Petri Nets and Bio-Modelling - and how to benefit from their synergy

In this talk we are concerned with the intrinsic similarities and differences between Petri nets on the one hand, and membrane systems and reaction systems on the other hand.

💡 Research Summary

The paper investigates the deep structural and semantic relationships between three formal models of computation: classical Petri nets, membrane systems, and reaction systems. Petri nets are introduced as a well‑established framework for modeling concurrent and distributed processes, equipped with a rich toolbox for analysis, verification, and causal semantics. Membrane systems (also known as P systems) are described as a biologically inspired model that captures the compartmentalized nature of cellular chemistry: membranes define nested regions, and multiset rewriting rules describe chemical reactions inside and across these regions. Reaction systems are presented as a more recent abstraction that focuses on the qualitative interaction of reactions, emphasizing facilitation and inhibition without counting the multiplicity of molecules.

The authors first observe that both membrane systems and reaction systems share the core idea of multiset rewriting, which aligns naturally with the token‑based dynamics of Petri nets. Exploiting this commonality, they propose a systematic translation from basic membrane systems into Petri nets. The key technical device is the introduction of localities: each membrane is mapped to a place annotated with a locality label, and tokens represent the multiset of objects residing in that compartment. Transitions model the application of membrane rules, including object transport across membranes. However, adding locality breaks the standard causal semantics of Petri nets, because the original semantics assumes a globally synchronized firing of transitions. To restore a meaningful notion of causality, the paper introduces locally synchronized executions, a new execution model where transitions belonging to the same locality may fire simultaneously while preserving the partial‑order of events across different localities.

Next, the paper turns to reaction systems, whose qualitative nature makes the traditional multiset‑based Petri net unsuitable. In reaction systems, a reaction either occurs or does not, based solely on the presence or absence of its reactants and inhibitors. To capture this behavior, the authors define a novel class of Petri nets called set‑nets. In a set‑net, each place holds a set of tokens rather than a multiset; a transition is enabled only when its entire input set is contained in the current marking. This formulation mirrors the “all‑present” condition of reaction system rules and naturally encodes facilitation (presence of required elements) and inhibition (absence of forbidden elements). The paper further shows that by equipping set‑nets with localities, one obtains a direct correspondence with membrane systems that use qualitative evolution rules, i.e., rules that do not depend on multiplicities.

Throughout the manuscript, the authors reference a series of earlier works (including their own on Petri nets with localities, synchrony vs. asynchrony in membrane systems, and modelling reaction systems with Petri nets) to situate the current contributions within a broader research agenda. They demonstrate that the translation techniques are not merely syntactic encodings but preserve essential behavioral properties such as reachability, deadlock freedom, and causal dependencies.

The concluding discussion emphasizes the two‑way benefits of this synergy. On the one hand, the extensive analysis methods developed for Petri nets—reachability graphs, invariant analysis, model checking—become applicable to biological models, enabling formal verification of properties like conservation of mass, absence of unintended reactions, or guaranteed termination of a computation. On the other hand, the biological models motivate extensions to the Petri net formalism itself: locality leads to a richer notion of compartmentalized concurrency, while set‑nets introduce a qualitative dimension that challenges the traditional quantitative mindset of net theory. These extensions open new research directions in both theoretical computer science (e.g., studying the expressive power and decidability boundaries of set‑nets) and systems biology (e.g., providing a rigorous semantics for qualitative models of cellular processes).

In summary, the paper presents a comprehensive framework that bridges Petri nets, membrane systems, and reaction systems. By establishing faithful semantic mappings and proposing novel net extensions (localities and set‑nets), it enables cross‑fertilization of analysis techniques and inspires further theoretical developments, thereby advancing the state of the art in both formal methods and biologically inspired computation.