First of all: it is really good to see GLPK becoming better and better. There are a number of primal heuristics which would be nice to have in GLPK.
Would it be difficult to implement the following proposals, regarding how often the problem of big M formulation comes up here in the mailing list? In case of CPLEX one can use the so-called indicators: http://www.ilog.com/products/cplex/news/whatsnew.cfm I find the proposal in page 14 interesting: J. N. Hooker, A principled approach to mixed integer/linear problem formulation, June, revised August, to appear in ICS Proceedings 2009. http://ba.gsia.cmu.edu/jnh/milpmodeling2.pdf http://ba.gsia.cmu.edu/jnh/papers.html I have a heuristic to be implemented for computing sharp big M-s. I hope i will find some time in the near future to try it. Best, Ali _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
