Existence of MLEs for the parameters of an adjusted linear preferential attachment model of random graphs with covariates

In this talk, we consider a variation on the preferential attachment model (PAM) of random graphs with covariates. It is similar to the usual PAM but the probability of existence of an edge depends on the node/vertex-level covariates. For this parametric model, the likelihood function becomes complex and hard to solve analytically. In this work, we prove the conditions under which the maximum likelihood estimates (MLEs) are guaranteed to exist.
[Joint work with my student: Sungchan Park]

Share this Event