Tight Bounds for Distributed Minimum-Weight Spanning Tree Verification

This paper introduces the notion of distributed verification without preprocessing. It focuses on the Minimum-weight Spanning Tree (MST) verification problem and establishes tight upper and lower bounds for the time and message complexities of this problem. Specifically, we provide an MST verification algorithm that achieves simultaneously O(m) messages and O($\sqrt$ n+D) time, where m is the number of edges in the given graph G, n is the number of nodes, and D is G’s diameter. On the other hand, we show that any MST verification algorithm must send {\Omega}(m) messages and incur {\Omega}($\sqrt$ n + D) time in worst case. Our upper bound result appears to indicate that the verification of an MST may be easier than its construction, since for MST construction, both lower bounds of {\Omega}(m) messages and {\Omega}($\sqrt$ n+D time hold, but at the moment there is no known distributed algorithm that constructs an MST and achieves simultaneously O(m) messages and O($\sqrt$ n + D) time. Specifically, the best known time-optimal algorithm (using O($\sqrt$ n + D) time) requires O(m + n 3/2) messages, and the best known message-optimal algorithm (using O(m) messages) requires O(n) time. On the other hand, our lower bound results indicate that the verification of an MST is not significantly easier than its construction.

💡 Research Summary

The paper introduces a novel model of distributed verification that does not rely on any preprocessing phase, and it focuses on the Minimum‑Weight Spanning Tree (MST) verification problem. In the classical CONGEST model, each node knows the weights of its incident edges and a local Boolean flag indicating whether each incident edge belongs to a candidate subgraph T. The goal is for every node to decide, without any prior certificates or labels, whether T is an MST of the underlying weighted graph G.

The authors present an algorithm that simultaneously achieves O(m) total messages and O(√n + D) rounds, where m is the number of edges, n the number of vertices, and D the diameter of G. The algorithm proceeds in several conceptual steps. First, it defines a modified weight function ω′ that incorporates the indicator bits: edges belonging to T receive a slightly smaller weight than any edge with the same original weight that is not in T. This transformation guarantees that T is an MST of the original graph if and only if it is an MST under ω′, thereby reducing verification to a standard MST computation under a uniquely weighted graph.

Next, the graph is partitioned into √n‑size fragments (clusters). Within each fragment the nodes locally compute a fragment MST using any linear‑time distributed MST routine; this costs O(√n) rounds and a linear number of messages confined to the fragment’s edges. The fragments are then abstracted into a “fragment graph” whose vertices represent fragments and whose edges correspond to inter‑fragment connections present in G. Representatives of each fragment exchange concise summaries of their internal MSTs over this fragment graph. Because the fragment graph’s diameter is bounded by the original diameter D, the inter‑fragment communication completes in O(D) rounds.

A second MST computation is performed on the fragment graph, again using a known time‑optimal sublinear‑time algorithm, which adds another O(√n) rounds. The resulting global MST is compared with the candidate T: the algorithm checks that no edge outside T is lighter than any edge on the unique path between its endpoints in T (i.e., T contains no cycle‑heavy edges) and that the total weight of T matches that of the computed MST. If both conditions hold, every node outputs 1; otherwise, they output 0.

The paper proves that the algorithm’s message complexity never exceeds O(m) because each physical edge participates in a constant number of transmissions across all phases. The round complexity is bounded by the sum of the fragment‑internal phase (O(√n)), the inter‑fragment dissemination (O(D)), and the final fragment‑graph MST phase (another O(√n)), yielding the claimed O(√n + D) bound.

To complement the upper bound, the authors establish matching lower bounds. The Ω(m) message lower bound follows from the classic communication‑complexity argument for MST construction: any verification must at least inspect each edge to ensure that no hidden lighter edge exists, which forces a linear number of messages in the worst case. The Ω(√n + D) time lower bound is derived by constructing a family of graphs where information about edge weights must travel across √n‑sized cuts and across the network’s diameter; any algorithm that finishes faster would violate information‑theoretic limits on how quickly distributed nodes can learn global weight ordering. Consequently, the lower bounds match the upper bounds up to polylogarithmic factors, establishing tight “Θ” results for both resources.

The significance of these results lies in showing that, despite intuition that verification might be substantially easier than construction, the fundamental limits for MST verification are essentially the same as for MST construction when no preprocessing is allowed. The work also opens a new line of inquiry into distributed verification without certificates, suggesting that similar tight bounds may be achievable for other classic graph problems (e.g., minimum cut, shortest paths) under the same model. Future research directions include extending the techniques to asynchronous or dynamic networks, exploring trade‑offs with limited preprocessing, and applying the methodology to randomized verification schemes.