Maximum-Entropy-Temporal-Networks Public

Maximum Entropy Temporal Networks — Code Repository Implementation of the framework introduced in “Maximum Entropy Temporal Networks”(P. Barucca, arXiv:2509.02098) 🧩 Overview Temporal networks — i.e., systems of time-stamped, directed interactions evolving continuously — pose modelling challenges that static network theory cannot capture.The paper “Maximum Entropy Temporal Networks” introduces a principled maximum-entropy (MaxEnt) framework for continuous-time temporal networks. By maximizing the path entropy of a marked point process under structural and temporal constraints, the model yields a factorised form: [ \lambda_{ij}(t) = \phi_r(t); w_{ij}, ] where (r) indexes a partition of edges by type, (\phi_r(t)) is a time-dependent rate process for that partition, and (w_{ij}) are static mark weights representing the network structure. This decomposition allows independent control of temporal dynamics and structural correlations, leading to: closed-form likelihoods and expectations for event counts, node strengths, and motif statistics; analytical control of nonstationary dynamics, reciprocity, and burstiness; tractable inference of both static and temporal components from data. 👉 For full theoretical details, see the paper:Maximum Entropy Temporal Networks (Barucca, 2025)