Decline and Fall of the ICALP 2008 Modular Decomposition algorithm
William Atherton, Dmitrii V. Pasechnik
TL;DR
The work examines the correctness of the ICALP 2008 linear-time modular decomposition method and presents a concrete counterexample graph $G$ that invalidates a crucial lemma. The counterexample shows that the refinement and promotion steps can mark children of a strong module not containing $x$, contradicting the lemma’s claim and leading to a missing prime node. The flaw is traced to Lemma 4 (Lemma 3.1 in the preprint), including a typographical issue ('not') that hides the true falsity; the example demonstrates the incorrect behavior on $G$ and on its complement $ackslashoverline{G}$. As a result, the authors note that Corneil 2024 adopts a LexBFS-based revision, underscoring the need for alternative strategies in linear-time modular decomposition.
Abstract
We provide a counterexample to a crucial lemma in the ICALP 2008 paper "Simpler Linear-Time Modular Decomposition Via Recursive Factorizing Permutations", invalidating the algorithm described there.
