Computational and Complexity Results for an Interior Point Algorithm on Multicommodity Flow Problems (Extended Abstract)

7 determine the maximum multiple of the demand vector that one can satisfy subject to the capacity constraints. The solution, , to the CFP problem is then used to scale the demands, thereby creating a feasible MCNF problem. We then randomly assign costs per unit ow for each commodity in each edge and solve the resulting MCNF problem. For the instances that we solved, the cost range was 1 to 1000 and the capacities ranged from 1 to 1000.