Reply to thread

Message: [QUOTE="JSH, post: 3597148"] Good catch. That is just a miss on my part, but it's easily fixable, by saying that a path is calculated for each possible starting node in turn. The algorithm is still polynomial time with that change. As that's fairly easy I'd like to focus first on getting a full path from an arbitrary start at just any node, say, chosen randomly. I realized an omission in my primary algorithm, as I simply forget about the cost for T_1 from N_1 to the node being considered, which is kind of an odd thing to forget. The solution and correction to the algorithm is that cost is multiplied as well. So if cost from N_1 to N_2 is cost_1 for T_1 and cost from N_1 to N_3 is cost_2 for T_2, then the calculation is cost_1*cost_2*distance, where here distance is 1, so it is cost_1*cost_2. With the correction you'd have two weight values for each, so cost_1*cost_2*time_1*time_2*distance If more than one path has the least value then you randomly pick. Oh, that's not good. I'll need to consider this carefully later. Kind of in a rush this morning... Thanks for the reply! I'll have to consider it in detail later as I'm in a rush this morning, but wanted to add the correction. If you are correct then that kills the usefulness of the idea for TSP, but doesn't mean I won't still program the algorithm out of curiosity. James Harris [/QUOTE]