An Improved Randomized Data Structure for Dynamic Graph Connectivity

We present a randomized algorithm for dynamic graph connectivity. With failure probability less than 1/n (for any constant c we choose), our solution has worst case running time O(log n) per edge insertion, O(log n) per edge deletion, and O(log n/ log log n) per query, where n is the number of vertices. The previous best algorithm has worst case running time O(log n) per edge insertion and O(log n) per edge deletion. The improvement is made by reducing the randomness used in the previous result, so that we save a log n factor in update time. Specifically, [KKM13] uses log n copies of a data structure in order to boost a success probability from 1/2 to 1− n. We show that, in fact though, because of the special structure of their algorithm, this boosting via repetition is unnecessary. Rather, we can still obtain the same correctness guarantee with high probability by arguing via a new invariant, without repetition. School of Engineering and Applied Sciences, Harvard University, [email protected]. Supported by NSF Grant CCF-1350670.