HybridCVRP is a metaheuristic solver for the Capacitated Vehicle Routing Problem (CVRP). It contains an implementation of the Hybrid Genetic Search with Ruin-and-Recreate [1].
A demo of the solver can be found at https://martsime.github.io/hybridcvrp.
For the demo webpage, the solver is compiled to WebAssembly and run in the browser. More information about the demo can be found here: https://github.com/martsime/hybridcvrp-web. Despite the demo being fully functional, running the solver compiled to WebAssembly in the browser may come with a signficiant performance drawback. Therefore, for any use other than demo purposes, it is recommended to install the solver and run it locally as described below.
As the solver is implemented in Rust, you are required to have a Rust installation to build it. A Rust version of (1.54.0 stable) or newer is required, although older versions could be supported.
The simplest way to run the solver is with the following command:
cargo run --release <instance-path>
For example:
cargo run --release instances/X-n101-k25.vrp
The best way to configurate the solver is by changing the parameter values in the config.yml file, which is parsed by the solver at startup.
There are also a few optional arguments on the run command. Run cargo run --release -- --help to see more information about run command and its arguments. Note that the provided arguments to the run command will take precedence over the parameter values set in config.yml.
Huge thanks to Thibaut Vidal for open sourcing an implementation of the Hybrid Genetic Search specialized for the CVRP. Access to the implementation was very helpfull during the development of the HybridCVRP solver.
[1] Simensen, M., Hasle, G. & Stålhane, M. Combining hybrid genetic search with ruin-and-recreate for solving the capacitated vehicle routing problem. Journal of Heuristics (2022). https://doi.org/10.1007/s10732-022-09500-9