Export Reviews, Discussions, Author Feedback and Meta-Reviews

This paper proposes a model of spatio-temporal dynamics that models a global latent process that governs the interactions between high level clusters of points together with a local observed process in which interactions are decoupled from points outside of one's cluster. Both levels can be thought of as vector autoregressive models. The authors apply their method to modeling data from a numerical simulation of a geologic model of fluid flow under the earth's subsurface. In general, this was a technically strong paper and shows off some state of the art optimization techniques. On the other hand it was also a difficult paper to get through as it was notation heavy and quite dense.