Usually for TSP, we know the number of nodes that would be possibly included in the subtours. I am modeling a CARP problem. Not all the arcs are used to travel. Therefore the number of arcs that are used are not known before solving the model. Thus, I cannot build a set of arcs, for SEC. Secondly, some of the tours I get are in form of two polygons with one of the nodes common to both. Technically, this cannot be categorized as subtour. This eliminates the possibility of using Fulkerson's SEC, because, the common node will have two different values for u_i. Do you have any idea how can I model the SEC for this problem.
asked 14 Feb '13, 16:16