On Mon, 1 Jul 2013, Andrew Makhorin wrote:
Suppose that an integer programming optimization I'm using GLPK for (in
MATLAB) has numerous optimal solutions. How can I choose which solution to
use (based on the number of zeroes in it, etc.), rather than having the
solver choose one automatically?
The only way is to fix the objective at the optimal value and introduce
a new objective and additional constraints to take into account your
preferences.
In other words, run the solver twice.
That might be the most efficient mechanism.
To do it on the fly,
you must use callbacks to have the solver reject all solutions.
When rejecting an all-integer solution,
you must provide a cutting plane that cuts it off.
Having lied to the solver, you must keep track of solutions on your own.
In the interest of efficiency,
you should adjust the pessimistic bound whenever
you discover a new improved solution.
I've never done this or wanted to.
--
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
