[jgrapht-users] Algorithms for flow graphs

[Joris K. <de...@gm...>]

Hello,

I'm about to build an algorithm for min-cost flow graphs. As you can
imagine, this needs a flow graph which typically has the following
properties:

1. Edges have a Capacity and a Cost
2. Vertices have a real postive/negative value indicating the
demand/supply associated with each vertex.
3. The graph is directed

Due to the generics in JgraphT I can easily create vertex/edge classes
which maintain these values. There is however a problem with the
currently existing algorithms in JgraphT. For example, in a flow
graph, a shortest path is usually (or at least in my case) calculated
using the edge costs as 'distances'. Furthermore, a max flow algorithm
utilizes the edge capacities. When I would use
org.jgrapht.alg.DijkstraShortestPath, the algorithm uses the
getEdgeWeight(E edge) function to obtain the distances. Similarly,
org.jgrapht.alg.EdmondsKarpMaximumFlow utilizes the getEdgeWeight(E
edge) function to read the edge capacities. The getEdgeWeight(E edge)
function can only return one value: either the edge capacity or the
edge cost. What would be the cleanest way to resolve this issue
(preferably something reusable by the jgrapht community)?