Progressive Filtering is an algorithm to solve instances of the Location Routing Problem. The algorithm is described in the paper "A progressive heuristic for the location-routing problem and variants" by Florian Arnold and Kenneth Sörensen (Computers & Operations Research, 2021).
ProgressiveFiltering can be turned into a runnable .jar and started via the command line. The following arguments can be provided:
path/instance.json The path to the instance to be solved. The instance format has to correspond to the the example in the input folder -m the maximal number of configurations to be considered in the first stage, e.g., "-m1000" would limit the number of configurations to 1000. Default value is 20.000 -t the number of threads to be used, e.g., "-t4" to use four threads. Default value is 1. -f the definition of the filter stages seperated by a commata, e.g., "-f100,5,1" would signify that three filtering stages are applied after which 100, 5 and 1 configurations are left over. Default value is 100,10,3,1
All dependencies are provided in the libs folder as .jar files. json-simple to read and write json files log4j for logging MDVRPSolver to solve MDVRP instances
Progressive Filtering has been implemented by Florian Arnold during his time as PhD student at the Unversity of Antwerp. The algorithmic ideas have been developed together with Kenneth Sörensen.
We have provided the sorce code to allow students and researchers to adapt the code to their need. We welcome a short note in case the source code is actively used or changed, however, this is not mandantory.
The software is for academic use only