Quasi-potentials in the Nonequilibrium Stationary States or a method to get explicit solutions of Hamilton-Jacobi equations

Abstract

We assume that a system at a mesoscopic scale is described by a field $ϕ(x,t)$ that evolves by a Langevin equation with a white noise whose intensity is controlled by a parameter $1/\sqrtΩ$. The system stationary state distribution in the small noise limit ($Ω\rightarrow\infty$) is of the form $P_{st}[ϕ]\simeq\exp(-ΩV_0[ϕ])$ where $V_0[ϕ]$ is called the {\it quasipotential}. $V_0$ is the unknown of a Hamilton-Jacobi equation. Therefore, $V_0$ can be written as an action computed along a path that is the solution from Hamilton's equation that typically cannot be solved explicitly. This paper presents a theoretical scheme that builds a suitable canonical transformation that permits us to do such integration by deforming the original path into a straight line. We show that this can be done when a set of conditions on the canonical transformation and the model's dynamics are fulfilled. In such cases, we can get the quasipotential algebraically. We apply the scheme to several one-dimensional nonequilibrium models as the diffusive and reaction-diffusion systems.