Predictive dual smoothing speeds up solving large linear problems
Predictive Dual Smoothing for Column Generation
Machine LearningArtificial Intelligence
Summary
Large mathematical problems with many parts can take a long time to solve because systems keep switching between guesses that don’t always improve the solution efficiently. The authors found a way to make these guesses smarter by predicting what future solutions might look like, guiding the process more steadily toward the best answer. This method uses learned insights from past problem-solving attempts to make solving similar problems faster and requires fewer steps. They tested this on common optimization problems and showed it works better than current methods.
What this means in practice
- •For logistics planners: Improve large-scale vehicle or resource allocation by reducing the time to solve optimization problems with predictive guidance on resource usage variables.
- •For manufacturing schedulers: Speed up cutting stock and assignment planning through a method that predicts future problem states to find efficient solutions faster.
Authors
Senne Berden, Noah Schutte, Andrea Lodi, Tias Guns
Abstract
Solving large-scale linear programs efficiently is an important challenge in many optimization settings. A key technique is column generation, which alternates between solving the master problem over a restricted subset of the variables, and using a pricing subproblem to identify new variables to add. The pricing subproblem is guided by the dual solution of the current restricted master problem, but oscillations in these dual solutions can substantially slow convergence. Dual stabilization methods address this issue. Dual smoothing is a common stabilization method, which guides the pricing subproblem using a combination of the current dual solution and duals from previous iterations. However, while past dual solutions can stabilize the dual trajectory, they do not necessarily guide pricing towards useful new variables. We therefore introduce predictive dual smoothing, which instead combines the current dual solution with a learned prediction of future duals to steer pricing towards variables that are more useful in subsequent iterations. The predictor is trained offline using supervision extracted from standard column generation trajectories and is used only to modify the pricing subproblem's objective function, while exact reduced-cost checks and fallback pricing with the unsmoothed duals preserve correctness. Experiments on cutting stock and generalized assignment problems show that predictive dual smoothing substantially reduces generated columns and wall-clock time relative to standard column generation and existing classical and learned stabilization methods. These gains extend to out-of-distribution instance sizes, and predictive smoothing provides further improvements when combined with strong classical stabilization.