Possible Graph Implementations

[maxfischer2781]

Get real with the purpose of the GraphI Interface: implement additional, optimised graph types.

• Adjacency Matrix: A literal NxN matrix storing the edge values between nodes (row/column). Probably works only with integer valued nodes.
  • Provide a graph[int, int]->number optimisation using numpy arrays.
  • Offer a wrapper to transform arbitrary nodes to indices.
• Adjacency Mapping: Combination of AdjacencyGraph and an Adjacency Matrix, for sparse graphs and any hashable nodes. Store a Dict[Node] -> List[Node] for neighbours of a node, and a Dict[Node, Node] -> Object for values of edges.
  • We can probably calculate/estimate the savings/cost. See stackoverflow for the behaviour of dict.

[maxfischer2781]

See also issue #3 to compare different types.