|Subject:||RE: [Help-glpk] Solve subproblem then use sol'n in larger problem|
|Date:||Sun, 10 Feb 2008 19:06:34 -0800|
> To do what you want, you need to provide
> an initial basis for the bigger problem.
> Copy the stati of the common variables and
> rows from the subproblem to the big problem.
> Place the new binary variables at their upper bounds.
> Make the new rows basic.
> Tell GLPK to use what you have provided.
> If one can, tell GLPK you have an incumbent.
> If one cannot, one can tell GLPK that one has a pessimistic bound.
> Invoke the simplex method.
> Invoke branch and bound.
Thanks so much for the response. Looking at the GLPK API, I think the magic words are going to be glp_set_col_stat()/glp_set_row_stat() based on your description. That makes sense. I think that lpx_warm_up() will then do the dirty work of setting up the basis based upon the stati I have set. I am hoping that glp_simplex()/lpx_integer() or just lpx_intopt() will be able to take over from there. If any of these thoughts seem completely wrong, hopefuly someone will correct me. When I get around to coding this within the next month or so, I may come back with more questions. Thanks for the help getting started.
Climb to the top of the charts! Play the word scramble challenge with star power. Play now!
|[Prev in Thread]||Current Thread||[Next in Thread]|