Title:Fluctuation theorems for multipartite processes

Abstract:Many systems are naturally modeled as multiple co-evolving subsystems. In this paper I analyze the stochastic thermodynamics of such systems, formalizing them as multipartite processes. In many multipartite processes the rate matrix of each subsystem $i$ only depends on a proper subset of the remaining subsystems, known as the \textit{community} of $i$. The intersections among the communities of the different subsystems forms a network. Here I derive fluctuation theorems (FTs) for multipartite processes, formulated in terms of this community network. I first derive several \textit{vector} FTs, governing the joint probability of the entropy production (EP) of all of the communities. These vector FTs in turn imply a set of conditional FTs, for the probability of the overall system's EP conditioned on the EP of any single community. Some of these FTs involve extensions of mutual information, to characterize the statistical coupling among the communities induced by the community network.