The integral simplex using decomposition (ISUD) algorithm was recently developed to solve efficiently set partitioning problems containing a number of variables that can all be enumerated a priori. This primal algorithm generates a sequence of integer solutions with decreasing costs, leading to an optimal or near-optimal solution depending on the stopping criterion used. In this paper, we develop an integral column generation (ICG) heuristic that combines ISUD and column generation to solve set partitioning problems with a very large number of variables. Computational experiments on instances of the public transit vehicle and crew scheduling problem and of the airline crew pairing problem involving up to 2000 constraints show that ICG clearly outperforms two popular column generation heuristics (the restricted master heuristic and the diving heuristic). ICG can yield optimal or near-optimal solutions in less than one hour of computational time, generating up to 300 integer solutions during the solution process.
Paru en juin 2017 , 21 pages