On Fri, 16 Dec 2011, Xypron wrote:
users of GLPK have had often had problems with the accuracy of big M formulations.
The trick to using big M is to do the math to get the smallest Ms that will work.
In CPLEX big M formulations can be replaced by indicator constraints. This is a constraint switched on and off by a binary.
What I suspect happens is: Initially, the indicator constraint is turned off. The indicator constraint is usually turned on only after a branch on the indicator variable. That might be good for accuracy, but it won't be so good for tree size. Of course, big big M methods have the same tree size problem, they just have lower accuracy. To get accuracy, use a small M. -- Michael [email protected] "On Monday, I'm gonna have to tell my kindergarten class, whom I teach not to run with scissors, that my fiance ran me through with a broadsword." -- Lily _______________________________________________ Help-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/help-glpk
