Deterministic Dynamic Maximal Matching in Sublinear Update Time

We give a fully dynamic deterministic algorithm for maintaining a maximal matching of an n-vertex graph in Õ(n8/9) amortized update time. This breaks the long-standing Ω(n)-update-time barrier on dense graphs, achievable by trivially scanning all incident vertices of the updated edge, and affirmatively answers a major open question repeatedly asked in the literature Baswana, Gupta and Sen [FOCS 2011], Bhattacharya,Chakrabarty, Henzinger and Nanongkai [SODA 2018], Solomon [Dagstuhl]. We also present a faster randomized algorithm against an adaptive adversary with Õ(n3/4) amortized update time. Our approach employs the edge degree constrained subgraph (EDCS), a central object for optimizing approximation ratio, in a completely novel way; we instead use it for maintaining a matching that matches all high degree vertices in sublinear update time so that it remains to handle low degree vertices rather straightforwardly. To optimize this approach, we employ tools never used in the dynamic matching literature prior to our work, including sublinear-time algorithms for matching high degree vertices, random walks on directed expanders, and the monotone Even-Shiloach tree for dynamic shortest paths.