In this paper we present a solution to the long-standing problem of characterising the coarsest liveness-preserving pre-congruence with respect to a full (TCSP-inspired) process algebra. In fact, we present two distinct characterisations, which give rise to the same relation: an operational one based on a De Nicola-Hennessy-like testing modality which we call should-testing, and a denotational one based on a refined notion of failures. One of the distinguishing characteristics of the should-testing pre-congruence is that it abstracts from divergences in the same way as Milner’s observation congruence, and as a consequence is strictly coarser than observation congruence. In other words, should-testing has a built-in fairness assumption. This is in itself a property long sought-after; it is in notable contrast to the well-known must-testing of De Nicola and Hennessy (denotationally characterised by a combination of failures and divergences), which treats divergence as catrastrophic and hence is incompatible with observation congruence. Due to these characteristics, should-testing supports modular reasoning and allows to use the proof techniques of observation congruence, but also supports additional laws and techniques.

Traditionally, in process calculi, relations over open terms, i.e., terms with free process variables, are defined as extensions of closed-term relations: two open terms are related if and only if all their closed instantiations are related. Working in the context of bisimulation, in this paper we study a different approach; we define semantic models for open terms, so-called conditional transition systems, and define bisimulation directly on those models. It turns out that this can be done in at least two different ways, one giving rise to De Simone's formal hypothesis bisimilarity and the other to a variation which we call hypothesis-preserving bisimilarity (denoted t fh and t hp, respectively). For open terms, we have (strict) inclusions t fh /t hp / t ci (the latter denoting the standard ``closed instance' ' extension); for closed terms, the three coincide. Each of these relations is a congruence in the usual sense. We also give an alternative characterisation of t hp in terms of nonconditional transitions, as substitution-closed bisimilarity (denoted t sb). Finally, we study the issue of recursion congruence: we prove that each of the above relations is a congruence with respect to the recursion operator; however, for t ci this result holds under more restrictive conditions than for tfh and thp.]

We investigate criteria to relate specifications and implementations belonging to conceptually different levels of abstraction. For this purpose, we introduce the generic concept of a vertical implementation relation, which is a family of binary relations indexed by a refinement function that maps abstract actions onto concrete processes and thus determines the basic connection between the abstraction levels. If the refinement function is the identity, the vertical implementation relation collapses to a standard (horizontal) implementation relation. As desiderata for vertical implementation relations we formulate a number of congruence-like proof rules (notably a structural rule for recursion) that offer a powerful, compositional proof technique for vertical implementation. As a candidate vertical implementation relation we propose vertical bisimulation. Vertical bisimulation is compatible with the standard interleaving semantics of process algebra; in fact, the corresponding horizontal relation is rooted weak bisimulation. We prove that vertical bisimulation satisfies the proof rules for vertical implementation, thus establishing the consistency of the rules. Moreover, we define a corresponding notion of abstraction that strengthens the intuition behind vertical bisimulation and also provides a decision algorithm for finite-state systems. Finally, we give a number of small examples to demonstrate the advantages of vertical implementation in general and vertical bisimulation in particular. 1