Geometric characterization of the Eyring-Kramers formula (Q6080998)

When studying diffusion processes (a particle moving randomly) in one or more dimensions, a special class is the one where there is a function \(F\) (called the potential) on the state-space that determines which sites are more or less likely to be visited. Such function \(F\) draws a landscape (with mountains and valleys) on the state-space such that the particle is attracted to the valleys and pushed away from the top of a mountain. The Eyring-Kramers formula is the mean time that particle takes of going from one bottom of one valley (local minima) to other local minima, assuming the existence of a saddle point in the way between the local minima. Such formula depends on the specific information of the potential, such as certain Hessian coming from \(F\), as well as an eigenvalue. In this paper the authors provide some extensions of the formula, when the potential has some degeneration not considered in the original formula, like having several saddle points of the same height, flat minima or flat saddle points. The exposition needs tools from potential theory and differential geometry