Adaptive Elimination in Curtailed Subset Selection: A Refined Procedure
Mathematics · 13.07.2026 · 14 (14) · 2518
DOI: 10.3390/math14142518 ↗Kurz gefasst
Fachartikel in der Zeitschrift Mathematics, online erschienen am 13.07.2026. Verfasst von Yifang Zhang und 1 weiteren Autor:innen. Die Arbeit nennt 15 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Subset selection from binomial populations is a classical problem in ranking and selection. In our earlier work, the curtailed procedure Rc introduced early stopping into this framework by terminating sampling once a current leader could no longer be overtaken. However, when several strong populations remain close, early stopping may be difficult to trigger and clearly inferior populations may continue to consume observations. To address this limitation, we propose an adaptive elimination refinement of Rc that permanently removes a population once its best possible future trajectory can no longer reach the current selection boundary. We study the additional error caused purely by elimination, defined as the probability that a population, especially the best population, is eliminated even though it would have been retained under the original curtailed procedure. Conservative bounds, approximation results, and simulation studies show that this elimination-induced risk remains very small, while the refined procedure yields noticeable reductions in sampling effort.
Zitieren
APA
Zhang, Y., & Chen, P. (2026). Adaptive Elimination in Curtailed Subset Selection: A Refined Procedure. Mathematics, 14(14), 2518. https://doi.org/10.3390/math14142518
Vancouver
Zhang Y, Chen P. Adaptive Elimination in Curtailed Subset Selection: A Refined Procedure. Mathematics. 2026;14(14):2518. doi:10.3390/math14142518
BibTeX
@article{Zhang2026,
author = {Zhang, Yifang and Chen, Pinyuen},
title = {Adaptive Elimination in Curtailed Subset Selection: A Refined Procedure},
journal = {Mathematics},
year = {2026},
volume = {14},
number = {14},
pages = {2518},
doi = {10.3390/math14142518}
}
EUVIMED