PHP implementation of the Hopcroft-Karp algorithm
The Hopcroft–Karp algorithm is the most efficient way to find a maximum cardinality matching in a bipartite graph.
Think of applicants for jobs, each applicant has skills for a certain sub-set of the jobs, and we need to employ the maximum number of applicants to jobs. Or scheduling appointments for a set of time slots when each person is only available at specific time slots.
- This is a quick and lazy implementation for now which works well for my use case.
- I am more than happy to improve it further as and when there is a need for that. Issues and pull requests are welcome.
- As per semver, there is no BC guarantee until 1.0
composer require motley/hopcroft-karp
You have to provide a set of edges describing your biparite graph. The vertex values can be object, string or int. The edge shall be one to many. Example:
$matching = HopcroftKarp::matching([
['left1', ['right2', 'right3']],
['left2', ['right1']],
['left3', ['right2']],
['left4', ['right2']],
['left4', ['right4']],
]);
The resulting matching object has a few helper methods to inspect it.
$matching->toArray(); // [['left1', 'right3'], ['left2', 'right1'], ['left3', 'right2'], ['left4', 'right4']]
$matching->getRightByLeft('left2'); // 'right1'
$matching->getRightByLeft('unmatched'); // null
$matching->getLeftByRight('right2'); // 'left3'
$matching->getAllLeft(); // ['left1', 'left2', 'left3', 'left4']
$matching->getAllRight(); // ['right3', 'right1', 'right2', 'right4']
You can pass in a previous matching as an optional argument and the algorithm will try to keep as many edges from it as possible. Example:
$previousMatching = new Matching([
new Edge('left1', 'right1'),
new Edge('left2', 'right2'),
new Edge('left3', 'right4'),
new Edge('left4', 'right3'),
]);
$matching = HopcroftKarp::matching([
['left1', ['right1', 'left2', 'right3']],
['left2', ['right1', 'left2', 'right3', 'right4']],
['left3', ['left2', 'right3', 'right4']],
['left4', ['right1', 'right2', 'right3', 'right4']],
], $previousMatching);
Issues and pull requests are welcome.