On the Performance of Dynamic Matching Models withThreshold-based Policies

We consider a dynamic matching model where a bipartite graph determines the compatibility among two set of nodes, usually referred to as ''supply'' and ''demand'' nodes. Items arrive to these nodes following a stochastic process and are matched by pairs - and leave the system - according to a matching policy. Items that are not matched queue up until they are matched later on, which entails a holding cost.