Dynamic Causality in Event Structures

TL;DR

This paper introduces Dynamic Causality Event Structures, allowing causal dependencies between events to change during execution, enhancing the modeling of dynamic system behaviors beyond static causality models.

Contribution

It proposes a novel class of event structures with dynamic causality, analyzing their expressive power and relationships to existing models.

Findings

Dynamic causality can involve removal or addition of dependencies.

Dynamic Causality ESs subsume some existing models.

They are incomparable with ESs for Resolvable Conflicts.

Abstract

Event Structures (ESs) address the representation of direct relationships between individual events, usually capturing the notions of causality and conflict. Up to now, such relationships have been static, i.e., they cannot change during a system run. Thus, the common ESs only model a static view on systems. We make causality dynamic by allowing causal dependencies between some events to be changed by occurrences of other events. We first model and study the case in which events may entail the removal of causal dependencies, then we consider the addition of causal dependencies, and finally we combine both approaches in the so-called Dynamic Causality ESs. For all three newly defined types of ESs, we study their expressive power in comparison to the well-known Prime ESs, Dual ESs, Extended Bundle ESs, and ESs for Resolvable Conflicts. Interestingly, Dynamic Causality ESs subsume Extended Bundle ESs and Dual ESs but are incomparable with ESs for Resolvable Conflicts.