Originally posted by: waldstein
When we use column generation to solve this problem, we need an initial feasible basic solution, which is composed by several paths of these flows. Then, column generation procedure can insert a new column to the model and finally obtain the optimal solution.
So how can we find this initial feasible basic solution?
In fact, there is an example of column generation, named of "cutstock". In this problem, several patterns are given before optimization. However, new patterns are generating when optimizing.
But how can we do that in network flow problem?
Thanks!
#DecisionOptimization#OPLusingCPLEXOptimizer