value \(\lambda_C\) such that a non-trivial invariant measure \(\mu\) exists if and only if \(\lambda>\lambda_C\). The present paper studies the invariant measures for the two-type contact process in the supercritical regime (\(\lambda>\lambda_C\)). Let \(\mu_1\) be the measure on \(\{0,1,2\}^{\mathbb{Z}}\) supported on particle configurations with no type-2 configurations and such that the projection onto \(\{0,1\}^{\mathbb{Z}}\) is \(\mu\), and define similarly \(\mu_2\) with the roles of type-1 and type-2 particles reversed. The author shows that there are four extremal invariant measures: both particles dying out (\(\delta_\varnothing\)), \(\mu_1\), \(\mu_2\) and finally a fourth invariant measure \(\nu\). The fourth probability measure \(\nu\) is obtained as the invariant measure of the 2-type contact process \(\zeta^{1,2}\) started with each site of \(\mathbb{Z}_{\leq 0}\) occupied by type-1 particles and each site of \(\mathbb{Z}_{>0}\) by type-2 particles. The measure \(\nu\) is supported on configurations in \(\{0,1,2\}^\mathbb{Z}\) with infinitely many sites occupied that are right-finite for type 1 and left-finite for type 2. The proof of the main results is based on the Mountford-Sweet renormalisation, with technical lemmas proved in a supplemental article. The Mountford-Sweet renormalisation yields tail bounds on leftmost type-2 and rightmost type-1 particle of, which gives invariant measure \(\nu\) for this initial condition. To recover the invariant measure for general initial condition, the author proves that there are four cases: it's possible that both particle types die out (leading to the trivial invariant measure) and that the particles of either type die out (in which case the process converges to \(\mu_1\) or \(\mu_2\) according to which type survives). Finally, this leaves the case when both types of particles survive, in which case the invariant measure is shown to be a convex combination of \(\delta_\varnothing\), \(\mu_1\), \(\mu_2\) and \(\nu\).