Iterative algorithm improves flow and cost decisions for large networks
Support Discovery With Iteratively Reweighted Least Squares for Fixed-Charge Network Flow
Artificial IntelligenceMachine Learning
Summary
Allocating flow in a network while deciding which paths to activate is a tough problem that mixes continuous and yes-or-no decisions. The authors introduce a new approach using a smooth approximation and repeated solving of simpler problems to find good solutions faster on large networks. They also improve these solutions by combining results from their method and other heuristics. Their experiments show this approach often finds better solutions than other scalable methods.
What this means in practice
- •For network schedulers: Optimize activation and flow decisions in large transport or communication networks by using fast, continuous optimization to handle fixed connection costs.
- •For supply chain planners: Improve design and cost allocation in distribution networks by efficiently identifying which routes should be used and how flow should be assigned.
Authors
Sindura Saraswathi, Christian Kümmerle
Abstract
The fixed-charge network flow problem (FCNFP) couples continuous flow allocation with discrete arc-activation decisions, making it a canonical but computationally challenging model for a variety of network design and resource allocation problems. Exact mixed-integer linear programming formulations capture the fixed-charge structure faithfully, but often become difficult to solve on large networks. We propose a scalable continuous-optimization algorithm for large-scale single-commodity FCNFP based on an iteratively reweighted least-squares (IRLS) framework. The method replaces the discontinuous fixed-charge and linear arc cost objective with a smooth nonconvex Lasry--Lions surrogate and solves a sequence of weighted quadratic flow subproblems. Each subproblem is solved by a warm-started dual semismooth Newton method whose Newton systems have weighted graph-Laplacian structure, enabling the use of modern Laplacian solvers. To further improve the discovered arc supports of the challenging underlying combinatorial problem, we also develop an algorithmic variant that incorporates objective-driven perturbation restarts and an anchor-union restricted search that jointly leverages supports discovered by IRLS and by complementary FCNFP heuristics. Computational experiments on 410 benchmark, synthetic, and large-scale instances show that our method obtains the best objective quality among the evaluated scalable FCNFP algorithms, with a mean gap of $1.316\%$ to a time-limited MILP reference and a win-or-tie rate of $90.0\%$ among the non-MILP methods. The results indicate that combining smooth continuous optimization with support-level search is an effective strategy for producing high-quality feasible solutions to large-scale FCNFP.