中文

Threshold Dynamics and Correlated Prophet Inequalities

计算机科学与博弈论 2026-07-10 v1 数据结构与算法 最优化与控制

摘要

Prophet inequalities have become a central tool for analyzing the performance of online algorithms. However, most existing results assume that input random variables are independent, which limits their applicability. Motivated by this gap, we study prophet inequalities under two correlation models induced by a latent state of the world variable ZZ. In the common-base model, the algorithm observes the sequence Z+X1,,Z+XnZ+X_1,\dots,Z+X_n. We analyze single-threshold algorithms with the constraint that they always accept the final item, guaranteeing a reward of at least ZZ. When ZZ is chosen adversarially, we characterize the optimal deterministic algorithm of this form, achieving a competitive ratio of 0.3810.381. We then show that randomizing improves the guarantee to 0.40.4. By a minimax argument, the same ratio is achievable when ZZ is random. We depart from standard techniques by establishing a stronger lower bound of 0.410.41 and an upper bound of 0.4750.475, ruling out the possibility that this class of algorithms attains the 1/21/2 ratio known for independent inputs. The core technical contribution is a new analytical framework that captures the reward dynamics of single-threshold algorithms. We introduce a differential equation characterizing the expected reward of a threshold in the worst-case instance, parameterized by the distribution of the maximum. This equation admits a closed-form and unifies known single-threshold prophet inequalities, yielding a simple threshold-optimality condition applicable to the common-base model. Finally, we study the common-scale model, where inputs take the form ZX1,,ZXnZ\cdot X_1,\dots,Z\cdot X_n. We show that this minimal multiplicative correlation yields strong impossibility results: no algorithm can achieve a competitive ratio exceeding 1/n1/n.

引用

@article{arxiv.2607.09887,
  title  = {Threshold Dynamics and Correlated Prophet Inequalities},
  author = {José Correa and Maximilian Fichtl and Reda Jlibene and Rida Laraki and Vasilis Livanos and Kevin Schewior and Victor Verdugo},
  journal= {arXiv preprint arXiv:2607.09887},
  year   = {2026}
}