pylibmgm

The pylibmgm module provides low-level building blocks for multi-graph matching problems. This API is useful when you need fine-grained control over the optimization process or want to implement custom solvers beyond what is provided in pylibmgm.solver.

Core Data Structures

Graph and Model Classes

These classes represent the fundamental data structures for graph matching problems.

`Graph`	Represents a single graph in a multi-graph matching problem.
`GmModel`	Pairwise graph matching (GM) model.
`MgmModel`	Multi-graph matching (MGM) model.
`CostMap`	Container for unary and pairwise costs in a graph matching problem.

Solution Classes

Classes for representing and manipulating solutions to matching problems.

`GmSolution`	Solution to a pairwise graph matching problem.
`MgmSolution`	Solution to a multi-graph matching problem.

Solvers and Optimization

Graph Matching - Solvers

Solvers for pairwise graph matching (GM) problems.

`LAPSolver`	Linear Assignment Problem solver.
`QAPSolver`	Quadratic Assignment Problem solver for graph matching.

Multi Graph Matching - Solution Generators

Generators for constructing initial multi-graph matching solutions.

`MgmGenerator.matching_order`	Order in which graphs are matched during solution generation.
`SequentialGenerator`	Sequential solution generator for multi-graph matching.
`ParallelGenerator`	Parallel solution generator for multi-graph matching.

Multi Graph Matching- Local Search Methods

Local search algorithms for refining solutions.

`GMLocalSearcher`	Local search method based on iteratively re-matching graphs.
`GMLocalSearcherParallel`	Parallel version of GM local search.
`SwapLocalSearcher`	Swap-based local search for multi-graph matching.

Utility Functions

`build_sync_problem`(model, solution[, feasible])	Build a synchronization problem from an MGM model and solution.
`omp_set_num_threads`(num_threads)	Set the number of OpenMP threads for parallel computation.