Project
FRESH Solutions
A general, reproducible hybrid metaheuristic solver for hard combinatorial optimization problems.
FRESH Solutions is a general, reproducible hybrid metaheuristic solver for hard combinatorial optimization.
It extracts the search architecture developed in WS4: compiled sparse problem state, exact incremental accounting, candidate and effect indexes, exact CPU local moves, batched GPU population search, weighted goal programming, and exact replay of reported incumbents.
The solver is the public artifact for the second paper in the FRESH Solutions papers project. It is not intended to be a universal mixed-integer programming solver or a proof of optimality.
Status
- Open software project
People And Roles
- FRESH Lab: project team
Outputs
- GitHub repository
- Hybrid metaheuristic solver
- Public benchmark evidence
- Paper artifact in preparation