4f00921114932db3f8662a41b44ee68f-AuthorFeedback.pdf

We greatly thank the reviewers for their constructive comments. Response to Reviewer 2. 1) The first-order Taylor approximation ofgk(α i) at bαi is for the purpose of giving a10 tangible form (polyhedron) of the confidence set. However, we can definitely evaluategk at everyαi and perform11 numerical inversion to find the confidence set. Figure 1: The CIs for two select edges at levelε = 0.05 over time T, for a Hawkes network with 5 nodes and influence functions are ϕij(t) = e t. In this picture, the solid blackline is trueαij and the dashed line is the zero level.