[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
## [Help-glpk] Stop criteria for MIP integer solution?

**From**: |
Pieter Thysebaert |

**Subject**: |
[Help-glpk] Stop criteria for MIP integer solution? |

**Date**: |
Tue, 13 Aug 2002 09:54:20 +0200 |

Hi,
I'm wondering if I can let the solver (for MIP) stop when it reaches the
first solution "better than the currently known optimum" ?
(I've found the time limit parameter...)
And in relation with this, can the solver continue from where it stopped
(after time exceeded, for instance) by calling lpx_integer() a second time?
The reason I ask is the nature of my specific problem: the (binary) decision
variables are time-indexed variables, and the objective is to minimize a
"time". So when a feasible solution is found (with objective value T), I can
tell for sure that the decision variables related to times > T are certain to
be 0 in the optimum (and in this feasible solution).
This knowledge would allow me to "reduce" the problem (= removing variables =
deleting columns from the constraint matrix). I could then try to find the
first feasible solution for the reduced problem with objective < T etc...
This way, I will iteratively try to find a solution for problems that keep
getting smaller, and I suppose that this could be a lot faster than trying to
find the optimum to the original large problem?
Pieter

**[Help-glpk] Stop criteria for MIP integer solution?**,
*Pieter Thysebaert* **<=**