FPT Parameterisations of Fractional and Generalised Hypertree Width

Abstract: We present the first fixed-parameter tractable (fpt) algorithms for precisely determining several central hypergraph decomposition parameters, including generalized hypertree width, fractional hypertree width, and adaptive width. Despite the recognized importance of these measures in complexity theory, databases, and constraint satisfaction, no exact fpt algorithms for any of them had previously been known. Our results are obtained for hypergraph classes of bounded rank and bounded degree. Our approach extends a recent algorithm for treewidth (Boja\'ncyk & Pilipczuk, LMCS 2022) utilizing monadic second-order (MSO) transductions. Leveraging this framework, we overcome the significant technical hurdles presented by hypergraphs, whose structural decompositions are technically much more intricate than their graph counterparts.