It seems to me that in this particular case (but not in general) you can get away with an ordinary LP using the method you suggested. By putting Y <= Z1 and Y <= Z2 in the program instead of Y=min(Z1,Z2), you only guarantee that Y <= min(Z1,Z2). However, in the optimal solution you will have Y = min(Z1,Z2), because any smaller value would mean an unnecessarily large value for X because of the first constraint.
Marcin On Tue, Jul 20, 2010 at 11:36 PM, xiaomi <[email protected]> wrote: > No. Not their sum. For example this problem: > > Object: minimize X > Constraint: X+Y>=8, Y=min(Z1,Z2) Z1=...., Z2<..... > > Do I describe Y as "Y<=Z1, Y<=Z2" ? That is not correct unless I maxmize Y > in my object at the same time as minimizing X. So I asked whether multiple > objects are functional. > > > > > > Mansour Moufid : >> >> 2010/7/20 xiaomi <[email protected]>: >> >>> >>> I am running a project that need to minimize more than one variable in >>> LP solver. Can GLPK do that? If yes, how to do so? >>> >> >> Minimize their sum? >> >> > > > _______________________________________________ > Help-glpk mailing list > [email protected] > http://lists.gnu.org/mailman/listinfo/help-glpk > -- Dwell not on close decisions, and thus, when you play against dwellers, you will make reciprocal gains in energy conservation and sanity preservation. Tommy Angelo _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
