XDEv11 / MGML

Maximum Graph Matching Library (NSD 2023 Spring NYCU Term Project)

MGML

Maximum Graph Matching Library (NSD 2023 Spring NYCU Term Project)

C++ API (MGML.hpp)

All vertices are numbered from 0.

std::vector<int> MGML::MCBM(const std::vector<std::vector<int>>& adjx, int ny);

Maximum Cardinality Bipartite Matching

• Parameter
  • Give adjx that for all x in $X$, adjx[x] stores all vertices in the other part, $Y$, connected to x and the number of vertices in $Y$, ny.
• Return Value
  • Return mx such that mx[x] is the matched vertex to x or -1 if unmatched.

std::vector<int> MGML::MWBM(const std::vector<std::vector<long long>>& weight);

Maximum Weight Bipartite Matching

• Parameter
  • Give a $n$ by $n$ matrix, $weight$.
• Return Value
  • Return mx such that mx[x] is the matched vertex to x

std::vector<int> MGML::MCM(const std::vector<std::vector<int>>& adj);

Maximum Cardinality Matching

• Parameter
  • Give adj that for all u, adj[u] stores all vertices connected to u.
• Return Value
  • Return mh such that mh[u] is the matched vertex to u or -1 if unmatched.

Python API

The module name is MGML and all vertices are also numbered from 0.

The Python API are basically almost the same as C++ API.

MGML.MCBM(adjx, ny)

MGML.MWBM(weight);

MGML.MCM(adj)