Petri nets and dynamic causality for service-oriented computations

2021 36th Annual ACM/IEEE Symposium on Logic in Computer Science (LICS)

2021

Reversible prime event structures extend the wellknown model of prime event structures to represent reversible computational processes. Essentially, they give abstract descriptions of processes capable of undoing computation steps. Since their introduction, event structures have played a pivotal role in connecting operational models (traditionally, Petri nets and process calculi) with denotational ones (algebraic domains). For this reason, there has been a lot of interest in linking different classes of operational models with different kinds of event structures. Hence, it is natural to ask which is the operational counterpart of reversible prime event structures. Such question has been previously addressed for a subclass of reversible prime event structures in which the interplay between causality and reversibility is restricted to the so-called causerespecting reversible structures. In this paper, we present an operational characterisation of the full-fledged model and show that reversible prime event structures correspond to a subclass of contextual Petri nets, called reversible causal nets. The distinctive feature of reversible causal nets is that causality is recovered from inhibitor arcs instead of the usual overlap between post and presets of transitions. In this way, we are able to operationally explain also out-of-causal order reversibility.

Section: Causal Netsmentioning

confidence: 99%

A distributed operational view of Reversible Prime Event Structures

Melgratti

Mezzina

2021 36th Annual ACM/IEEE Symposium on Logic in Computer Science (LICS)

2021

“…Unravel nets. Unravel nets [9] are nets in which each computation gives rise to an acyclic subnet where each place has at most one incoming arc (hence an occurrence net). Given a net N = P, T, F, m 0 be a net and let T ⊆ T be a subset of transitions, then N T = P , T , F , m 0 where…”

Section: -Occurrence Occurrence and Unravel Nets Seen As Spread Netsmentioning

confidence: 99%

“…Another way to address this problem is to define an equivalence on some behaviors of (safe) Petri nets, which implies that the data structure adopted cannot be any longer the one devised for unfoldings or prefixes. The notion of unravel net introduced in [9] and [10] goes in this direction requiring that each execution is a partial order, but the overall structure gathering all executions does not need to be a partial order. Focussing on the dependencies to obtain a more compact data structure implies that the net representing more concisely the non sequential behaviours has to fulfill some properties that are posed on the net as a whole, e.g.…”

Section: Introductionmentioning

confidence: 99%

Spreading nets: A uniform approach to unfoldings

Journal of Logical and Algebraic Methods in Programming

Fabre

2020

Self Cite

This paper is devoted to the study of the notion of spread net. A spread net is a (safe) Petri nets where each place is annotated with some information, taken from a suitable information domain, about how such place can get marked. Spread nets generalize various kinds of nets used to represent the non sequential behaviours of a safe net, like unfoldings, merged processes or trellis processes. The spreading of a net may allow to produce more compact (partially unfolded) nets representing the non sequential behaviour of a net, depending on the chosen information domain. In particular in a spread net it is possible to merge not only conflicting runs, but also to partially refold time, as spread nets allow loops in time.

“…Another way to address this problem is to define an equivalence on some behaviors of (safe) Petri nets, which implies that the data structure adopted cannot be any longer the one devised for unfoldings or prefixes. The notion of unravel net introduced in [3] and [2] goes in this direction requiring that each execution is a partial order, but the overall structure does not need to be a partial order. Overcoming the request that (at least locally) the behavior should be represented using a partial order has led to the introduction of a reveal relation playing the role of causality [1,9].…”

Section: Introductionmentioning

confidence: 99%

Toward a Uniform Approach to the Unfolding of Nets

Fabre

Electron. Proc. Theor. Comput. Sci.

2018

Self Cite

In this paper we introduce the notion of spread net. Spread nets are (safe) Petri nets equipped with vector clocks on places and with ticking functions on transitions, and are such that vector clocks are consistent with the ticking of transitions. Such nets generalize previous families of nets like unfoldings, merged processes and trellis processes, and can thus be used to represent runs of a net in a true concurrency semantics through an operation called the spreading of a net. By contrast with previous constructions, which may identify conflicts, spread nets allow loops in time.