Spectral gap for stochastic energy exchange model with nonuniformly positive rate function (Q2354148)

In this paper, an interacting particle model called stochastic energy exchange model is studied. In one dimension, the model is described as a vector \((x_{1},\dots,x_{N})\) of positive values representing energies in \(N\) different nodes that evolve in time. After a rate \(\Lambda(x_{i},x_{i+1})\) the energy at site \(i\) changes from \(x_{i}\) to the value \((x_{i}+x_{i+1})\alpha\) and at site \(i+1\) to \((x_{i}+x_{i+1})(1-\alpha)\), where \(\alpha\in [0,1]\) is a random number. One property of this model is that the energy of the system, denoted by \(\mathcal{E}\), is preserved along the time. This process describes a continuous time Markov chain with a generator denoted by \(L\). In this paper, it is assumed that \(\Lambda(a,b)\geq c (a+b)^{m}\) for constants \(c\) and \(m\). This means that the rate function is not uniformly bounded from below, and it is argued that this property for \(\Lambda\) is important to analyze interacting particle systems. One of the main purposes of studying particle systems are the so-called hydrodynamic limits, i.e., the convergence of the model after space and time scaling. For the particular model studied here, an important step towards a rigorous analysis of the hydrodynamic limit is to have sharp estimates of the spectral gap \(\lambda(\mathcal{E},N)\), which can be interpreted as the value of the first eigenvalue associated to \(L\). The main result of the paper is the inequality \(\lambda(\mathcal{E},N)\geq C\mathcal{E}^{m}/N^{2}\), for some constant \(C\). The strategy to obtain such an inequality is to reduce the model to compare it with other particle systems, where it becomes possible to analyze the spectral gap. In this case, the so-called long-range model is used to carry out the comparison. The paper also contains examples where its results can be applied.