Edge- and Node-Disjoint Paths in P Systems

In this paper, we continue our development of algorithms used for topological network discovery. We present native P system versions of two fundamental problems in graph theory: finding the maximum number of edge- and node-disjoint paths between a source node and target node. We start from the standard depth-first-search maximum flow algorithms, but our approach is totally distributed, when initially no structural information is available and each P system cell has to even learn its immediate neighbors. For the node-disjoint version, our P system rules are designed to enforce node weight capacities (of one), in addition to edge capacities (of one), which are not readily available in the standard network flow algorithms.

💡 Research Summary

The paper presents distributed algorithms for computing the maximum number of edge‑disjoint and node‑disjoint paths between a source and a target cell in a P system, specifically using a restricted model called simple P modules. A simple P module consists of a finite set of cells (each with a unique ID, a multiset of objects, a state, and a finite ordered set of rewriting rules) and a binary relation defining duplex communication links between parent‑child cell pairs. Initially, cells have no knowledge of their neighbors; therefore the authors first devise a breadth‑first search (BFS) phase that propagates cell IDs throughout the network, allowing every cell to locally reconstruct the underlying digraph (its “neighborhood table”).

After the topology discovery, the core of the work adapts the classic Ford‑Fulkerson maximum‑flow method to the distributed, synchronous setting of P systems. The authors employ a depth‑first search (DFS) to locate augmenting paths in the residual digraph. Each cell maintains the current set of flow‑paths that pass through it and updates a local view of the residual graph after every augmentation. When an augmenting path is found, the algorithm reverses the arcs along that path (the “push‑back” operation) and merges the new path with the existing flow, thereby increasing the cardinality of the edge‑disjoint path set by one. This process repeats until no further augmenting path exists, at which point the algorithm terminates with a maximal edge‑disjoint collection.

To obtain node‑disjoint paths, the paper does not use the conventional node‑splitting transformation (which would double the number of cells, impossible in a P‑system where each node corresponds to a physical cell). Instead, the authors enforce a capacity‑one constraint on every non‑terminal cell directly in the rule set: each cell is allowed to forward at most one unit of flow. This is achieved by rules with a “min” application operator (ensuring a rule fires at most once per round) and a “l” (local broadcast) transfer operator that replicates a token to all neighbors. Consequently, any set of edge‑disjoint paths that respects these capacity rules is automatically node‑disjoint.

The algorithmic complexity is analyzed as follows. The initial neighbor discovery requires O(|V|) synchronous rounds, each cell broadcasting its ID once. The augmenting‑path phase runs for at most k+1 stages, where k is the final number of disjoint paths; each stage involves a DFS that may examine O(|E|) arcs, yielding an overall worst‑case time of O(k·|E|) rounds. Because all cells operate in parallel and only local information is stored (O(degree) per cell), the approach scales well to large networks.

The authors discuss several motivating applications. In bandwidth‑critical routing, edge‑disjoint paths enable parallel data streams for multi‑core streaming tasks. In Byzantine agreement, Menger’s theorem guarantees that 2k+1 node‑disjoint paths between any pair of nodes suffice to tolerate k faulty nodes; the presented algorithm can construct such paths automatically in non‑complete graphs. The method also applies to neural or vascular remodeling scenarios where alternative routes must be identified after damage.

In conclusion, the paper successfully translates classical network‑flow techniques into the native language of P systems, handling both edge‑ and node‑disjoint path problems without prior knowledge of the network topology. It demonstrates that distributed, rule‑based computation can achieve optimal routing structures in a fully decentralized manner, and it opens avenues for future work on asynchronous P systems, weighted graphs, and dynamic topology adaptation.