Flows and Matching

Overview

Some problems only look combinatorial until the constraints line up and the clean model is flow. This category is for those reductions: send units through a network, enforce conservation, and let the residual structure do the work.

Current Focus

The current baseline covers the three core tools I expect to reuse most often: Dinic for max flow, bipartite matching when the graph structure is already clean, and min-cost max-flow when the model needs an optimization layer on top of feasibility. The next additions here should be more specialized routines such as Hungarian or general matching.