The large deviation principle for interacting dynamical systems on random graphs (Q2113491)

This paper studies large deviation results for interacting dynamical systems on random graphs. The \(n\) interacting particles follow the evolution \(\dot{u}_i^n=f(u_i^n,\xi_i^n,t)+n^{-1}\sum_{j=1}^nX_{ij}^nD(u_i^n,u_j^n)\), \(u_i^n(0)=g_i^n\), where \(1\le i\le n\), \(u_i^n\) is the state of particle \(i\), \(f\) is the intrinsic dynamics and \(D\) is the pairwise interaction between particles. Here, \(X_{ij}^n\) represents the adjacency matrix of the topology via W-random graphs, the state of each particle \(i\) is generally in \(\mathbb R^d\), \(\{\xi_i^n\}_{i=1}^n\) are random parameters and \(\{g_i^n\}_{i=1}^n\) depict the initial configuration. Some large deviation principles are established for this model and in particular in the setting of W-random graphs. The contraction principle concerning large deviation results is also established for the combinatorial and dynamical systems. Some generalizations and an open problem are provided concerning the main results. The lower semi-continuity of the rate function is also derived.