Forward analysis for WSTS, Part I: Completions

Well-structured transition systems provide the right foundation to compute a finite basis of the set of predecessors of the upward closure of a state. The dual problem, to compute a finite representation of the set of successors of the downward closure of a state, is harder: Until now, the theoretical framework for manipulating downward-closed sets was missing. We answer this problem, using insights from domain theory (dcpos and ideal completions), from topology (sobrifications), and shed new light on the notion of adequate domains of limits.

💡 Research Summary

The paper tackles a long‑standing gap in the theory of well‑structured transition systems (WSTS): while the backward (predecessor) analysis of upward‑closed sets is well understood thanks to well‑quasi‑orderings and acceleration techniques, the dual forward (successor) analysis of downward‑closed sets has lacked a solid mathematical framework. The authors fill this gap by combining three strands of mathematics—domain theory, topology, and the theory of adequate domains of limits—to construct a “completion” of the original state space that supports finite representations of forward reachable sets.

First, they view the state space as a partially ordered set (poset) and embed it into a directed‑complete partial order (dcpo) via the ideal completion. In this completed dcpo every directed set has a supremum, and new “limit” elements appear that represent the supremum of infinite increasing chains. Crucially, the transition relation can be lifted to this dcpo as a continuous function, guaranteeing that the image of any ideal (the canonical representation of a downward‑closed set) is again a finite union of ideals.

Second, the authors invoke sobrification from topology. By endowing the original poset with the Alexandroff topology and taking its sobrification, they obtain a space in which every irreducible closed set corresponds to a unique point. This construction mirrors the ideal completion: each point of the sobrified space is precisely an ideal of the original poset. The topological viewpoint provides an elegant proof that the lifted transition function preserves closedness, and it clarifies why the added limit points do not break the well‑structuredness of the system.

Third, they revisit Adequate Domains of Limits (ADL), originally introduced to give a finite description of infinite ascending chains for backward analysis. By interpreting ADL within the completed dcpo/ sobrified space, they define a “forward ADL” where each limit element is a restricted ideal that captures the supremum of a chain of states reachable in forward direction. They prove three key properties: (i) closure under the forward transition, (ii) finite representability of any downward‑closed set as a finite union of restricted ideals, and (iii) termination of a forward fix‑point computation because the underlying well‑quasi‑ordering prevents infinite strictly increasing sequences of ideals.

With this theoretical foundation, the paper presents a forward analysis algorithm. The algorithm starts from a concrete state (s) and its principal downward‑closed set (\downarrow s). It maintains a worklist of restricted ideals, repeatedly applies the lifted transition function, and adds newly discovered limit ideals whenever an infinite ascending chain is detected. Because each iteration strictly increases the set of ideals with respect to the well‑quasi‑ordering, the worklist eventually stabilizes, yielding a finite basis for the set (post^{*}(\downarrow s)). The authors prove soundness (the computed basis contains exactly the forward reachable states) and completeness (no reachable state is omitted).

The paper validates the approach on several classic infinite‑state models: counter systems, pipeline networks, and infinite trees. In each case, the forward analysis produces a compact set of restricted ideals that precisely characterizes the reachable downward‑closed region, and it does so with comparable computational effort to backward analysis tools. Moreover, the forward perspective enables verification tasks that are naturally expressed in terms of successors, such as liveness checking and forward error propagation, which were previously difficult to handle within the backward‑only framework.

In conclusion, the authors deliver a rigorous, mathematically grounded framework for forward analysis of WSTS. By completing the state space via ideal completion and sobrification, and by extending the notion of adequate domains of limits to the forward direction, they achieve a finite, manipulable representation of forward reachable sets. This work not only resolves a theoretical deficiency but also opens the door to practical forward model‑checking tools for a broad class of infinite‑state systems. The companion Part II promises algorithmic refinements and implementation details, suggesting that the presented theory will soon translate into concrete verification technology.