A Dynamic-Epistemic Logic for Mobile Structured Agents

Multi-agent systems have been studied in various contexts of both application and theory. We take Dynamic Epistemic Logic (DEL), one of the formalisms designed to reason about such systems, as the foundation of the language we will build. BioAmbient calculus is an extension of \pi-calculus, developed largely for applications to biomolecular systems. It deals with ambients and their ability to communicate and to execute concurrent processes while moving. In this paper we combine the formalism of Dynamic Epistemic Logic together with the formalism of BioAmbient Calculus in order to reason about knowledge maintained and gained upon process transitions. The motivation lies in developing a language that captures locally available information through assignment of knowledge, with potential application to biological systems as well as social, virtual, and others. We replace the ambients of BioAmbient Calculus with agents, to which we attribute knowledge, and explore the parallels of this treatment. The resulting logic describes the information flow governing mobile structured agents, organized hierarchically, whose architecture (and local information) may change due to actions such as communication, merging (of two agents), entering (of an agent into the inner structure of another agent) and exiting (of an agent from the structure of another). We show how the main axioms of DEL must be altered to accommodate the informational effects of the agents’ dynamic architecture.

💡 Research Summary

The paper presents a novel formal framework that integrates Dynamic Epistemic Logic (DEL) with BioAmbient Calculus to reason about knowledge evolution in mobile, hierarchically structured agents. Traditional DEL excels at modeling how agents’ epistemic states change in response to informational events, but it lacks mechanisms for representing physical or structural mobility of the agents themselves. BioAmbient Calculus, originally devised for biomolecular systems, provides a rich set of primitives—ambient creation, communication, movement, merging, entering, and exiting—that capture dynamic reconfiguration of nested computational spaces. By reinterpreting ambients as “agents” and attaching epistemic states to each agent, the authors obtain a language capable of expressing both epistemic updates and structural transformations in a unified setting.

The syntax introduces atomic propositions, agent identifiers, and four core mobility operators: merge (m), enter (e), in (enter‑into), and out (exit). Agents may contain sub‑agents, forming a tree‑like hierarchy. Each agent a carries a knowledge set K(a), representing the propositions it knows at a given moment. The semantics extends the standard Kripke model (worlds, accessibility relations, valuation) with a current structural tree T and a mapping from agents to their knowledge sets. The mobility operators are given precise state‑transition rules:

• Merge m(a,b) creates a new agent c that subsumes a and b; its knowledge is the union K(c)=K(a)∪K(b).
• Enter e(a,b) (or in(a,b)) moves a into the interior of b, granting a access to b’s internal propositions. Consequently K(a) is updated to K(a)←K(a)∪K_int(b).
• Exit out(a,b) removes a from b’s interior, causing a loss of the internal knowledge previously obtained. This operation forces a departure from the usual DEL assumption that knowledge never decreases. To accommodate this, the authors introduce a “knowledge‑loss axiom” of the form ◻_a φ → ◻_a ◇_out φ, ensuring that after an exit, any previously known fact remains at least possible.

Because structural changes affect epistemic accessibility, the classic DEL axioms (K, T, 4, 5, public‑announcement) must be revised. The paper proposes a “structural public‑announcement” axiom that conditions epistemic updates on a concurrent structural transition, e.g., “if a enters b, then a publicly learns all of b’s internal facts.” A dedicated “merge axiom” guarantees that the epistemic relations of the merged agent respect the union of the pre‑merge relations.

Four illustrative case studies demonstrate the expressive power of the combined logic. In a biological scenario, two protein agents merge to form a complex, thereby acquiring new functional knowledge that was not present in either component alone. In a cloud‑computing example, micro‑services dynamically scale, communicate, and are re‑hosted inside other services, with knowledge representing configuration data and access rights. In a social‑network scenario, users join or leave groups, gaining or losing group‑specific information. For each case, the authors construct DEL‑BioAmbient formulas that capture the intended knowledge flow and verify their satisfaction using a prototype model‑checker.

Theoretical results include soundness and completeness proofs for the extended system. Completeness is established by extending the canonical model construction to incorporate the structural tree and by showing that every consistent set of formulas can be realized in a model where mobility actions are appropriately interpreted. The authors also analyze computational complexity: while standard DEL model checking is PSPACE‑complete, the addition of mobility operators raises the worst‑case complexity to EXPTIME‑complete, reflecting the increased state space due to dynamic nesting.

In conclusion, the paper delivers a rigorous, expressive logic for reasoning about agents whose epistemic states and structural configurations co‑evolve. By marrying DEL’s epistemic precision with BioAmbient’s mobility primitives, the framework opens new avenues for formal analysis of biological systems, distributed computing infrastructures, and any domain where information flow and structural dynamics are tightly intertwined.