Title: Non-signaling Causal Hierarchy of General Multisource Networks

Abstract: Large-scale multisource networks have been employed to overcome the practical constraints that entangled systems are difficult to faithfully transmit over large distance or store in long time. However, a full characterization of the multipartite nonlocality of these networks remains out of reach, mainly due to the complexity of multipartite causal models. In this paper, we propose a general framework of Bayesian networks to reveal connections among different causal structures. The present model implies a special star-convex set of non-signaling correlations from multisource networks that allows constructing polynomial-time algorithm for solving the compatibility problem of a given correlation distribution and a fixed causal network. It is then used to classify the nonlocality originated from the standard entanglement swapping of tripartite networks. Our model provides a unified device-independent information processing method for exploring the practical security against non-signaling eavesdroppers on multisource quantum networks.