This package(Christofides) provides a way to implement Christofides algorithm for solving Travelling Saleman Problem(TSP) to obtain an approximate solution on an undirected graph(Distance Matrix) provided as an upper Triangular matrix. The Distance from a node on to itself is assumed 0.

Use the compute() function which takes as input a distance_matrix and returns a Christofides solution as follows:

The Distance Matrix is an upper Triangular matrix with distance from a node on to itself 0, since Christofides algorithm could only be applied for undirected graphs. Also the distance between a node on to itself is practically 0. Example for distance_matrix is as follows, distance_matrix = :

The above distance_matrix should be provided as an input to christofides.compute(), when we want to calculate TSP for distance = :

christofides.compute(distance_matrix) returns a Dictionary with following Keys: Christofides_Solution,
Travel_Cost, MST, Odd_Vertices Indexes, Multigraph, Euler_Tour

• Christofides_Solution: A list consisting of approximate tour for TSP. Use: TSP['Chistofides_Solution']
• Travel_Cost: The cost of TSP tour generated. Use: TSP['Travel_Cost']
• MST: The minimum spanning tree generated during the Christofides algorithm. Use: TSP['MST']
• Odd_Vertices: A list of odd vertices of minimum spanning tree. Use: TSP['Odd_Vertices']
• Indexes: List of edges from minimum cost perfect matching of odd vertices. Use: TSP['Indexes']
• Multigraph: Edges of multigraph formed after Indexing. Use: TSP['Multigraph']
• Euler_Tour: The Euler Tour of the Multigraph. Use: TSP['Euler_Tour']

• _csr_gen_triples(csr_matrix)
• _odd_vertices_of_MST(distance_matrix, number_of_nodes)
• min_Munkres(distance_matrix, bipartitie_graphs)
• Munkres_cost(indexes, bipartite_graph)
• bipartite_Graph(distance_matrix, bipartite_set, odd_vertices)
• create_Multigraph(distance_matrix, MST, indexes, odd_vertices)
• Euler_Tour(multigraph)
• shortcut_Euler_Tour(tour)
• cost(christofides_tour, distance_matrix)

Download Package and Install using:

or:

scipy, numpy, networkx, munkres