cs.LGNov 6, 2026

Learning to Restart: Differentiable Restart Policies for Optimization Dynamics

Authors: Yiman Fong, Zhonglin Xie, Zaiwen Wen

Organizations: Department of Industrial Engineering, Tsinghua University · Beijing International Center for Mathematical Research, Peking University

Abstract

Adaptive restart mechanisms, which reset the momentum or other auxiliary state of an iterative optimizer, play a central role in the empirical performance of accelerated first-order methods, yet existing restart rules remain almost entirely hand-designed. The obstruction to learning restart policies is that, unlike the typical Learning to Optimize (L2O) setting, the discrete restart events make the resulting trajectory non-differentiable in the algorithm parameters and thus block gradient-based optimization of the restart policy. We resolve this obstruction by developing a differentiable framework for restart policies through their continuous-time limit. We introduce surrogate gradients for the restart events and optimization trajectories that are efficient to compute and provably converge to their continuous-time counterparts. We validate the effectiveness of our surrogate gradients in L2O settings, training restart policies that improve upon standard handcrafted adaptive restart criteria on quadratic and linear programs, and that transfer zero-shot from synthetic quadratics to logistic regression on real datasets.

Figures & tables

Appendix figures & tables5 assets

Supplementary material from the paper’s appendix.

Appendix

Explore similar work

CardsList
  1. When to restart? Exploring escalating restarts on convergence

    Mar 4, 2026Ayush K. Varshney, Šarūnas Girdzijauskas, Konstantinos Vandikas +1Deep Learning OptimizationLearning Rate Scheduling

  2. Does "Do Differentiable Simulators Give Better Policy Gradients?" Give Better Policy Gradients?

    Apr 20, 2026Ku Onoda, Paavo Parmas, Manato Yaguchi +1Policy GradientRobotic RL

  3. Learning to Optimize by Differentiable Programming

    Jan 23, 2026Liping Tao, Xindi Tong, Chee Wei TanPrimal-Dual OptimizationLearning to Optimize