Optimal distance query reconstruction for graphs without long induced cycles
share
Let G=(V,E) be an n-vertex connected graph of maximum degree Δ. Given access to V and an oracle that given two vertices u,v∈ V, returns the shortest path distance between u and v, how many queries are needed to reconstruct E? We give a simple deterministic algorithm to reconstruct trees using Δ nlog_Δ n+(Δ+2)n distance queries and show that even randomised algorithms need to use at least 1/200Δ nlog_Δ n queries in expectation. The best previous lower bound was an information-theoretic lower bound of Ω(nlog n/loglog n). Our lower bound also extends to related query models including distance queries for phylogenetic trees, membership queries for learning partitions and path queries in directed trees. We extend our deterministic algorithm to reconstruct graphs without induced cycles of length at least k using O_Δ,k(nlog n) queries, which includes various graph classes of interest such as chordal graphs, permutation graphs and AT-free graphs. Since the previously best known randomised algorithm for chordal graphs uses O_Δ(nlog^2 n) queries in expectation, we both get rid off the randomness and get the optimal dependency in n for chordal graphs and various other graph classes. Finally, we build on an algorithm of Kannan, Mathieu, and Zhou [ICALP, 2015] to give a randomised algorithm for reconstructing graphs of treelength k using O_Δ,k(nlog^2n) queries in expectation.
READ FULL TEXT