Project
FalCom
A Sampling Method for Districting and Hierarchical Facility Location
About
FalCom introduces a Markov chain Monte Carlo sampling method for solving districting and hierarchical facility location problems. The algorithm generates diverse, near-optimal solutions by exploring the combinatorial space of feasible districting plans, enabling rigorous statistical analysis of solution quality and fairness properties.
Software
A Python library implementing a hierarchical and capacitated ReCom algorithm that simultaneously partitions a dual graph into hierarchical service districts, locates facilities within districts, and allocates expert teams to facilities — while satisfying capacity-demand balance and user-choice constraints such as budget.
A Python library providing a wide range of plotting functions to analyze falcomchain inputs and outputs.
Interactive Visualizer
The FalCom Visualizer animates the Markov chain sampling process, showing how the algorithm traverses the solution space, the evolution of district boundaries, and convergence behavior in real time.
Open VisualizerCase Study — London Ambulance Service
An interactive map of the LAS three-level operational hierarchy on which we run FalCom: 5 sectors (per-sector colour palettes), 19 groups (light-to-dark tones within each sector), 63 ambulance-station candidates, and 7 super-facility candidates (sector HQs and emergency operations centres). The base graph is Greater London at the LSOA scale (5,042 polygons). Both sector and group boundaries are derived by Voronoi-by-station and repaired to contiguity.
View LAS Map