A Case Study on Evaluating Encodings Between Process Calculi∗

Two process calculi may be compared in their expressive power using encodings between them. Not every such translation is meaningful and therefore quality criteria are necessary. There are several frameworks to assess the quality of encodings. A rather prominent approach consists of five criteria for valid encodings collected by Gorla. An alternative notion of validity up to a semantic equivalence was proposed by one of the authors. Here the quality of the encoding is measured by the strength of the semantic equivalence that makes it valid. In this paper we analyse two well-known translations from the synchronous into the asynchronous π-calculus, both meeting Gorla’s criteria, and evaluate the encodings’ validity up to different semantic equivalences. While Honda and Tokoro’s translation is valid only for the weakest of the considered behavioural equivalences, Boudol’s encoding also respects stronger ones. One could hence argue that Boudol’s translation is more faithful.