Open-Sourcing Rebalancer: A Generic, High-Performance Library for Solving Assignment Problems

Open-sourced Rebalancer, a generic, high-performance library for solving assignment problems. Rebalancer separates concerns of problem specification, storage, solving, and debugging. It provides a language for describing problems using objects, bins, constraints, and objectives, transforming them into expression graphs. It supports both optimal solvers (MIP) and local search solvers, with local search being preferred for large-scale problems at Meta. Rebalancer has been used internally for over nine years to solve millions of assignment problems daily across various infrastructure optimization tasks. A debugging UI tool, Rebalancer Explorer, is also released.