Almost Optimal Fully Dynamic $k$-Center Clustering with Recourse

Authors: Sayan Bhattacharya, Martin Costa, Ermiya Farokhnejad, Silvio Lattanzi, Nikos Parotsidis

ICML 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical Our contribution: We give an algorithm for dynamic kcenter that answers this question in the affirmative, obtaining the following result. Theorem 1.1 (informal). There is an algorithm for dynamic k-center that maintains a 20-approximation with O(k log5(n) log ) update time and O(1) recourse. We obtain our result by combining a simple variant of the algorithm of (Bateni et al., 2023) with the dynamic sparsification algorithm of (Bhattacharya et al., 2023a).
Researcher Affiliation Collaboration 1Department of Computer Science, University of Warwick 2Google Research. Correspondence to: Sayan Bhattacharya <EMAIL>, Mart ın Costa <EMAIL>, Ermiya Farokhnejad <EMAIL>, Silvio Lattanzi <EMAIL>, Nikos Parotsidis <EMAIL>.
Pseudocode Yes Algorithm 1 Reconstruct 1: for i = 1 to ℓdo 2: COUNT[i] = COUNT[i] + 1 3: end for 4: j = smallest index s.t. COUNT[j] |Uj|/4 5: for i = j to ℓdo 6: COUNT[i] = 0 7: end for 8: repeat 9: for m = 1 to M = Θ(log n) do 10: (Am, Bm) Almost Cover(Uj) 11: end for 12: m arg min1 m M cl(Am, Uj \ Bm) 13: (Sj, Uj+1) (Am , Bm ) 14: j = j + 1 15: until |Uj| 16k 16: ℓ= j 17: U := S1 Sℓ 1 Uℓ
Open Source Code No The text is ambiguous or lacks a clear, affirmative statement of release.
Open Datasets No The paper focuses on theoretical algorithm design within an abstract 'metric space (V, d)' and does not describe or use any specific datasets for empirical evaluation.
Dataset Splits No The paper is theoretical and does not conduct experiments on specific datasets, therefore, no dataset splits are mentioned.
Hardware Specification No The paper is theoretical and focuses on algorithm design and analysis, not empirical experiments. Therefore, no hardware specifications are mentioned.
Software Dependencies No The paper discusses theoretical algorithms and components like 'Dynamic MIS' and 'Almost Cover' but does not specify any programming languages, software libraries, or their version numbers used for implementation.
Experiment Setup No The paper is theoretical and does not present empirical experiments. Consequently, there are no details on experimental setup, hyperparameters, or training configurations.