Smoluchowski-Kramers approximation and large deviations for infinite-dimensional nongradient systems with applications to the exit problem (Q317476)

This paper deals with the stochastic wave equation in a bounded regular region \(D\) in \(R^d\), \(d\geq 1\), depending on two small parameters \(\varepsilon\) and \(\mu\) (representing the constant density of the particles), the solution of which describes the displacement field of the particles of a material continuum in the domain \(D\), subject to a random external force field and a damping force proportional to the velocity field. The Laplacian describes interaction forces between neighbouring particles, in the presence of a nonlinear reaction described by additional term depending on the solution. The stochastic term is described by a cylindrical Wiener process, which is white in time and coloured in space. In the literature, it was shown before that for fixed \(\varepsilon>0\), the solution of this equation converges to the solution of the stochastic first-order equation as the density \(\mu\to 0\), and this approximation is known as the Smoluchowski-Kramers approximation. In the present paper, the authors are interested in comparing the small noise asymptotics, as \(\varepsilon\to 0,\) for the stochastic wave equation and the stochastic first-order equation. They show that the Smoluchowski-Kramers approximation, that works on finite time intervals, is good as well in the large deviations regime.