On Schematological Equivalence of Partially Interpreted Dataflow Networks

We consider the problem of determining when two dataaow networks over uninterpreted nodes and some primitive nodes always have the same input-output behavior. We deene a set of behavior-preserving transformations on networks and show that this set is \schematologically complete", i.e., networks have the same input-output behavior under all interpretations if and only if they can be transformed into isomorphic networks. As a by product, we obtain a polynomial algorithm for deciding schematological equivalence of dataaow networks. This paper extends our results in 12], were scehmatologically complete set of transformations was provided for the nets which consist only of uninterpreted nodes.