We consider the dynamic complexity of some central graph problems such as Reachability and Matching and linear algebraic problems such as Rank and Inverse. As elementary change operations we allow insertion and deletion of edges of a graph and the modification of a single entry in a matrix, and we are interested in the complexity of maintaining a property or query. Our main results are as follows: 1. Reachability is in DynFO; 2. Rank of a matrix is in DynFO(+,×); 3. Maximum Matching (decision) is in non-uniform DynFO. Here, DynFO allows updates of the auxiliary data structure defined in first-order logic, DynFO(+,×) additionally has arithmetics at initialization time and non-uniform DynFO allows arbitrary auxiliary data at initialization time. Alternatively, DynFO(+,×) and non-uniform DynFO allow updates by uniform and non-uniform families of poly-size, bounded-depth circuits, respectively. The first result confirms a two decade old conjecture of Patnaik and Immerman [16]. The proofs rely mainly on elementary Linear Algebra. The second result can also be concluded from [7].
CITATION STYLE
Datta, S., Kulkarni, R., Mukherjee, A., Schwentick, T., & Zeume, T. (2015). Reachability is in DynFO. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 9135, pp. 159–170). Springer Verlag. https://doi.org/10.1007/978-3-662-47666-6_13
Mendeley helps you to discover research relevant for your work.