Meta open-sources Rebalancer, a C++ assignment solver with a Python interface
Overview
Meta has open-sourced Rebalancer, a C++ library with a Python interface for solving constrained assignment problems, under Apache 2.0.
The release includes documentation, a PyPI package and a debugging UI called Rebalancer Explorer. According to MarkTechPost, Meta has used the library for resource allocation for over nine years and runs about 40 million problems a day, with a P99 solve time of 12 seconds on 265,000 objects and 3,200 bins. PyPI still classifies the package as Alpha.
A later MarkTechPost tutorial uses Rebalancer 1.0.4 for task placement and compares its local search with the exact HiGHS mixed-integer solver on toy models, synthetic clusters and 60 OR-Library instances. The author reports that in 14 of 24 solves the solved flag read True while memory stayed over capacity, so the author checked each plan by recounting the returned assignment.
Written by AI from the articles below · updated Oct 13, 7:29 PM ET
Check the sources:
Developments
2 developments
- Oct 13, 6:14 PM ET · 1 articleTutorial on using Meta's Rebalancer for task placement and optimizationMarkTechPost: Meta's Rebalancer tutorial shows task placement with local search and HiGHS
- Oct 7, 2:29 AM ET · 1 articleMeta open-sources Rebalancer, a C++ assignment solver with Python interfaceMarkTechPost: Meta open-sources Rebalancer, a C++ assignment solver for placement problems
Article timeline
The articles in this story. Times are ET.
- MarkTechPostNewsMeta's Rebalancer tutorial shows task placement with local search and HiGHS
AIA tutorial implements Meta's Rebalancer 1.0.4, with a C++ core and Python binding, for task placement and optimization. It compares local search with the exact HiGHS mixed-integer solver on toy models, synthetic clusters, and 60 OR-Library instances. The author checks each plan by recounting the returned assignment, because in 14 of 24 solves the solved flag read True while memory stayed over capacity.
- MarkTechPostNewsMeta open-sources Rebalancer, a C++ assignment solver for placement problems
AIMeta has open-sourced Rebalancer, a C++ library with a Python interface for solving assignment problems under constraints and objectives, released under Apache 2.0. The article reports that Meta has used it for resource allocation for over 9 years and runs about 40 million problems a day, with P99 solve time of 12 seconds on 265k objects and 3.2k bins. The package can be installed with pip install rebalancer, though PyPI still classifies it as Alpha.
Heat trend
Not enough continuous observations to show a trend yet.