Reproduce “Contextual Bandits with Packing and Covering Constraints: A Modular Lagrangian Approach via Regression” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Aleksandrs Slivkins, Xingyu Zhou, Karthik Abinav Sankararaman, Dylan J. Foster

The authors present a modular algorithm for contextual bandits with linear constraints, combining Lagrangian primal–dual resource management with regression-based action selection. It supports packing and covering constraints, including negative resource consumption. With large budgets, a known positive feasibility margin, and bounded regression error, it achieves near-optimal square-root outcome-regret scaling; without the margin, the guarantee weakens to T^{3/4} scaling.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.