> I am implementing strong branching using GLPK. > While determining the branching variable from the set of candidate > variables, I calculate LP > relaxation at each child node. > > > I want to fix the bounds of the variable if one of the child nodes is > infeasible. > I tried doing this as follows: > glp_set_col_bnds(glp_ios_get_prob(tree), j, GLP_FX, 1.0, 1.0); > (where j is the candidate variable for which one of the child nodes is > infeasible. I am doing this > in a subroutine where I determine set of candidate variables and > select a variable to branch on. This subroutine is called > by callback routine using GLP_IBRANCH) > > > However, When I try to do this, I get error by glpk pointing to the > branch_on routine in glpk source code. > > > Can you please guide me how can I fix the bounds in strong branching. >
You cannot change the active problem object in that way, because it is used by the mip solver. To implement a strong branching you need to use a copy of the current subproblem created with glp_copy_prob. _______________________________________________ Help-glpk mailing list [email protected] https://lists.gnu.org/mailman/listinfo/help-glpk
