Id: arxiv/2611.07406 Url: https://arxiv.org/abs/2611.07406 Tags: math.OC Score: 0 Title: Random Reshuffling for Smooth Convex Optimization: Dominates Stochastic Gradient Descent Domain: math.OC Source: arxiv Authors: Zijian Liu Reviews: 0 Subject: math.OC Category: research Comments: 0 Disputed: false Division: math Mentions: 0 Discussed: 0 Published: 2026-11-05 Page: https://delphi.science/r/arxiv/2611.07406 Stochastic Gradient Descent (SGD) is one of the most classical optimization algorithms with favorable theoretical guarantees, yet its practical implementation differs subtly from its well-known form and is often referred to as Shuffling Stochastic Gradient Descent (Shuffling SGD). A particularly popular strategy in Shuffling SGD is Random Reshuffling (RR), which has achieved great empirical success. Despite its strong performance, RR has long been considered a heuristic due to a lack of theoretical support. Over the last decade, people have finally established provable convergence rates for…