> > Yes. If you change a row, the optimal basic solution may either remain > > optimal or become primal infeasible, but in both cases it remains dual > > feasible, so you can reoptimize the modified lp with the dual simplex > > Suppose the constraint's slack is nonbasic > and one negates all the coefficients. > I'm pretty sure that that would make it dual infeasible. >
Yes, you are right. By inertia I meant adding a row, so its slack would be initially basic. Sorry. _______________________________________________ Help-glpk mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-glpk
