arXiv preprint ·
Exact Sum-of-Radii Clustering under Any Constant Stability
- Published
- Source
- arXiv preprint, posted before peer review
- Original
- Read it at arXiv
- Topics
Abstract
We establish a separation between the stability thresholds of Minimum Sum-of-Radii (MSR) clustering and two closely related objectives. Known hardness results show that, on general metrics, exact recovery for Minimum Sum-of-Diameters and symmetric $k$-center is hard for every fixed $1 1$, extending Gao and Reyzin's result for $γ\ge2$. All three results use the same Bilu--Linial stability model: the optimal partition remains uniquely optimal when each pairwise distance is multiplied by any factor in $[1,γ]$, without requiring the perturbed distances to satisfy the triangle inequality. The number of clusters is part of the input. For $1<γ<2$, our algorithm runs in $n^{O((γ-1)^{-5})}$ time. The main structural insight is that a cheap merge in one part of the instance would allow a profitable split elsewhere under a suitable perturbation, contradicting stability. The resulting exchange inequality controls local cluster density, which gives linear growth and hence bounded treewidth in local interaction graphs. We use random partitions to obtain these regions, solve them by dynamic programming, and then derandomize the construction. For rational input and fixed rational $γ>1$, recognizing stability in this model is coNP-complete. Verifying that a supplied partition remains uniquely optimal under all allowed perturbations is coNP-complete for both arbitrary symmetric and metric perturbations.
The authors’ abstract, as arXiv lists it. arXiv’s listings, abstracts included, are free for anyone to reuse.
Read it on ReadSearch
Each opens after you create an account, and brings you back to this paper.
More in Data Structures & Algorithms
Get the few papers worth your time in Data Structures & Algorithms every morning.
ReadSearch reads the new papers in your field each day and brings you the few worth reading, each with its key finding in a sentence.