Time Delays in Membrane Systems and Petri Nets

Timing aspects in formalisms with explicit resources and parallelism are investigated, and it is presented a formal link between timed membrane systems and timed Petri nets with localities. For both formalisms, timing does not increase the expressive power; however both timed membrane systems and timed Petri nets are more flexible in describing molecular phenomena where time is a critical resource. We establish a link between timed membrane systems and timed Petri nets with localities, and prove an operational correspondence between them.

💡 Research Summary

The paper investigates the incorporation of explicit time delays into two well‑established formalisms for modeling distributed, parallel systems with explicit resources: membrane (or P) systems and Petri nets equipped with localities. Both formalisms naturally capture the hierarchical compartmentalisation of a cell (membranes) and the concurrent execution of rules or transitions, but they originally lack a direct notion of time. The authors first extend membrane systems with a global clock and a function e that assigns an integer execution time to each evolution rule. A rule that starts at clock tick j finishes at tick j+e(r); the objects produced become available only at the next tick. The evolution of a timed membrane system proceeds in two phases at each tick: (i) a maximal parallel rewriting step (mpr) where all enabled rules are applied simultaneously, and (ii) a parallel communication step (tar) that moves the produced messages across membranes. If no rule can fire, the system halts.

The central theoretical contribution is Proposition 1, which shows that timing does not increase the expressive power of membrane systems. For any timed membrane system Π, the authors construct an untimed membrane system Π′ that simulates Π exactly on the original alphabet V. The construction introduces auxiliary symbols a_j (0 ≤ j < m, where m is the maximum rule delay) to encode the passage of time. Rules with delay e(r) > 0 are replaced by a chain of “clock” rules a_j → a_{j‑1} together with a rule that finally produces the original object after the required number of steps. By induction on the number of clock ticks, each timed transition of Π corresponds to a finite sequence of untimed transitions of Π′, and the multisets of original objects coincide at every step. Hence, timed and untimed membrane systems generate the same class of configurations.

Next, the paper defines timed Petri nets with localities. A locality is a label attached to each transition, mirroring the membrane label in a P system. Transitions are equipped with an integer delay function d: T → ℕ; when a transition fires, it consumes its input tokens, waits d(t) time units, and then produces its output tokens. The net’s places correspond to the compartments of the membrane structure, and arcs encode the movement of objects (here, out, in j) exactly as in the membrane rules.

The authors then establish an operational correspondence between timed membrane systems and timed Petri nets with localities. The translation maps each membrane i to a set of places, each rule r∈R_i to a transition t_r with locality i, and the rule’s left‑hand side multiset to the transition’s input arcs, while the right‑hand side multiset (including target annotations) becomes the output arcs. The delay e(r) of the rule is assigned as the transition’s delay d(t_r). Under this mapping, a single timed step C ⇒ C′ in the membrane system is simulated by a sequence of timed transition firings in the Petri net that respects the same global clock, and conversely any maximal concurrent firing sequence of the net corresponds to a valid maximal parallel rewriting step in the membrane system. This bidirectional simulation proves that the two timed formalisms are computationally equivalent while preserving the explicit timing information.

The paper concludes that adding time does not increase computational power for either formalism, but it greatly enhances modeling convenience for biological processes where timing is a critical resource (e.g., protein lifetimes, immune cell maturation). The established link enables the transfer of analysis techniques and tool support between the two communities: Petri‑net based verification tools can be applied to membrane‑system models, and the hierarchical intuition of membranes can guide the design of structured Petri nets. Future work is suggested on stochastic timing, non‑deterministic delays, and large‑scale case studies.