During my time at ETH Zürich, I gained extensive experience in solving optimization problems using local search-based meta-heuristics. What started as a hobby project evolved into a Rust library called RapidSolve, which provides a framework for implementing multiple meta-heuristics to tackle combinatorial optimization problems. Over time, I developed and integrated the following algorithms into the library:
The library is designed to be highly extensible, and I plan to incorporate additional algorithms in the future.
One of RapidSolve’s standout features is its support for hierarchical objectives, allowing the meta-heuristics to address problems in multiple stages automatically. For example, if the primary objective is to minimize a violation penalty, the meta-heuristic first seeks a feasible solution (eliminating violations) before optimizing the remaining objectives.
Before releasing the library on crates.io, I added comprehensive documentation including example code demonstrating how to apply every supported meta-heuristics to the classic traveling salesman problem.