Dynamic Sum-Radii Clustering
Real networks have in common that they evolve over time and their dynamics have a huge impact on their structure. Clustering is an efficient tool to reduce the complexity to allow representation of the data. In 2014, Eisenstat et al. introduced a dynamic version of this classic problem where the distances evolve with time and where coherence over time is enforced by introducing a cost for clients to change their assigned facility. They designed a \(\varTheta (\ln n)\)-approximation. An O(1)-approximation for the metric case was proposed later on by An et al. (2015). Both articles aimed at minimizing the sum of all client-facility distances; however, other metrics may be more relevant. In this article we aim to minimize the sum of the radii of the clusters instead. We obtain an asymptotically optimal \(\varTheta (\ln n)\)-approximation algorithm where n is the number of clients and show that existing algorithms from An et al. (2015) do not achieve a constant approximation in the metric variant of this setting.
KeywordsFacility location Approximation algorithms Clustering Dynamic graphs
- 5.Tantipathananandh, C., Berger-Wolf, T.Y., Kempe, D.: A framework for community identification in dynamic social networks. In: SIGKDD, pp. 717–726 (2007)Google Scholar
- 11.An, H., Norouzi-Fard, A., Svensson, O.: Dynamic facility location via exponential clocks. In: SODA, pp. 708–721 (2015)Google Scholar
- 12.Fernandes, C.G., Oshiro, M.T., Schabanel, N.: Dynamic clustering of evolving networks: some results on the line. In: AlgoTel, pp. 1–4, May 2013Google Scholar
- 16.Dinur, I., Steurer, D.: Analytical approach to parallel repetition. In: STOC, pp. 624–633 (2014)Google Scholar
- 17.Lee, Y.T., Sidford, A.: Path finding methods for linear programming: solving linear programs in Õ(vrank) iterations and faster algorithms for maximum flow. In: Proceedings of the 2014 IEEE 55th Annual Symposium on Foundations of Computer Science, FOCS 2014, pp. 424–433. IEEE Computer Society, Washington (2014)Google Scholar